The force transmitted by the vehicle tyre to the ground is limited by a number of factors, but principally by vehicle corner weight and the amount of weight transferred as a result of vehicle dynamics. Tangential to this force at the contact with the ground is the tractive force, which in the static case of a given vehicle-tyre combination is a fixed constant but which will increase as the tyre under power begins to lose traction and begin to slip. Generally referred to as the ‘slip ratio’ – the ratio of the speed of slip to that of the driven wheel – this coefficient increases up to a point, after which it falls away again, as shown in Fig. 1. Highly dependent on the tyre, its temperature and the surface upon which the car is running, this best relative slip speed for a given tyre/surface is somewhere around 10-15%. The role of a vehicle launch system therefore is to assist the driver in maintaining this level of tyre slip without exceeding it and without overspeeding the engine.

Generally included as part of the traction control system on many aftermarket engine management systems, at their simplest level they will need to incorporate a method of determining tyre slip, a sensor in the clutch-actuating mechanism and a method of initiating the system at the start line. Inputs to the algorithm necessary to control this most basic of system are therefore wheel speeds, throttle position, clutch position and engine speed. Tyre slip is readily calculated using wheel speed inputs from a non-driven wheel (usually a front wheel or the average of the two front wheels) and that of the driven wheel or wheels.   

On a command to activate the system (a simple on-off switch) as the clutch is depressed, the driver will then open the throttle until the engine reaches the desired engine launch speed. This will most likely be a ‘soft’ speed limit beyond which the fuel to one or more cylinders would be cut. As a safeguard, at around 250 rpm higher, a ‘hard’ rev limit will be set, beyond which the engine will not go.

Held at this condition in readiness for the start, in the case of a turbocharged engine this could be for a couple of seconds while the boost pressure builds up. Once the engine speed has stabilised (along with the boost pressure if needed), on re-engaging the clutch and with the vehicle beginning to move, the system will cut in until the desired level of slip is created. By now, at full throttle, the system will modulate the fuel to the cylinders until the point where second gear is requested.

At this point it must be noted that with such a simple system there is no control over the clutch or its speed of engagement other than by the driver. It is assumed that for this part of the process the clutch will be either ‘in’ or ‘out’ and that the driver, through skill and/or experience, will be able to operate this without the engine stalling or ‘bogging down’. To have a fully automatic launch control system and therefore the possibility of a two-pedal (brake and throttle) driving layout will need additional sensors, an actuator and a clutch control system controlling engine speed and clutch position at optimum tyre slip.

Launch control systems may be effective and may, it is argued, make the first few seconds of the race marginally safer. However, as a bit of a purist, from the point of view of the sport I do feel it downgrades the skills of the driver.

Fig. 1 - Friction as a function of tyre slip

Fig. 2 - A simple launch control flow chart 

Written by John Coxon

If the FIA were to open Formula One to driverless cars next year, what are the chances – disregarding the available development time, hypothetically speaking – that one of the top teams would field a car quicker over either a qualifying lap or a Grand Prix distance than a human-driven car?

Nick Wirth knows a thing or two about unmanned technology having in the past created sophisticated robots as well as Formula One cars. He discounts the possibility of a driverless Formula One car outrunning its manned equivalent in the short term.

By contrast, a Formula One team technical director, who wishes to remain nameless, says, “In my opinion, based purely on experience with closed-loop lap simulation results, there is significant evidence to suggest that a virtual ‘driver model’ can cope with a significantly more ‘unstable’ car compared to reality. The big question though is whether this theoretical increase could be converted into reality.

“Control system recognition and response is the key factor here,” he says. “I think there is enough evidence to suggest that, in theory, a digital controller could function at a higher bandwidth compared to a human being. If this can be achieved then the digital controller should outperform its human counterpart.

“Obviously, both the development resource and the time required to achieve this is significant, but I believe it could eventually be done. Therefore, all the big-budget teams would ultimately be interested in this development thread if it were legal, which it obviously isn't.”

Mike Lancaster is the brains behind one of the top suppliers of control electronics to professional racing, and he says, “The quick answer is that I have little idea, and I suspect that will be the same for most of us. That said, my initial thoughts are that the human driver is a complex creature endowed with a vast array of subtle feedback systems, and controlled ultimately by the most sophisticated parallel processing system that as far as we know exists anywhere. Despite this, the organic brain is very slow, not only to process information in real time but sluggish in machine terms to react appropriately.

“Lined up against the organic computer is a vastly faster albeit single-minded ‘brain’ that knows nothing at all beyond following sequential instructions. Assuming the driverless car was able to drive around each corner alone and unimpeded, and with sufficient and no doubt considerable time to adapt to circumstances, in the end it would be faster than any human in my opinion.

“The analogue is a modern fighter aircraft, which is fundamentally unstable and controlled by computer, without which it could not be flown by a human. Assuming the same budget was applied to racing a land-based vehicle without a driver then the writing is on the wall. All of that said though, I doubt if the current driverless technology for production vehicles is at the level needed to go faster than a human… yet.”

Written by Ian Bamsey

At the time we didn’t fully appreciate the reliability bit, but anyone who owns a classic vehicle from the period will know how thick and heavy many of the wiring looms were. Little did we realise at the time how important the concept of multiplexing would become, and how just about every single vehicle (including race vehicles) made in the 21st century would rely on the technology.

These days of course, more and more electrical/electronic functions are being put into the car, most of which require some form of control and many of which require some form of local intelligence. One of the earliest protocols devised to transfer serial data between systems is known as CAN (Controller Area Network) or more exactly, the CAN bus, and it is still the most popular. Essentially it is a pair of twisted wires replacing up to 100 or more ‘normal’ wires, with control devices connected in a daisy-chain fashion.

Designed to handle short messages up to 8 bytes in length at rates of up to 1 Mbit/s, the protocol uses a method known as Carrier Sense Multiple Access (CSMA) to transmit the data frames, allowing each control unit or node in the system to transmit the data in frames at any time. The data will include a node address to which the message is to be delivered (a sort of mail post or zip code) as well as the message itself. Passed between nodes in turn, if the postcode does not match that of the node then the message is simply ignored and passed on. Should two or more messages be sent at the same time, a collision will be detected and the message with a higher priority indicator will continue, with the one of lower priority repeated later.

Although satisfactory for most powertrain applications, for x-by-wire safety-critical systems –braking and throttle – another and more powerful system is being touted. Known as FlexRay and similar to CAN, this system is faster (up to 10 Mbit/s) and can handle up to 254 bytes of data in a frame. Wired in much the same way as CAN, FlexRay can be configured if desired to support two separate cable paths, giving the opportunity to transmit the same, repeated data to provide safety-critical redundancy in the system or a completely new set of data to increase the overall throughput should better control be desired.

Whereas with CAN frames vary in length, for FlexRay the frames are split into two, with the first portion the same for all. Since these occur at well defined times in the message they can be used to synchronise the execution of the message in the multiple control units at the individual nodes. Referred to as Time Division Multiple Access (TDMA), each frame is assigned a dedicated time slot in the signal, and since no two frames are assigned the same slot in time, collisions between messages are impossible.

Currently more expensive than CAN, FlexRay’s supporters claim that with wider adoption, prices will fall. But as brake-by-wire systems become more integrated, particularly into regenerative hybrid applications, perhaps this technology will be a better option than just CAN.

And for our door application? Well, designed as a low-cost option for applications not requiring the same degree of control, we have LIN – Local Interactive Network. But that’s another story.

Fig. 1 - The fixed-length FlexRay message

Written by John Coxon

Whether they are for automotive or motorsports use, electrical connectors are generally made from copper alloys plated with tin, silver or gold. Consisting of an outer female part that retains a bullet- or spade-shaped inner part under the action of spring loading, the reliability of the contact is controlled by the mechanical properties of the base material and the conductive properties of the surface layer. Should anything interfere with these properties – be it contamination, oxidation, the formation of sulphides or corrosion – then the resistivity of the joint may be impaired to the point where the electrical circuit is unsustainable. Thus, not only is the normal contact force between male and female parts important but so is the type of plating and its thickness.

A large part of what drives the selection and thickness of the coating is eventually down to cost, but in recent times the drive to reduce the amount of lead in all its forms in the manufacturing environment has led to the replacement of traditional tin-lead coatings with pure tin. In common with most other plating surfaces (except silver in some cases – see below) nickel is used as an underplating of 1-1.5 microns in thickness to prevent the copper migrating through from the base material to form a copper-tin compound, which can reduce the ability of the contact to be soldered if required and could induce a level of compressive stress in the tin deposit to form a type of ‘whisker’, which will eventually lead to system failure.

If tin is selected as your coating, the options are bright tin or matt tin. Bright, apparently polished tin has a more pleasant appearance and a lower coefficient of friction of the two, which may be important if the level of force required to press-fit male and female parts together is large. Matt tin on the other hand is easier to plate but softer and therefore less durable should parts need to be routinely dismantled.

A more expensive plating is pure silver. Favoured for its higher current power applications, silver has the highest electrical (and thermal) conductivity of any metal. Because of the high-power applications therefore, forces normal to the connectors when mated together should be very high. For these high values, and when the thickness of the silver is up to 20 microns, no underplating is generally required. Lower power ratings tend to need a nickel underplating, but because of the low thickness of the silver, such connectors are far less durable in  assembly and dismantling. Let’s not forget also that over time silver will tarnish, so this may not be the best option if the working environment is even slightly corrosive.

The best coating material for both electrical contact and durability though is of course gold. When applied using a nickel underplating, gold will supply a stable and low contact resistance throughout the whole operating life of the connection – but at a cost. Almost inert chemically, gold will not form an insulating film between contacts, and because of this the assembly (and disassembly) forces can be comparatively low. As a result, metal-to-metal contact is easily established with low contact forms, and is the reason why the coating is the preferred choice for applications using low voltages or currents.

Tin-, silver- or gold-plated contacts? Just another decision to be made when designing your wiring loom. 

Fig. 1 - Tin-plated contact

Written by John Coxon

The function of the CE is to control the ac three-phase power of the Motor Generator Units (MGUs) and the dc power from the battery. Currently Formula One uses two MGUs, one for the kinetic system and one for the heat system, where the kinetic system is an upscaled version of the recent KERS and the heat system is attached to the turbocharger. To control these two systems two separate CEs are used, appropriately termed CUK and CUH.

Inside the CE there are two sides to its operation, a low-voltage side, with logic boards controlling the high-voltage side with switches, and capacitors switching the power between the MGU and the battery.

When harvesting energy, the relevant MGU will send its ac power via three high-current cables to its CE. Inside the control unit’s casing is a series of high-current switches called IGBTs (insulated gate bipolar transistors) that switch the current, which also passes through capacitors. The electronics here convert the current to dc format.

The CE then sends the dc power via two cables to the battery. By inverting this process the CE can also take battery power to spin the MGUs.

This process of converting from ac to dc and back creates heat and therefore losses. Keeping the electronics at their working temperature is critical for reliability, and air cooling alone is not sufficient to manage the heat rejection so this is achieved with a dedicated water cooling circuit, which requires a small pump and radiator mounted in the sidepod. 

Conveniently, the CEs provide the car with its 12-24 V supply for the car’s other electrical systems, so the car does not need an alternator or its own battery.

Where to mount the CEs in the car is decided between the power unit manufacturer and the teams, so installations vary, with the units mounted either with the battery in a recess under the fuel tank or on the sidepods, depending on the team’s philosophy on fuel tank size and its impact on wheelbase length, cooling and sidepod packaging. 

New rules limits team to five complete power units for each season, and from 2015 just four. For practicality purposes the power unit is split into six elements, one of which is the CE for the ERS. Should a team fit a sixth CE then there will be a grid penalty for the driver, with further penalties for further new units being used.

This places great importance on reliability, despite these power units being all-new for the 2014 season. At the season’s halfway point, several drivers had already used four CEs, the CE itself proving to be an unexpectedly unreliable part of the complex new power units. Teams have reported that any failure in the CEs tends to be on the high-current side, not the lower-voltage logic boards. The result of high-voltage spikes tends to be catastrophic on the internal components and render the CE unusable.

Although the reliability of the CEs had improved in the latter part of the first half of the season there will be drivers who suffer penalties as a result of these units. 

Fig. 1 - A typical CE unit with the capacitors and high-current cabling evident 

Written by Craig Scarborough

At the simplest level, the main channels on any data logger are likely to be engine speed and throttle angle. To understand how the power unit works in conjunction with the chassis, inputs such as steering and lateral acceleration are also a must. Using these five channels and not having to understand anything else about the engine, a driver can generate more than enough data to improve his driving skills by comparing data lap after lap. Indeed, some kind of chart-overlaying feature is the first stage in any data analysis process, and is the one most commonly used, while a popular add-on feature these days is the ability to synchronise video data files to those logged directly from the engine controller.

As power unit engineers, however, we are much more interested in engine control, so in addition to the above recorded channels we might wish to include a number of pressures (oil, fuel and so on), a smattering of temperatures (coolant, intake air, oil and exhaust gas, say), injector pulse widths, lambda (a function of air-to-fuel ratio) and ignition timing. Logging all these against time for a typical 15-minute club race will generate far more data than many engineers or drivers can ordinarily handle, so some simple way of analysing the data has to be found. Fortunately, even the simplest data logger has a number of ways in which it can assist.

To start with what is generally the default setting, multiple channels can be plotted against the time or distance travelled. This enables the driver or engineer to pinpoint certain events during the lap and read off other parameters happening at the same time.

Many analysis software packages can also generate histograms or sometimes more usefully x-y plots. In the latter case, one channel can be plotted against another to infer whether there is any causal relationship between the two. One example of this is a plot of gallery oil or fuel pressure against the lateral g-force when a slight fall-off in the pressure on right- or left-hand corners could indicate the presence of oil or fuel surge.

A particularly useful feature in many systems is the ‘maths’ channel. By using the data measured on one channel, a new function can be computed to help our understanding. A good example of this, and one that is highly topical in Formula One, is the subject of fuel consumption. By using our knowledge of the pulse width applied to the fuel injector, along with fuel pressure and temperature, we can determine the amount of fuel going into the engine per injection pulse and, summed over a unit of time, the rate of fuel consumption can be calculated. Summed over the whole race, the total fuel consumed can be derived. How accurate this is compared to, say, a fuel flow meter is open to debate, but the whole idea of maths channels increases the flexibility and potential usefulness of any system. 

As powertrains become more complex, the need for data logging and subsequent analysis of the data can only increase.

Fig. 1 - A simple histogram, summarising coolant outlet temperature

Written by John Coxon

Tyres are of course critical to lap time in Formula One, and with the recent trend for high-degeneration tyres, this importance is growing.  Sauber’s Willem Toet in his Lanchester lecture to the Royal Aeronautical Society provided some insight into the proportionate benefits in improvement tyre use. He said that for a given percentage increase in tyre grip, there is a proportionate percentage decrease in lap time by a factor of three; by contrast, aerodynamics provides merely a 1% lap time benefit for a 1% improvement.

Although the interaction of the tyres between the car and the track is a complicated one, a key indicator of how the tyre is being used is its surface temperature, and measuring this accurately gives the race engineers an insight into how the tyre is behaving. To achieve this, teams and their partners have developed increasingly complex tyre temperature sensors.

Taking temperature readings from the surface of a rotating tyre clearly requires a non-contact sensor.

Black rubber tyres have high emissivity, which is the amount of radiated heat they reflect. A mirror may have an emissivity of 0 and perfectly a black object a value of 1. As the tyres radiate very little heat, the temperature sensors need to read in the high emissivity range typically 0.8-0.9595 to take accurate readings.

Then, complicating the issue, is the wide tread of the modern Formula One tyre (330 mm front and 375 mm rear), so either multiple sensors or those with a wide field of view (FOV) are required. With the front wheels being steered, and aerodynamics critical in the area around the tyres, packaging is a challenge, not least the usual installation issues of weight and ruggedness on a Formula One car.

Teams started routinely using infrared tyre temperature sensors in testing some 20 years ago. That was during the tyre war, when each race was followed by a test, usually with the run sheets emphasising tyre assessment.

The first sensors to come into common use were simple infrared modules inside small alloy cases. These sensors convert the radiated heat into a voltage, which is non-linear but is continuous and repeatable, and this analogue output is fed back through the CAN bus to the data acquisition system.

Installation for the rear tyres take the form of a rear-facing sensor embedded in the floor ahead of the rear tyre or sometimes in the flick-up on the sidepod, while the front sensor was set into the trailing edge of the front wing endplate. Since both the sensor and the installation allowed for only a single temperature reading on each tyre (and when the front tyres were not being steered), this was not a perfect solution, so for some tests teams fitted outriggers to the suspension uprights, which mounted three or more sensors to take readings at several points across the tread, even when the front tyres were being steered. This set-up clearly affected aerodynamics though, and would not be legal to run over a race weekend.

More recently, the sensors have been developed further to provide a wider FOV, such that several points on a tyre’s tread can be measured simultaneously. This can be achieved either with multiple sensors in the same module or with a single module with a wide FOV, which then takes readings from multiple positions within the FOV.

With these sensors a team can measure the temperature across the tread, to understand the effect of tyre deformation and suspension camber on tyre temperatures. With some sensors able to measure up to 64 channels from its FOV, from 16 points across the tyre by four positions radially, the variation in temperature] of the tyre can be also be measured.

Installing these sorts of sensors is somewhat easier than simpler infrared sensors, as the distance between the sensor and the tyre can be far greater, allowing the front tyre sensor to be mounted in the sidepod front or the wing mirror pod. With this set-up, the front tyre temperature can also still be read when some steering lock is applied.

There is also an optical means of infrared temperature sensing, which was first used by McLaren more than ten years ago. Using an infrared camera, the entire surface of a tyre can be seen in real time, and such technology has been used by the FIA for the thermal images seen on TV. As this is a real-time device, the interface is USB and requires a separate data logger to support the volume of data and the USB connection. As with the other types of infrared sensors, installation is relatively easy, as the small unit (61 x 42 x 45 mm) can again be fitted in a mirror or sidepod.

Fig. 1 - The rear tyre temperature sensor is typically recessed into the floor

Written by Craig Scarborough

In order to save time though, the actions with the change mechanism have to be synchronised, and the way to initiate this in a typical flat shift system is to use a gearshift load cell. Positioned in one of two places – either in the change link rod to the transmission or as part of the gear knob on top of the gear stick – the electrical output of the cell, triggered by the forces generated by the driver wishing to change gear, provides a signal to the engine ECU to initiate the sequence of events. Originally used in Indy or Touring cars when not using a paddle shift gear change, these load cell are now found in all sectors of motorsport where flat shift gear changes on sequential gear change systems are required.

When fitted in the change link to the transmission, the load cell is generally a simple strain gauge set-up that measures compressive or tensile loads in the rod. As part of a Wheatstone bridge circuit, and together with signal conditioning and amplification, the electrical output will be a simple, temperature-compensated 0.5-4.5 V linear output that can be fed directly to the relevant pin on the engine ECU. The linear output makes setting it up in the engine control software a simple matter, for once the upper and lower limits are selected, the ECU will initiate the ignition cut as soon as the correct level of force (as set in the software) is obtained in the rod.

In the other option – as part of the gear knob – the arrangement of strain gauges is slightly different, but the voltage output is much the same. Here the strain gauges will be so arranged to measure the bending moment in the gear stick, so they are bonded onto the stainless steel inner sleeve in a different orientation. In both types of load cell, to protect the gauges and offer a more aesthetic product, the whole assembly will be sealed into an outer casing made from aluminium alloy. Trailing out of the sensor will also be a fly lead typically 0.5 m long with three 26 AWG sleeved wires: supply, ground and output.

However, irrespective of which method you use, the minimum force necessary to unload the gear teeth and initiate the gear change will be different in the lower gears than that required at higher speed. The loading between the engaged gear teeth and the speed of the shafts all have an influence on the optimum performance. Where the engine management system allows, a rotary sensor attached to the barrel of the sequential change mechanism can be used to indicate the gear, and the minimum force required fine-tuned to each actual gear change. In this way the minimum time is lost for any and all gear changes.

These sensors may be small but they certainly ‘punch’ well above their weight when it comes to changing gear.

Fig. 1 - Gear lever load sensor (Courtesy of KA Sensors)

Written by John Coxon

Most people who make their own wiring harness have something a little less complicated than a full rally car harness with all the lights and relays, but whatever the application, what useful advice could I offer?

Building your own loom has many advantages of course. Apart from being able to select the terminals that are most convenient for your application, the wiring can be placed wherever’s most convenient for you, along with the connections. Wiring runs can be shorter, making the finished product lighter, and by making the whole thing modular, replacement becomes so much easier if, for instance, part of it is damaged some time later in its life. Making the loom modular also helps if complete assemblies have to be replaced when, say, the engine has to be rebuilt.

Attaching the many sensors to a loom is so much easier with the engine on a purpose-built engine stand, rather than scraping your knuckles in the confined place of the vehicle engine bay, and no matter how much care is taken you can be certain that at least one terminal will be in a position that’s impossible to get at.

If making the loom shorter is a benefit, it may also be possible to make the wiring thinner to save even more weight. Be careful here though, since the reduced voltage drop from the shorter loom will be offset by the higher resistance of the thinner wires, but carefully thought out, significant weight savings (in kilogrammes) can be accumulated.

In doing this, however, much care will be needed to ensure robustness of the loom. For example, tight radii will have to be avoided, and the wires will almost certainly need to be fastened securely around the vehicle frame or engine bay. Although secure such that they can’t move and chafe, they can’t be too rigidly attached so that movement when needed is impossible. Where movement is possible, some prefer to run the wires through individual sleeves, rather like shrinkwrap. This enables the wiring to be located but allows the individual wires to move slightly should they need to.

But the main enemy of all electrical systems are heat and vibration, so routing wires well away from the exhaust system is an obvious decision. Not so obvious, however, is the need to keep wires well away from unsupported componentry cantilevered off other parts where high g loadings can destroy anything attached to it – however firmly – in a matter of minutes.

Above all, keep wires bundled together for strength, and where they are attached to terminals or parts of the engine or chassis, make sure they are protected by heat-shrinkable tubing. That will give added robustness in the areas that need it.

Fig. 1 - Engine bay of a sports racer. Note the obvious lack of wiring which, if you look carefully, can be found towards the middle right in the picture

Written by John Coxon

At its simplest, the throttle pedal exists as an air valve; opening and closing according to the demand by the driver, the air is mixed with fuel and burned to deliver torque to the vehicle wheels. At its most basic level therefore, an electronic fuelling system is a device for delivering torque over a range of engine speeds and loads using a series of ‘maps’ specifying lambda (the air-to-fuel ratio) and ignition timing to give optimum performance. Carefully calibrated and stored electronically in what’s called a look-up table, this torque can be controlled away from its maximum by the simple expedient of retarding the ignition timing, which can be done quickly and accurately. This in principle therefore was the basic building block behind the original engine management systems.

However, as engines become more complex and the complexity of control therefore increases, the process of calibration becomes more onerous, and the more complex things become, the more onerous it gets. In the early days the cylinder charge was affected by the throttle angle, the fuel by the calculated injection time and the engine efficiency by the control of the ignition. If we then add a desire to include boost pressure, camshaft phasing or variable intake systems – not forgetting powertrain control, of course, since many of these interactions will occur simultaneously – it is difficult to predict the overall effect and consequently the output of the engine. And an engine whose torque output is inconsistent does not inspire confidence in the driver.

What is needed is a clear control structure and a system of prioritisation to ensure that the torque demanded by the throttle pedal is the same as that produced at the vehicle wheels. Fortunately, the invention of the electronic throttle – sometimes referred to as a fly- or drive-by-wire throttle – has made all this possible, and we now use for engine control what is commonly known as the torque-based approach.

The first stage in any system is to convert the driver demand into a torque requirement gleaned from the pedal map. A function of engine speed and pedal position, the resulting output is the precise torque to be delivered, and how that is to be delivered will be down to what is known as the system architecture.

Modern engine control can use many different devices to alter the output torque. This architecture must therefore provide the basis by which all engine actuators (throttle position, spark advance, variable intake devices and so on) are coordinated to produce the required output demanded by the pedal map.

Furthermore, control has to be prioritised between fast and slow response in order to achieve the correct transient demand, and all of this must be achieved using the minimum of fuel.

The overall system also includes feed-forward and feedback subsystems. The feed-forward system provides the calculation to produce the estimated desired actuator position to give the torque requested using compressible flow models. Inputs of desired manifold pressure and desired air per cylinder are fed into the computer model to produce the required throttle position commensurate with the required torque. The feedback system then corrects this feed-forward subsystem based on this estimated torque. In the end, the driver should get precisely the torque requested.

While perhaps not as important in most motorsport applications in the past, the new breed of Formula One powertrains will never be able to function without some form of torque-based control.

Fig. 1 - It’s all in this little box

Written by John Coxon

We have covered the ECU in a previous F1-Monitor. The triangular box is often sited inside the cockpit in the ‘V’ below the driver’s legs, while some teams mount it in the sidepod. Being air-cooled, the sidepod location is convenient for cooling and wiring, but being a rather large and moderately heavy device, sidepod mounting requires aerodynamic and centre of gravity compromises.

The ECU is part of a package of devices, the most important being the ignition/injection power box, which provides the outputs for the engine and other electrical systems. Thus the unit is connected to the electrical cut-off switch to isolate the electrics in the event of an accident. The power box also regulates the power supply from the alternator and passes it to the car’s battery. Being connected to the alternator, the power box also provides inputs from the alternator thermocouples and passes them back to the ECU via its CAN bus connection. At some1.5 kg in weight and more than 25 cm long, the Power box is a large item to package, but requires only air cooling.

Supporting the control/data acquisition function of the ECU are various interfaces for specific sensors. A primary interface is the connection to the linear variable displacement transducer (LVDT) sensors, which are used for measuring the damper and thus suspension travel. An interface is required as the sampling rate is high for the suspension; for low frequencies a connection directly via the CAN bus can be used. The LVDT interface unit connects up to four five-wire sensors, and outputs the results through a CAN interface. 

Another sensor-specific interface is the tyre pressure management system (TPMS). With sensors either attached directly to the wheel or as part of the valve, they can send tyre pressure data wirelessly to an aerial mounted centrally on the car, which in turns sends the signals back to the ECU via a CAN interface. As the TPMS is not always fitted for the entire weekend, and its use is often restricted to free practice sessions, the aerial is mounted only temporarily, either on the sidepod or in many cases within a specific wing-mirror housing.

Teams may also fit an additional high speed data recorder for free practice sessions, but this must only take additional inputs or piggyback over existing sensors, as the team cannot disconnect any of the inputs to the SECU. Mounting this unit is troublesome because of its 0.6 kg weight, and its 12 x 15 cm size means it is hard to find space in the sidepods. One team mounts the unit to its front bulkhead, inside the nosecone, for testing and free practice.

The car must also be equipped with an FIA safety data recorder (SDR), which stores data from all the car’s key sensors and two FIA-specific accelerometers. These allow the FIA to use the SDR as an aircraft-style black box recorder to gather data on accidents. Should the car be in a major crash, the SDR will trigger an LED on the cockpit top to warn marshals of the severity of the accident, to allow them to take appropriate care with the driver.

Since its introduction in 2009, KERS has required specific electronic hardware, and with the expansion in 2014 of energy recovery systems (ERS) the requirement for supporting systems is growing. At the core of the ERS electronics is a control unit, which controls the power inverter and battery management. Unlike the other ERS components it manages, the controller is simply air-cooled and mounted either in the sidepod or in a complete ERS pod in the fuel tank area. There are a huge number of other smaller electronic units supporting ERS – I’m told that one 2014 system has more than 30 individual electronic units requiring air cooling.

Unconnected with the car’s operation are the driver radio and FOM camera hardware; these spec systems are provided by the FIA and are mandatory. The driver radio is mounted inside the cockpit, and of course is linked to the earpieces and microphone sewn into the driver’s balaclava. As the pit/driver radio audio is now a key part of Formula One TV coverage, the output from the radio is split and passed to both the FOM camera unit and the aerial for transmission to the team back at the pits.

For the TV feed, the car is mounted with cameras in pods around the car. The feed from the cameras is passed through a conditioning unit before passing into the FOM camera interface unit. This interface also takes feeds from the car’s GPS aerial, the car’s telemetry, external microphones as well as the driver radio. The combined output is then transmitted from an aerial on the front of the car. This allows the TV producers to have the video, audio and data to present on screen. In contrast to the car’s other electronics, the FOM camera interface is the sole area where coaxial cable is used to interconnect devices.

Fig. 1 – Formula One’s ECU was designed to be mounted in the cockpit, but sidepod mounting is also used 

Written by Craig Scarborough

In the past we have had the McLaren ‘spygate’ and the Renault ‘crashgate’ affairs. More recently, and on a more technical note, we have had ‘exhaust blowing’ – either hot or cold. But as I write, the Formula One paddock is all a-rumble about another form of unfair advantage – traction control, and whether Red Bull has devised some devious method of circumnavigating the regulations by modulating the engine output torque at the rear wheels other than via the driver’s right foot. 

The premise would seem to come down to the apparently inescapable logic that since the Red Bull cars are consistently the slowest of the top teams in the speed traps, yet have higher than average speeds along the straights then they must have better traction into, out of or all the way through the curves that precede them.

At this point I assume that few people would disagree. However, of all the teams, Red Bull Racing would seem to be the only one that still has ongoing issues with its KERS system and, like adding 2 and 2 together to make 5, or even 25, the supposition is that the team is somehow using KERS harvesting to modulate the engine output torque to deliver some form of traction control system outside of that banned by the regulations since 2008.

Whenever teams come up with a superior chassis or engine package it gives them something like a 2 s per lap advantage at certain tracks, so it is only natural that tongues start to wag. But the disclosure of the staccato effect of the rubber laid down as Mark Weber’s car was accelerating out of the hairpin at Singapore has only added fuel to the fire, with the claim that the team is exploiting tyre grip up to its maximum of a small amount of slip by continually harvesting KERS into, through and out of the curves, giving the effect of traction control. Modulating the torque so that the tyre would slip slightly – giving maximum traction, then breaking traction altogether, at which point more KERS would be harvested, thus reducing torque to the rear wheels only to re-establish traction, with the whole thing repeated again every few milliseconds – does indeed sound to be highly plausible given modern sophisticated electronic sensors.

Traditional traction control systems involve cutting the spark ignition to individual cylinders and ramping the timing of this spark back in again via the engine ECU and its control software. In the McLaren engine control system for Formula One this function is switched off inside the ECU.

Although many people have asked if such as a system is fully legal, perhaps more importantly it doesn’t seem to be illegal. And plausible doesn’t mean to say it is possible within the regulations as written, or indeed that Red Bull – true to its denial that such a system exists – is exploiting such a system. Whatever the case, it must be certain that other teams developing their new cars will be looking very closely, since control of traction with the new 1.6 litre turbocharged engines will be even more critical for 2014.  

But such is the secrecy in Formula One that perhaps we will only find out when Red Bull technical chief Adrian Newey finally puts down his most famous pencil and folds away surely the only drawing board left in the sport to write his memoirs. Could it be a best seller for 2025 I wonder?

Fig. 1 - Traction control-free zone

Written by John Coxon

In calibrating the engine, engineers can of course plan it so that a precise amount of fuel is injected according to the prevailing conditions in the cylinder at any time, but to ensure that the engine is running at the optimum condition requires some kind of oxygen or lambda sensor.

Developed in the early days of electronic fuel management, lambda sensors are truly only required where fuelling precision is of great importance. Gasoline engines for instance will run quite happily at air-to-fuel ratios anywhere from 10:1 right up to 20:1-plus, but to get best performance then 13:1, depending on the engine, is the right amount, while the best fuel efficiency is obtained nearer 16:1. Mounted in the correct place along the exhaust pipe – neither too far down (where it won’t heat up quickly or become fouled) nor too close (where it might overheat) – the lambda sensor, as part of a feedback loop, is essential for accurate fuelling.

Early sensors were designed to run with three-way catalysts, and as such could only control to stoichiometry, where lambda = 1.0. Operating according to galvanic principles, porous platinum coatings are used to form a sandwich with a solid zirconium dioxide electrolyte separating the exhaust gas from the atmosphere. If the exhaust gas is denuded of oxygen – that is, the mixture is rich – then the cell’s output will be around 0.2 V. As soon as oxygen enters the exhaust and the mixture is lean, this changes more or less instantaneously to 0.8 V. Therefore, in order to pinpoint the precise position of stoichiometry, the control system has to effectively flip-flop between rich and lean settings where the average voltage output will be of the order of 0.5 V.

Referred to as a ‘narrowband’ sensor for obvious reasons, these are no good for competition engines or those that need to work at settings richer or leaner than lambda = 1.0. For these, so-called wideband or broadband sensors have been developed. Similar in a way to the narrowband sensor, wideband sensors incorporate an additional cell where the current output directly indicates the oxygen content of the exhaust gas. Eliminating the flip-flop control of the narrowband sensor, the output is therefore a continuous curve more suitable to modern fuelling control strategies, enabling a much higher level of accuracy when controlling outside of the lambda = 1.0 range. Indeed, special motorsport sensors have evolved that are more capable for the rich (lambda less than 1.0) end of race engine control strategies, offering increased resilience over roadcar units designed more suitably for the lambda = 1.0 and lambda > 1.0 regions.

And what is also quite useful, when fitted with some kind of digital readout, wideband lambda sensors can be used to ‘tune’ engines fitted with good old carburettors. Now that really is something to be thankful for.

Fig. 1 - Lambda sensor 

Written by John Coxon

Instrumentation around the hub is typically to record wheel speed, brake temperature, brake wear and pushrod load. This is repeated for each of the car’s hubs, the sensors and interface units being common to all four corners.

A differential Hall effect non-contact sensor is used to detect wheel speed. Pointed at a ferrous feature on the brake disc bell, such as the disc mounting bolts, as each bolt passes no more than 1.5 mm over the sensor, the sensor output changes in order to calculate wheel speed. Due to the importance of this data, teams will run two wheel speed sensors to safeguard against one failing.

A vital factor in surviving a race is the management of brake temperatures. High brake temperatures leads to oxidisation of the carbon brake disc surface, which in turn leads to excessive wear.  Monitoring brake temperature is achieved with an infrared temperature sensor, which is typically mounted on the carbon fibre brake ducts around the upright, with a clear line of sight at the brake disc’s surface. The sensor picks up the thermal radiation produced by the glowing brake disc and outputs this as an analogue voltage.

It’s also critical with Formula One’s carbon-carbon brakes that disc/pad wear is not excessive.  Thinning brake discs will run hotter, which will in turn accelerate wear, leading eventually to catastrophic failure of the brake disc. Brake wear is monitored by measuring the displacement of the pads in the caliper via a small LVDT mounted between the caliper body and the pad. This has to be a specifically developed LVDT, due to the high temperatures found in such close proximity to the brake discs – even the back of the pad can reach temperatures of 400 C.

Connecting to the back of the upright, the instrumentation for the front suspension pushrod (or pullrod) is also passed through into the wheel. Pushrod loads are recorded for tracking suspension movement, and peak loads are also measured to ensure the safety of the composite suspension elements. An overload could lead to the carbon fibre suspension failing, so a load sensor is built in line with the pushrod, the cabling for which runs down inside the pushrod into hub area, rather than externally into the chassis where the pushrod meets the rocker.

Cables for these sensors need to be fed back into the car’s CANbus. With four cables, each with three to five wires of 22-24SWG, the cable bundle would be significant, and to pass this much though the wishbones – which already carry wheel retaining tethers and brake lines – is impractical. So the sensors are all connected to a Hub Interface Unit, which acts as a ‘satellite’ to the car’s main network and allows connectivity to the ECU via a single cable passing through the wishbone. Measuring just 55 x 39 mm and weighing 55 g, it takes up minimal space inside the wheel. This means the loom can pass back into the chassis loom via a simple two-wire link.

Fig. 1 - Red Bull’s front hub showing the cabling for the brake wear and temperature sensors (Photo: Craig Scarborough)

Written by Craig Scarborough

With modern diagnostics systems, problems with engine management, wiring or sensors are generally easy to diagnose, even when they’re intermittent. However, when the vehicle involved is somewhat elderly then the issue is almost invariably electrical, with the diagnosis remaining slightly more challenging. The particular case I am talking about above was thought at the time to be down to either a broken wire somewhere in the loom, which was not immediately obvious, or more likely a loose or ill-fitting terminal. And when you consider that the vehicle was fitted with a number of 1960s-type spade terminals, those of us familiar with this machine knew just where to look.

The problem with spade-type terminals and others like it is that they are perfect cantilevers. By that I mean that supported at one end, the area with the highest bending moment, the other unsupported end will be free to vibrate. And mounted in a racing machine where the engine is bolted directly into the chassis, the resulting vibration can cause havoc. Attaching the female part of the connection simply makes matters worse because, as well as the uncertainty of the electrical contact between male and female parts, the added weight on the cantilever will create a bigger bending moment at the base of the male part. At the same time this will reduce its natural frequency – and, knowing my luck, just into the frequency band produced by a four-cylinder engine at a speed to create resonance! Once the bending stress at the base of the spade goes beyond its critical limit it won’t be long before the cyclic loading cracks the terminal, and then catastrophic failure follows.

It need not be like this though. All you have to do is look at modern sensor technology and apply a little logic. Modern automotive sensors – the type used to generate input and receive output signals to the engine ECU – no longer have single fixed bayonet-type terminals. Connections are generally made using moulded multi-pin plugs or, where that is not possible, the sensor will be fitted with a flying lead of some description [Fig. 1]. This flying lead will be bonded into the sensor at the critical end and attached to a moulded two-, three- or four-pin plug at the other. Mating with the opposite male or female part of the main wiring loom and clipping firmly together, the resulting connection will be durable and, if the appropriate sealing method is used, watertight. Simple, relatively cheap and above all reliable, especially if the connector is clipped away to a supporting panel somewhere out of harm’s way.

Oh, and in case you wanted to know, the car was repaired in five minutes and, starting last on the grid of 30 finishing a creditable 11th. If only it had been fitted with more modern connectors!

Fig. 1 - A flying lead sensor

Written by John Coxon

The distributor, the mechanical method by which high-tension voltage (up to 45 kV) is applied to each spark plug in turn, is no more. Replaced at first by the double-ended ‘wasted’ spark-ignition coil and, later, individual coils as part of the spark plug cap, the contact breaker points needed to fire the high voltage down to the plug could not cope with currents greater than 5 A. This not only limited the energy stored in the coil but the mechanical nature of the switching also affected the engine speed, durability of the ignition system and the precision of the spark timing. These problems were virtually eliminated when more electronic methods were used.

But for the past 15 years or so virtually all coils in new engine designs have been of the coil-on-plug variety. Originally these had the coil positioned above the spark plug on top of the engine in the centre of a four-valve layout and were referred to as ‘compact’ coils, but in some later designs these coils have been relocated into the recess machined just above the spark plug. With these later types, referred to as ‘pencil’ designs, the coil is positioned inside the cylinder head, minimising the overall engine height and making the whole coil to spark plug component much more compact.

But these ‘pencil’ designs also have limitations, particularly regarding their size and power output. Because of the need to ‘package’ the coil inside the space available, diameters tend to be limited to a minimum of 23 mm, which coincidentally is the size of the socket used to insert and remove the spark plugs. At the same time, in order to dissipate the heat, power, at these diameters outputs tend to be capped at around 80 MJ.

Operating like compact coils with primary and secondary windings, the structure of the ‘pencil’ designs therefore differs totally from compact designs. The coil windings are therefore located centrally around a metallic core with the primary coil outside that of the secondary one. This means that the high-voltage coil is central simplifying the insulation. Furthermore, instead of the ‘soft’ ferromagnetic cores used to concentrate the magnetic fields in older designs, ‘harder’ permanent magnets are used to maintain or possibly increase the magnetic flux density for the application.

The ultimate choice of pencil or compact coil is down to the designer. Where engine height is the limiting factor then the pencil design may be more optimal, but where power is needed then the compact coil version should be the one to go for. Whichever version is chosen, however, gone are the days when we could see a faint blue light tracking along the high tension leads in damp weather.

The distributor may have gone, and the high-tension leads connected to and from it may also have gone, but the challenges nevertheless remain.

Fig. 1 - Comparison of compact and pencil coils

Fig. 2 - ‘Wasted’ spark coil

Written by John Coxon

In recent years this connectivity has been driven by new systems, such as a standard ECU and KERS. There is then a further rise in complexity as the new 2013 ECU has an even greater capacity for sensors. Then next year the cars will have far more complex engines and energy recovery systems, driving even more complication in the cars’ cabling.

Connector technology has progressed immensely over the years; as soon as on-car electronics became common, the need to move from simple spade or block connectors to the multi-pin Military (Mil-spec) type of connectors became apparent. But such are the demands for space and performance in Formula One that the common Mil-spec connectors were soon not up to the job.  Demands for smaller connectors with denser pin arrangements, all with lower weight, drove the connector industry to develop bespoke motorsport ranges during the 1990s.

Most current connectors are made in aluminium, with the connector pins being gold-plated. The mated pair of connectors are secured with a bayonet-type fitting and, once closed; they are environmentally sealed to prevent the ingress of dirt or moisture.

Although they are not made to Mil-spec standards they are still subject to the same tests to ensure they can withstand the harsh environment of moisture, heat and vibration. Unlike Mil-spec connectors they do not use a screw thread fitting to hold the loom to the connector; instead heatshrink adhesive-lined boots are used. For harsher environments, such as the gearbox, vibration-resistant connector variants are available. 

Connector types

Throughout the car, specific connector formats are needed to package the required density of connections in the space available.

The various ECUs, power, ignition and telemetry control boxes all use specialised high-density connectors to interface with the other devices on the CANbus. With the advent of KERS and direct-injector technology, connectors beyond the usual scope of commonly found connectors are required. For these applications new connectors are being developed to cope with the high current demands.

One common connector is the steering wheel-to-column interface. To allow rapid and repeated removal of the wheel, a special connector pair is designed to fit into the end of the steering column and the back of the steering wheel.

A similar requirement is for the nose cone, which is frequently removed for access to the footwell of the car. The electronics inside the nose, such as the Formula One management (FOM) cameras, ride height sensors and pressure sensors need a connection to the main wiring loom, so a connector is required to break the loom between the car and the nose. A specific blind mating connector is used, which allows the refitting of the nose without risk of damaging the connector’s pins, even if the connectors are misaligned in three axes.

Also, the pit umbilical cable has a bespoke connector to mate to, and this is fitted inside the cockpit to allow the car to join the local computer network inside the pit garage.

Inside the fuel tank, special hermetically sealed connectors are used to connect to the various pumps and sensors to the car’s main loom.

The gearbox and its electro-hydraulic control systems demand robust sensors to cope with its harsh environment

Written by Craig Scarborough

In practical terms there are two types of LSD, often referred to as ‘passive’. Torque-sensitive differentials bias the torque across the axle as a function of the torque available at the slipping wheel. Having a fixed bias ratio, the torque sent to the non-slipping wheel will be a fixed ratio of that at the slipping wheel. However, if the slipping wheel is in the air, for instance if it’s bouncing off a kerb, then all traction is lost, and along with it any forward mobility. To offset this a certain amount of pre-load can be introduced, but so as not to affect vehicle handling this has to be limited [Fig. 1].

The other type of passive method is the speed-sensitive system. Biasing a fixed amount of torque proportional to the speed difference across the driving wheels, these have to be ‘tuned’ to the specific requirement of each vehicle, with inevitable compromises depending on the requirement for either good vehicle handling or excellent forward mobility [Fig. 2]. Furthermore, when a vehicle travels through a curve that normally produces a difference in wheel speeds, some level of undesirable torque transfer will also take place, regardless of the external conditions or traction available at the track surface.

Since both these systems have undesirable limitations, optimal vehicle handling can only be attained when the differential is electronically controlled, at which point the differential can be said to be ‘actively’ controlled. 

Consisting of a gerotor-style hydraulic pump that spins when a speed difference between wheels is detected, fluid is drawn out of an accumulator to pressurise a hydraulic system and actuate the operating mechanism of a multi-plate clutch pack in the driveline to one of the driving wheels (on a two-wheel drive system). The hydraulic pressure (and hence the operation of the clutch) is then modulated by some kind of solenoid-controlled spool valve from the pulse width modulated (PWM) signal of an electronic controller (ECU) according to various inputs such as individual wheel speeds, steering angle, throttle position and vehicle speed.

For much of the time the system will need to operate as an open differential since neither of the wheels will be traction-limited, but with increasing speed and load transfer, torque will need to be transferred back to the inside wheel. For low-speed corners, where the wheel speed difference is high, the torque transfer should not be so great, since the vehicle could be plunged into terminal understeer. So, rather than the compromised settings of a passive differential, the highly ‘tuneable’ active version gives just that bit more scope for performance enhancement.

With an ECU for the engine, one for the gearbox and another for the differential, electronic control just got that little bit more complex.

Fig. 1 - Traction effort for torque-sensitive limited slip differentials

Fig. 2 - Torque transfer for a speed-sensitive LSD

Written by John Coxon

The most obvious example of this is when two differing properties of a substance are required at the same time, such as the pressure and temperature of the intake air in an engine where space might be limited. In this case an integrated circuit of, say, a piezo-resistive pressure-sensing element could be combined with a sensor of negative temperature coefficient – a material whose electrical resistance falls with temperature – to monitor the pressure and temperature of the incoming airflow. With the integration providing temperature compensation for the piezo-resistive device, we now have two highly responsive signals for the size of one, and convenient 0-5 V signals to input into the engine ECU. In addition to the reduced number of connectors and wiring needed, with no loss of signal accuracy, the combined unit could well be cheaper too.

While this may not be the world’s most exciting example, contrived to save only weight and perhaps a little money no doubt, a greater and more worthy application of the concept is in safety-critical components and particularly those involved with engine control.

Because of the much greater demands on their efficiency and refinement, modern engines are said to be ‘torque controlled’. To smooth out the fluctuating loads within the engine and deliver the required torque to the driving wheels, the driver’s accelerator pedal has to be connected to the engine throttle valve not by a mechanical link or bourdon cable but by an electrical wire passing the demand signal to the engine ECU. Such systems, now ubiquitous in all modern engines, therefore need a means of signalling the driver demand from the accelerator pedal, a means of opening the throttle plate (using an electric motor) and a means of measuring its position – all at the same time. And if any of the sensors were to fail and, say, the throttle was driven wide open when the driver wanted to brake, the results could be catastrophic!

To combat this eventuality, system designers include a degree of sensor redundancy into the system. Within the pedal or engine throttle sensors there will be two separate rotary potentiometer circuits, each providing a separate signal to the ECU. One circuit might for instance be supplying a 0-5 V (actually more like 0.5-4.5 V) signal while the other circuit might be supplying half this (voltage) but in the opposite sense. In other words, if the signal is increasing on one circuit as the throttle is opened then on the other, as well as being half its maximum, the voltage will fall as the throttle is opened. In this way the engine should always know where it is, and should the worst happen and a sensor fail then catastrophic failure will be totally avoided, with the engine stopping totally – or, more likely, going into ‘limp home’ mode.

Although during my childhood ‘two-into-one’ didn’t go, these days I for one am grateful that it now does.

Fig. 1 - Duplex rotary throttle potentiometer

Written by John Coxon

Connecting up the FIA-spec ECU and power box to all the electronic subsystems, sensors and actuators around the car, means the loom literally extends to every corner of the car. Although most of the major electronic boxes are mounted in the sidepods and the cockpit, the loom also needs to be routed around the powertrain, reach as far as inside each suspension corner, pass inside the front wing itself and climb upwards into the rear wing. The loom even has to be pass up inside the steering column to reach the steering wheel, complete with all its switches and rotary controls.

The cabling system itself tends to be fairly uniform across the grid, in either 55A or the lighter/stronger 55M wire specification. This is a silver-plated copper wire meeting the MIL-W-22759 and MIL-C-27500 standards. For a more flexible loom the individual cables are contra-wound, then sheathed in DR25 sleeves. Where the loom ends in a connector, a heatshrink boot is fitted; this is preferred over the MilSpec screw thread, which is rarely used in Formula One.

As connectors are to be covered in future F1-Monitor article, we will not go into details here, but the wire will be stripped and crimped into the connector blades, as soldering is a less consistent method. 

To ease the mechanics’ task, the loom is separated into several subsections to allow the car to be rapidly disassembled. Thus one large central loom is installed in the monocoque, and this will be connected to smaller sub-looms for the engine, gearbox and each suspension upright. Other looms will be used for other removable or inaccessible areas, such as the fuel tank or the front/rear crash structures.

With so much cabling, the looms’ design needs to focus on size, weight and reliability. Gone are the days when the car chassis was finished and cabling would be retrofitted wherever space could be found; now the loom itself will be designed in 3D CAD along with the rest of the car. Every sensor, actuator and electronic box will be positioned on the CAD model to allow the loom designer to plan for every cable, taking into account space for bends, supporting structures and connectors.  Sufficiently large apertures need to designed into the monocoque at an early stage to allow the loom to pass through. 

Bespoke carbon fibre cable guides and junction boxes are made to ensure the loom is supported throughout its length. Owing to the heat around many parts of the car, the loom needs careful routing, and where diversion is not possible then sheathing with heat-reflective sleeving is used instead.

Fig. 1 - Every black box and its associated wiring is positioned in the master 3D CAD model at an early stage (Photo: Craig Scarborough)

Written by Craig Scarborough

One way of course to rid ourselves of this parasite (for that’s what it is) is to dispense with as much wiring as possible – shorter wiring ‘runs’, thinner wires, even omitting the wires totally if possible. No doubt few of us will be as drastic as this, but when the loom has been pared down to the minimum, I would argue that one ought to consider reducing the size or weight of the connectors themselves. The modern racecar though, much like its road-going contemporary, is festooned with electronics, so however well intentioned our wishes may be, the challenge to minimise electrical wiring is one that few will relish.

But choosing the correct connector is not just about weight. To finish first in any competition, first you have to finish, and so product performance, reliability and suitability for the task intended are surely the primary requirements. Thus the connector not only has to cope with the expected currents – including any unexpected ‘surge’ currents – it also has to do it with the minimum voltage drop across the electrical circuit, and do so without failing in any number of ways. Racecars are well known for their high levels of teeth-chattering vibration, and the temperature environment of a typical race engine will challenge any electrical component, especially if for reasons of necessity it is mounted too close to the exhaust. Vibration can cause connectors to work themselves apart sometimes, so some form of locking feature is also usually essential.

The hydrocarbon fluids of oil and fuel are never too far away either. Not only do they have low surface tension characteristics, ensuring that they will find their way into the smallest of crevices, but more often than not, given time, they might react with the insulating materials to either soften or harden them depending on any one of the complex chemical reactions possible.

Moisture also is a great enemy of the wiring loom. Once present, galvanic cells may be created, which in turn produce chemical corrosion, eventually reducing the effectiveness of the electrical connection. Gold-plating the connector pins is clearly a must for these challenging conditions, but that inevitably adds to the cost in these budget-conscious times. Sealing against the ingress of moisture in the first place is also clearly a good idea, and while shrinkwrap assists and strengthens the wiring into the back of the connector, detail design of the shrouding around the mating male and female parts of the connector itself will preserve the pins inside. And in all this designers still want that extra pin out of the ECU or an extra channel into the data logger, as well as the growing requirement of redundancy in sensor technology.

Reducing the size of the electrical connectors may be desirable to save weight, but as part of the overall criteria in any electrical circuit it comes fairly low on the list.

Fig. 1 - Automotive electrical connectors

Written by John Coxon

Thinking on though, in high-level motorsport perhaps the holistic approach is precisely what is needed. In medicine, ‘holistic’ normally refers to treating the entire person rather than the simple physical symptoms of a disease, and in a modern formula car the vehicle ECU is very much comparable to that of the brain in a human. First it receives coded messages from various sensors, the motor and exterior surfaces, and converts them into some kind of binary language. It then computes a response, sending messages back along the same communication system to other areas of the car, where it is decoded and acted on. In a way this mimics nature, and using the best available technology should be what motorsport at this level should be all about.

So it was encouraging when the individual concerned went on to say that racecar designers see a car as an entity, rather than in terms of separate systems that are subsequently bolted together. This latter approach is what we tend to find in road cars. No, the designer of a racecar looks at things differently and seeks to create an advantage over the opposition, which is why the system processors used in the latest generation of Formula One ECUs are the best available for the job and will never be found anywhere near any other type of transport application.

These processors were actually designed for network switching applications, and consequently the new Formula One ECU, the TAG320, controls everything on board the vehicle. The driver may steer the vehicle using a mechanical system (the steering wheel) and stop it using an hydraulic system, but everything else – throttle, clutch, gearbox, fuel pumps (of which there are many), differential and all the data passing back and forth – is controlled by the ECU. So while there may be many other boxes around the vehicle (including one at every corner taking data to and from each wheel system), all these are connected to the central ‘brain’ of the system via four wires – power, ground and two CAN lines.

But while the latest ECU, in use for the 2013 Formula One season, will work faster, handle more data and offer more control power than the previous version, the true benefit may not be visible until the start of the first race. You see, the older ECU’s architecture was such that you could never be absolutely sure that certain engine input/output data was not being used to operate illegal launch or traction control strategies. By completely segregating every single application and limiting the input and output signals to a specific application, launch and traction control can be completely eradicated in future and any accusations of cheating consigned to history.

An holistic approach it may be, but at the same time it just might be the tonic that the patient requires.

Fig. 1 - 2012 Formula One ECU (Courtesy of Mclaren Electronic Systems)

Written by John Coxon

2013 will bring in a new ECU to control the Formula One powertrains. Provided to all teams via McLaren Electronic Systems, the TAG-320 is the first step towards the major change in powertrain rules for 2014.

As a method of controlling costs and driver aids, the FIA elected to introduce a Single ECU (SECU) supplier into Formula One in 2006. With suppliers invited to bid for the contract, it was McLaren Electronic Systems (MES) who won the tender process; MES duly introduced its TAG-310B ECU into the sport, which controlled both engine and chassis functions. 

For some customer-engined teams, this was the first time they had an integrated ECU to work with. The common practice beforehand was that the engine ECU was supplied by the engine manufacturer, with chassis and data logging functions being managed by a separate ECU supplied by the team.

After some six years now the TAG-310B will be replaced by MES' next generation of ECU, the TAG-320. Designed specifically for the demands of Formula One and in particular the change in powertrain rules for 2014, the unit will therefore no longer support ten cylinder engines but will support the advanced energy recovery systems for next year.

The main advance over the TAG-310B is the processing performance and the number of interfaces. At 4000 MIPS (millions of instructions per second) its processing power is now more than double the outgoing model, with RAM increasing to 512 Mbytes. There are more sensor inputs and CAN interfaces, sampling rates and logging channels have risen nearly tenfold and the unit gains an internal accelerometer. Even the internal data logging capacity is higher, from 1 to 8 Gbytes, while there is also a slightly greater range of control outputs. 

This allows the unit to manage far more controls with greater accuracy, which will be critical with complex high-speed systems such as the seamless-shift gearbox, direct injection and the advanced energy recovery systems in 2014.

Physically the unit retains the unique shape of the unit used by McLaren Mercedes, which predates the 2006 SECU. The triangular layout allowed McLaren to fit the ECU under the driver’s thighs, on the floor behind the dash bulkhead. The pointed end of the ECU faces forwards, making it easy to direct the wiring loom inside the cockpit towards the three CAN interfaces on the back of the ECU. Some teams still choose to mount the SECU in this position under the driver’s thighs, but many teams now fit it in the sidepod, to allow cooling air to be diverted into it from the radiator ducts. For easier packaging the TAG-320 retains the same external dimensions at the TAG-310B.

Lastly, the transition to the TAG-320 is aided by compatibility in the code and the external applications.  The code written by the teams to control the chassis and powertrain functions on the TAG-310B will be compatible with the TAG-320, while the two key applications – SystemMonitor, which configures the control systems, and Atlas, which teams use to analyse the telemetry data – are also consistent between both ECUs, easing the team’s software department’s transition to the new ECU. 

Although the ECU’s increased speed and greater number of interfaces are aimed very much at the complex powertrains of 2014, teams will have a year to bed-in with the unit with the current powertrains during 2013, although it must be said there were no reliability issues reported with the TAG-310B.

So we can expect a seamless switch to the new unit when Formula One racing resumes in March.

Fig. 1 - The new TAG320 ECU is the same physical size as the outgoing TAG-310B, but has far greater performance (Courtesy of McLaren Electronic Systems)

Written by Craig Scarborough

These days of course, many gearboxes have barrel gear selector mechanisms and electronic control such that changing up in the gearbox no longer requires hand-to-foot coordination, and the simple flick of a paddle moves to the next gear with only a momentary cut of the ignition and hence forward progress. Systems exist that reduce this ignition cut time to as little as 20 ms, and while some may claim it is ’seamless’ as far as the driver is concerned, with the overall lap time saved counted in some cases in many seconds, technically I might take issue. During all this the clutch remains untouched, and the driving force between the gear teeth is momentarily reduced by the action of the ignition cut.

The real test of gearbox control, however, must surely be in the downshift. Since the driver is slowing the vehicle on the brakes there is less of an inherent advantage in terms of lap time saved, but in synchronising the input and output shafts as we move down the box, just like the practice of ‘heel and toeing’ in years past, if it’s done correctly it can spare the gearbox much abuse and hence stand a greater chance of lasting the race.

Considering the use of motorcycle engines and gearboxes into hill climb or track day cars, during a braking event when the driver wants to change down, even in the simplest of systems the gearbox control unit (GCU) will first check to ensure that first gear is not already selected. If this is the case, the down change will be aborted and will prevent the driver from selecting neutral in a typical motorcycle gearbox when downshifting at slow speeds.

Following a demand from the gear change paddle to change down, the clutch will be powered open. Once open, the gear actuator will be powered and a short delay introduced as the next gear is selected. At this point, and depending on the gear selected, the throttle or throttles will be opened again using what is generally known as a ‘throttle blipper’. This revs the engine to synchronise the road speed to the engine speed before powering the clutch closed again to complete the gear change.

Using sensors to determine the relative shaft speeds, gear changes can be aborted if the engine is likely to be over-revved, but above all by taking care of the complete gear-change process the driver is left with two hands on the wheel and can concentrate more on the braking of the vehicle and the correct line through the corner. Many aftermarket systems can use pneumatic actuators because of their forgiving nature, but more often solenoid-operated systems can be more sympathetic to the forces needed and better controlled by the electronic GCU.

Systems of this nature may not be of the complexity of the real seamless shift variety, but for thousands of racers using motorcycle-based engine-gearbox combinations the results can be highly satisfying.

But not as much as I would think of the perfect ‘heel and toe’.

Fig. 1 - Typical gear change paddle

Written by John Coxon

However, the techniques for determining such data, once solely the province of the research laboratory, are now widely available to the point where even the humblest of engine development establishments will routinely produce it on a regular and ongoing basis. Referred to simply as 'combustion analysis', the heart of the technology lies in the measurement of the precise pressure inside the combustion chamber at every 0.1 crank angle degree or so throughout the cycle. For this, and apart from an accurate crank angle encoder, pressure transducers capable of measuring accurately up to 300 bar have to be incorporated into the engine with the ability to withstand up to 500 C and have an output signal that varies little with temperature.

While many pressure measurement transducers still use strain gauge technologies, when high combustion loads are to be recorded at such demanding temperatures the only practical way is to use a piezoelectric sensor ,which at its core uses a single crystal measuring element. If a load or pressure is applied to this crystal, an electric charge proportional to the load is produced, which when amplified and converted to a 0-10 V dc output can be fed into any suitable data logger alongside the associated crank angle encoded position. The drawback to this electric charge is that it soon dissipates, so piezoelectric transducers, while excellent for transient pressure measurement, are not so useful if pressures are quasi-static.

The working element of the transducer uses a single quartz (silicon dioxide), gallium orthophosphate (GaPO4) or similar crystal, and will have been grown from scratch under carefully controlled conditions taking many months. Once of sufficient size (and since the orientation of the lattice is important to the properties of the sensor) the crystal ingot will be carefully X-rayed to determine the major axis and then carefully sliced into thin wafers. Depending on the crystal type, the result will be a compact transducer of perhaps no more than 4 mm in diameter at its base but with a pressure sensitivity of 2-12 pico Coulombs per Newton (pC/N) and which is reasonably insensitive to variations in temperature throughout its designed operating range (generally 0-500 C).

In the world of 'combustion analysis' such transducers invariably have to be small since space around the combustion chamber is invariably tight. Getting the transducer close to the chamber to obtain the best possible signal is therefore not easy. In some cases, and where access to the casting patterns is available, certain cylinder head cores may need to be 'rubbed' to produce just that little bit extra casting thickness. This is necessary to machine a suitable access point for the transducer and not have to pass through the cooling jacket. In other cases, for instance when modifying more standard cylinder head castings, this might not be possible and transducers have to be obtained that can 'bridge' these waterways, against the coolant at both the lower and upper ends. Since in my experience these are rarely successful in high-performance engines, other methods such as welding in special sleeves may have to be developed.

While not always as easy as many would have you believe, once combustion pressure data is available you wonder how you ever managed before without it.

Fig. 1 - Transducer installation methods

Fig. 2 - Typical transducer

Written by John Coxon

Electrical gremlins can be behind many a DNF, and because they are often so apparently minor - a broken sensor or chaffed wire for instance - they are all the more frustrating. But some of them are avoidable. Better preparation is the watchword here, but for others perhaps the answer is a power control module (PCM).

A PCM is a solid-state module that replaces relays and circuit breakers, and as a kind of intelligent controller it can protect your electrical circuitry much as a fuse would. But a PCM goes much further than that. Unlike a fuse, a PCM can decide if the problem was just a spike in the current supply and switch the circuit back on again or, if the sensor or its circuitry was terminal, decide on some other course of action. A rally car for instance, using several hundred amps of forward lighting to find the way, can shut down just a sufficient number of the lights so that a failing generator can power the engine management. Without such a device maybe the generator would overload and fail all the sooner, leaving you with just the battery to get you home (or not).

But a PCM is even more than that. It can exchange data with all the other power modules around the car and can monitor the activity of the complete electrical system to even out the current supply and avoid unnecessary or unproductive current drain. Take for instance the initial cranking and firing of an engine. With the starter taking many hundreds of amps, the power to all other circuits (including the engine management) can be shut down until the engine has reached the required cranking speed. At that point the EMS can be switched back on again and the engine fired.

Sensibly integrated with rest of the electrical system, the installation of a PCM can even save weight, replacing maybe not miles of untidy wiring but a considerable amount. And if you've ever weighed an engine or vehicle wiring loom then you'll know that the weight is often recorded in kilogrammes not grammes. The provision of something more than just an intelligent circuit breaker can therefore often save you weight. Unlike a fuse, however, for which you need to have access, a PCM can be installed safely out of harm's way and in a place that is most beneficial to the vehicle and not the driver and/or his mechanic.

It may be fashionable to have the latest piece of electronic gizmo in the car but with a PCM it can only make sense.

Fig. 1 - The power control module

Written by John Coxon

The necessity of changing plugs in those days was dictated by the somewhat narrower heat ranges of the plugs on offer. Attempts at firing from cold on the racing plugs was never attempted, for it was certain to foul the plugs through excessive oiling while cold or through excessively rich mixtures. The ability to change plugs in an instant and diagnose misfiring cylinders quickly therefore became second nature.

Racers these days may have never had it so good with the advent of the wide heat-range plug but even so, the ability of any engine builder to select the correct plug for the task is still a skill worth having.

Essentially it's all down to the design of the plug, and its ability to retain the heat of combustion at low speed and dissipate it as the speed rises. If you like, a spark plug is a compromise: if it dissipates heat too easily - which, paradoxically, we call 'hot' or cold-running plugs - then combustion deposits will build up on the electrodes, causing the plug to 'shunt' or lose the power of its spark. If it doesn't dissipate the heat fast enough (a 'cold', or hot-running plug) then the electrode may overheat and introduce pre-ignition to the combustion mixture, resulting in reduced performance, electrode wear and possibly even engine damage. The trick therefore for any engine tuner is to keep the spark plug temperature in the 500-900 C range or thereabouts throughout.

Given thermocoupled spark plugs and suitable instrumentation, determining the correct temperature range is easy enough, but a skilled engine tuner can actually 'read' the plug and extract so much more than just its temperature. For instance, looking at the annealing zone on the ground electrode can give not only an indication of the heat in the plug at the time, but viewing the condition of the porcelain centre cover can also indicate if the ignition timing is correct. Running what is generally known as a 'plug cut' - when the engine is switched off at wide-open throttle - little black specks can indicate a plug that's slightly too hot or an engine consuming oil. On the other hand, silver specks on the porcelain can indicate the onset of detonation, something that is often difficult to detect for the inexperienced. Retarding the engine slightly and checking the ground electrode for heat range might seem to be a good idea at this point.

But 'plug cuts' are also well known for reading the mixture. With modern, cleaner-burning race fuels, however, this is more difficult than it once was, and the shades of grey or brown on new plugs that will need to be used have to be carefully interpreted against previous experience.

There's more to a spark plug than you might at first think.

Written by John Coxon

Likewise the Hall effect sensor can be used to detect the position of the crankshaft relative to the point of mixture ignition and, when counted over time, used to indicate the average rotational speed. Not only can this be used in the ignition and fuelling map, when added cumulatively the data can be useful in switching events or fuelling strategies. The transfer between initial cranking and cold-start fuelling strategies is perhaps an example of this.

One of the best and most ingenious examples of this approach, strangely enough also using a Hall effect device, is to be found on the latest mechanical flywheel energy storage device being used at Le Mans this year. The unit, as installed in an LM P1 car, is mounted between the engine and gearbox, and complete with all its ancillaries - hydraulic pumps, lube pumps, scavenge pumps, oil tank, vac pump and control valves - it weighs 37.9 kg

The control system, an in-house design, relies on two Hall effect high-speed sensors, and generates a huge amount of data. But if you imagine that all that is generated is rpm then think again.

Originally the energy-storing flywheel was connected to the driveline via a continuously variable transmission, or VCT. The latest version, however, is connected to the gearbox input shaft using a system of three gears clutched in and out. Using a six-speed gearbox this means 18 gear ratios between the engine input speed and the road wheel speed.

Considering the role of the two sensors, the designer offers this explanation. "The first measures the flywheel speed, and hence its energy content, while at the same time the rate of change of this is proportional to the power transmitted. If you know the speed and the power then the shaft torque can be calculated. At the other side of the clutch we have the other speed sensor, and since all three shafts are geared together we know how fast each is rotating and hence the speed difference across the clutch at any instant."

Knowing the energy in and that coming out gives an accurate indication of the energy dissipated through the clutch, and when mapped against the coolant flow rates and the heat transfer characteristics, the instantaneous clutch temperature can be derived. In theory, therefore, so long as this temperature is kept below a certain figure, the clutch controlling the energy flow in and out of the flywheel, will never burn out.
Not bad for a couple of inexpensive 'off-the-shelf' speed sensors as well as a lot of painstaking development.

Fig. 1 - Energy storage using a clutched flywheel and a couple of simple sensors

Written by John Coxon

Designed to provide the engine control ECU with information relating to how fast the engine is rotating as well as the position of the crank (or cam) throughout its cycle, the crankshaft sensor is essential in determining the amount of fuel to be injected as well as the timing of its ignition. Typically using some form of disc or wheel with slots cut into it at known angles relative to the ignition firing of individual cylinders, the two most common sensor technologies used are either impedance-based, known generally as variable reluctance (VR), or those based on the Hall effect. But if both rely on sensing the instantaneous changes in a magnetic field, that is surely the limit of their similarity.

The simplest form of VR sensor consists of a coil of wire wrapped around a magnetic former. As the slots in the toothed disc or wheel pass the magnet, the flux surrounding it changes and causes an EMF (electromotive force) to be generated in the coil; the frequency with which this is generated is proportional to the speed of rotation of the disc or wheel. The higher the rotational speed the greater the amplitude of the signal, which is fine for high-speed engines but at low speed the circuitry to produce the clean signal necessary for the engine ECU is difficult to design. So despite the low cost of the sensor, the rugged nature of its construction and its extreme high-temperature capability (up to 300 C in some cases), VR sensors are increasingly being replaced by those that use the Hall effect.

Using a principle discovered by 19th century physicist Edwin Hall, the Hall effect sensor relies on a semiconductor element between the magnet and the toothed wheel. Sensing the change of magnetic flux, rather than the rate of change as in the VR method, the Hall effect sensor not only measures the crankshaft speed in terms of revolutions per minute but actually counts individual revolutions.

That makes it particularly useful if engines are running slowly - as in the case of initial cranking, say, or in setting the ignition timing of the engine when the digital on-off nature of the signal makes this much more precise when the engine is rotated by hand. With semiconductors at the heart of the sensor, signal conditioning is generally also incorporated, so the output at the connector will be compatible with the digital logic used in the engine's ECU. The presence of these electronics, however, makes the unit less robust than its VR counterpart, so Hall effect sensors are usually rated only up to 150 C.

Being less sensitive to electromagnetic interference as well, the Hall effect makes modern engine management so much more functional, despite having been around for more than 140 years.

Fig. 1 - An older style VR pick-up

Written by John Coxon

The CAN (Controller Area Network) bus is effectively a pair of twisted wires, one carrying a 'high' voltage and the other a 'low' voltage, which transmit a series of digital messages according to a given protocol. By connecting sensors, actuators and their controlling devices (ECUs) with messages loaded and downloaded at various 'nodes' along the way, what were once bulky vehicle electric harnesses have been reduced to but a single cable.

First defined in 1984 by component supplier Robert Bosch, and incorporated into ISO standard 11898 in 1994, since the mid-2000s CANbus has been fitted to just about every new passenger car, providing a single conduit for data transfer across the vehicle. The problem is that with a maximum transmission rate typically of the order of 1 Mbit/s, there is only so much information it can carry. Using a system whereby the priority of the message varies according to a designation given once the bus is operating above around 35% of its overall capacity, statistically there is no guarantee that the signal will be received at its destination in a timely manner if other, higher priority data is in the way.

A simple analogy is that of a car joining a busy road at a 'T' junction. The car can only access the busy road at a given time when there is space for it. If there is no space then any request to join the carriageway will be denied, and repeated requests ignored until there is either space or an approaching car (message) with a lower priority.

For time-sensitive control data (as in a powertrain) clearly this is a worry since the data about, for instance, an ignition timing pulse on an engine running at high speed needs to be delivered precisely at the optimum time, not five or even ten crankshaft degrees later. Of course, where a lot of time-sensitive data is used, multiple CANbus systems can be (and are) most helpful and, linked together, offer what many control engineers might say is an inexpensive if not very elegant solution.

A much better approach is Flexray. Described as a "time-deterministic protocol for X-by-wire applications", Flexray typically operates at up to ten times the speed of the fastest CANbus and, because of that, the data messages can be time triggered within a matrix according to their frequency and urgency.

It may not be as swashbuckling as Buck Rogers or capture the imagination as Flash Gordon does - for some people at least - but in the next five to ten years Flexray, as science fact, will surely make its presence felt in more ways than we can even guess at the moment.

Fig. 1 - Time-triggered network principles

Written by John Coxon

been the norm) but the maximum fuel flow rate will no doubt have ramifications, since measuring gasoline fuel flow accurately onboard a vehicle has never been that easy.

In a test bed environment, where it is somewhat easier to control the environment, the instrument of choice was at one time the fuel balance. It was a system whereby the time taken for fuel (cooled to precisely the correct temperature) to empty a small chamber perched on a simple balance arrangement was measured, and average fuel consumption calculated, and it was both accurate and repeatable. Replaced by the coriolis device (see RET-Monitor, July 2011) having fewer moving parts, both these are bulky and therefore totally unsuited for in-vehicle installations.

For in-vehicle use, until recently perhaps, the main choice would have been a turbine flow meter. Consisting of a stainless steel, low-inertia turbine wheel or rotor within a non-magnetic housing, the speed of the turbine rotation is recorded by a magnetic pick-up attached to the side of the device. The AC voltage output generated is then proportional to the fluid flow rate, while the total number of pulses counted gives an indication of total mass flow. Since the rotor is immersed totally in the flow, turbine flow meters can exert a small but definite back pressure within the fluid, and as the flow needs to be laminar, great care is needed not to introduce turbulent flow immediately upstream.

Although highly reliable and accurate throughout most of their steady-state operating range, the mechanical nature of these devices and the inevitable friction within their bearings means that at low flows their accuracy can be impaired. In highly pulsating flows, such as those found in engine fuel systems, the inertia of the turbine wheel - however small - may lead to inaccuracies, and the fact that turbine wheels can jam and that the local pressure variations around the turbine blades under some conditions can create vapour lock does not truly recommend them to regular in-vehicle use.

However, the coming emphasis on fuel flow measurement in motorsport has encouraged the development of new types of flow meters, this time using ultrasonic technology. With no internal moving parts, and therefore no restrictions on the flow and its pulsating nature, the latest such devices can measure a wider range of flows with minimal impact on the pressures in the system.

Anodised aluminium for lightness and corrosion resistant to aggressive fuels (like ethanol), ultrasonic meters can measure the flow several thousand times per second, giving them the sensitivity to capture, so it is said, individual fuel injector flows at low engine speeds. Although not designed specifically for motorsport, such flow meters will have a much wider application where accurate fluid flow rates in hostile conditions are required.

Although it was possibly not their intention, the rule makers in Formula One may have accidentally spurred the way forward for many other categories to adopt fuel flow rate regulations and be relevant not just to the wider automotive world but to many other industries outside the sport.

Fig. 1 - 2014 Formula One fuel flow envelope (revised 10/02/12 @ 15:10)

Written by John Coxon

First, second, third, fourth - the gears would fly by as I accelerated the class IV gearbox machine into the distance and with the humble Villiers 9E engine and equally humble four-speed gearbox I could easily cope. However, when moving on to the more modern 125 Yamaha, six-speed unit, things became more difficult. Moving the lever up to change up and then down to change down was never the problem; it was simply remembering which gear you were in out of the six - was that fourth or fifth? Without a rev counter (for in those days such things were rare), in the heat of the moment there was no way of knowing; and anyway, in the next split-second or so, you were ready to change up (or down) and so there was no time to worry.

Many years later, and in most forms of motorsport today, the four-speed gearbox is but an ancient memory, and six speeds are the accepted norm. In Formula One of course, transmissions have seven speeds, and in 2014 will move up a gear (if you'll pardon the expression) to a mandatory eight. It is little wonder therefore that it is now considered imperative -not only for the driver but also for the engine controller - to know which gear we are in.

In theory, of course, there is little difficulty in working it all out - a simple tailshaft sensor to compare its speed with that of the engine speed can tell you almost immediately. The clutch has to be engaged and the vehicle moving, and the calculation will take but a few milliseconds. Indeed, many aftermarket systems incorporate this as part of their standard package. Another way is to mimic the driver by counting up the box using an inline microswitch in the gear change system, and down the box in a similar way. Storing this information digitally in its memory, the ECU should know where it is at any time, even if the car is stationary.

By far the best method, however, is to mount a rotary potentiometer at the end of the gearbox change barrel. Similar in many ways to the unit mounted at the end of the throttle shaft to determine the degree of throttle opening, rotary potentiometers are simply rotary voltage dividers calibrated to the exact position of the selector barrel. Indicating the position of each gear, in the more sophisticated systems the system can also plot its position at all stages through the gear-changing process.

In sequential systems, neutral can appear in one of two places. Motorcycle-derived gearboxes invariably have neutral between first and second gears, while vehicles that are devoid of any historical precedent tend to slot this in at the beginning. Thus, for a motorcycle, the arrangement on the barrel may be 1-N-2-3-, while for a car it is more likely to be N-1-2-3-. However, knowing where on the barrel the gear should be is one thing, knowing where it actually is, is another, since mechanical wear and the random movement of parts can affect this to a small degree. Limits to the position of the gear therefore have to be added.

It goes without saying that the rotation of the potentiometer - be it 90º, 180º or 360º - must be at least equal to the rotary movement of the barrel. And while the mechanical wiper technology of traditional potentiometers can be affected by the high levels of shock and vibration, not to mention the extreme temperature experienced by a gearbox, newer non-contacting rotary sensors are now available that use Hall effect technology. With these sensors also accommodating duel redundant options, offering the ability for the sensor to be used in two modes - control and position monitoring, or control and gear indicator - there seems to little excuse for not knowing which gear you are in.

But there again, why make it easy for yourself?

Fig. 1 - Non-contacting dual independent output rotary sensor

Written by John Coxon

Apparently simple at first but increasingly complex when we start delving into it, electromagnetic radiation exhibits both wavelike and particulate properties as one of a transverse oscillating wave of electric as well as magnetic fields, and whose oscillations are at right angles to the direction of the energy transfer. These electric and magnetic components oscillate perpendicular to each other in phase and at right angles to the direction of travel.

But unlike all the other forms of pollution mentioned, not all electromagnetic radiation is undesirable. Light, for instance, is a perfect example, as are the radio waves that increasingly energise our communications networks, mobile phone telemetry systems and the like. No, the major concerns are over the forms of electromagnetic waves inadvertently produced and captured by our electronic systems, with sometimes devastating results. One such system historically at risk is the engine management when early systems could be corrupted by unwanted radio waves.

I remember one such incident about 30 years ago when a number of test vehicles undergoing mileage accumulation broke down on the same section of road within a short time of each other. Recovering them back to base, the engine was cranked and started immediately, giving no indication of a fault and running perfectly thereafter. It was only when the vehicles travelled down that certain section of road (which was close to a high-power radio mast) that the problem re-occurred. Eventually the problem was traced and demonstrated to all by flooding the engine bay of the car with high-frequency radio waves in a specially constructed chamber: the engine cut out and stalled every time the radio power was increased.

The problem with that particular installation, so I was informed, was that the control system wiring loom - which consisted of long, straight wires in the form of a loop and assembled around the engine bay - acted much as a radio aerial. Picking up the electromagnetic waves generated by the radio mast, as your car radio would, only served to confuse the early engine control electronics, and eventually caused the engine to stop.

Since then of course, much legislation has been introduced to ensure that these kinds of events no longer happen. Vehicles intended for the road have strict requirements for both electromagnetic susceptibility and, equally important, electromagnetic emission. Vehicles intended for the track or totally off-road may be exempt from on-road legislation, but with the advent of hybrid technology and more powerful kinetic energy recovery systems placing large amounts of electrical or electronic circuitry into confined spaces, the issue of interference, or 'crosstalk' - no, not the art of witty conversation - may be more relevant than ever.

High frequencies, large currents and the size of the loops of wiring therein, and the inevitable closeness of proximity of systems, make for the perfect breeding ground for electromagnetic interference. Once recognised, however, steps can be taken to counteract it - unlike of course the early practitioners of electronic engine control, who were more than just a little perplexed at the time.

Fig. 1 - Representation of an electromagnetic wave in 3D, showing a linearly polarised plane propagating from left to right

Written by John Coxon

One such decision to be made by anyone undertaking any form of vehicle wiring is whether to solder or crimp.

Let me start by saying that virtually without exception, the motorsport industry, the aviation industry, the automotive industry and even the marine industry all seem to adopt crimping as the norm. Safe and secure crimps produced using high-quality crimping tooling sometimes costing many hundreds of pounds and used by trained individuals to a set and closely monitored procedure can rarely be faulted. And with gauging technology to determine any wear in the tooling above an acceptable amount, it is perhaps easy to see why the industry has adopted these methods.

But ask any electronics engineer and they will readily admit that the highest quality of any electrical connection is best served using a soldered joint. After all, we still connect electronic components to a printed circuit board using solder, and if failures in these are now somewhere nearer one in 100,000 or greater, it surely can't be the actual fault of the solder. Or can it?

A typical electrical connection on a vehicle, aircraft or boat requires three things - low electrical resistance, good mechanical strength and protection against the environment. While low electrical resistance may seem to many to be the overwhelming necessity, in actuality the largest issue is likely to be one of mechanical strength. An electrical lead attached to any connector is in fact a cantilever, and like any cantilever it is therefore subject to all manner of vibration-induced bending moments. In a crimped connection, the central electrical connection will be crimped to the wire with a separate outer grip crimped to the protecting elastomer outer sheath. The former provides adequate electrical connectivity but the latter restrains the sheath.

In soldered joints the mechanical security of the joint is obtained by the 'wetting' action of the tin within the solder on the substrate beneath, but capillary action with the wire strands will draw the molten metal up away from the joint. So while we may have a perfect electrical connection, the hardened solder within the wire will create a weak point, and the wire will eventually break due to fatigue.

The application of heat, even if done so carefully, can also soften the outer sheath and may damage its properties, further impairing its flexibility. Using traditional tin-lead solders (37% lead, 63% tin) with a melting point of 183 C, soldering is difficult enough. Now that lead in solder is outlawed and a new lead-free product - tin-antimony or tin-copper, to name but two - will have to be used, this melting point will be higher, and the risk of damage to the protecting sheath therefore greater.

So it seems logical that crimping is the way to go.

And me? Well I have a copious supply of tin-lead solder and I like the apparent security of a properly soldered joint, despite the difficulties and the risk. Using crimped joints I apply just the smallest amount of solder on the first electrical crimp and use the second outer crimp for mechanical security, ensuring of course that all soldering fluxes are carefully removed afterwards. It's a bit of a 'belt-and-braces' or 'hedging my bets' approach, but it always seems to work for me. There again though, with my record in making decisions, what do I know?

Fig. 1 - A selection of inexpensive crimping tools

Fig. 2 - Solder or crimp?

Written by John Coxon

The loggers of the past were large and unwieldy devices often strapped in the back of the vehicle somewhere using only a very few non-programmable, dedicated channels to record temperatures and, if you were lucky, pressures and speeds. All these were carefully calibrated in the electronics laboratory by a group of dedicated technicians, and the information generated often spewed out in the form of a chart or simply a list of numbers. How things have changed.

Since then the technology of data acquisition has taken over where the simple datalogger left off, and when at one time we were only interested in a handful of channels, today 50, 70 or even 500 is perhaps unusual but not that uncommon. The problem today is not so much acquisition of the data but knowing what you have and using it intelligently afterwards. But that aside, with only 50 analogue channels feeding directly into the back of a logger, that is still a fairly sizeable cable.

The great revolution in the data acquisition world to date is surely the concept of the data bus, a method of converting the various data channels into digital form and collecting the signals using what amounts to just a couple of wires; the data itself is converted into a series of noughts and ones transmitted as a series of square-wave voltage pulses. Immortalised in the familiar RS-232 standard of computing fame, many data acquisition systems still use this as a means of capturing information from an engine ECU. However, the increasing amount of electronics on board modern vehicles, and the need to communicate with more than one ECU, has led to the principle of the CAN bus, which these days is in use just about everywhere where high-speed machinery data needs to be acquired and processed quickly and reliably.

Originally devised by engine component supplier Bosch, the CAN (Controller Area Networking) bus is more than just way of transmitting data, it is a complete communication system linking sensors, actuators and control devices. Consisting of a number of 'nodes' connected by a common wire, each node is able to receive and transmit messages from the sensor/actuator, which it does via a host processor and CAN controller. While the host processor 'sees' all the messages on the bus, it decides which are relevant and which to ignore, as well as which messages it wants to transmit from its own sensor.

Using one of any number of protocols, standardised or not, the controller loads the message onto the bus transmitting at up to 1 Mbits per second. Each individual message consists of a number of frames indicating, for instance, the start of the message, its priority, the number of bytes taken up by the data transmitted as well as the data itself, a redundancy check code and an acknowledgement field.

With all this information travelling the length and breadth of the vehicle it seems silly not to tap into it in some way, if only to save money on sensors. Increasingly it seems that dataloggers are therefore being configured to access such information via the vehicle CAN bus by plugging into a convenient node. For vehicles with an OBD (On-Board Diagnostics) port, this might be a convenient point of access, but since this data is primarily intended for fault finding, unless your software is configured specifically, precise decoding may prove difficult.

So where at one time an army of electronics gurus may have been necessary to glean your information from an array of dataloggers, switching on and loading the software is now probably all that is needed. That is a pity because as an engineer I rather enjoyed that moment when, having connected all the sensors, calibrated them and switched everything on, the numbers just fell onto the screen.

Fig. 1 - Wideband lambda controller with CAN bus output (Courtesy of AIM Technologies)

Written by John Coxon

But despite these many hours of practice I doubt very much that while much personal satisfaction was obtained, the change itself would have been far from ideal. Little wonder then that in a world of iPods and Apps, the art - or should I say now, the technology - of changing gear has been taken over by yet another onboard ECU. But as I have discovered only recently, not all that is electronic is necessarily the best and that, very much like a stand-up comedian, it's all in the timing.

But while the difficulties of synchronising input and output speeds on the traditional H-type gearbox were simple once mastered, the complexity of ensuring the perfect change, time after time, really does take some form of electronic brain. Because not only do the shaft speeds have to be equalised in order for the change to take place, all this has to be coordinated with the engine management system and the selector mechanism inside the box, with the whole process producing the minimum of shock loading and in the minimum amount of time.

Although motorcycles tend to incorporate manual selection systems inside their sequential gearbox assemblies, the forces necessary to generate rapid movement of the selector forks within will often require more force than most people can manage inside the confines of a racecar. For that reason, gear-change actuator mechanisms, as well as that required to 'blip' the throttle on downshifts, generally need to be electrical, pneumatic or hydraulic. While professionals often prefer the more expensive hydraulic approach, and electric solenoid technology can be used, the most popular method - because of its simplicity and ruggedness - is to use pneumatics. At only modest pressures (8-10 bar), pneumatics can produce surprisingly high forces for very little overall weight which, and unlike solenoid technology, do not fall away with temperature and any fall in battery voltage due to repetitive use.

But the secret of any successful system is how it handles the ignition 'cut' facility. Designed to unload the gear dogs of a straight-cut box by cutting the engine sparks momentarily, these units are generally described as having a 'flat shift' capability that enables gear changing 'up' the box at the same time as keeping the throttle wide open. Taking its signal to cut the sparks from a switched input, a gear lever or steering wheel paddle arrangement, the ignition is cut for a predetermined time before being re-introduced at a variable time, 50 ms or 100 ms later.

In operation, once the ignition-cut signal is given, the driveline has to unwind sufficiently to allow the drive dogs to disengage and the forks to move across to engage the next gear before the sparks can be resumed. In open-loop mode, the time taken for this has to be estimated and, if too long, the engine speed may have dropped too much, resulting in unnecessary shock loading as well as loss of time. If the time allowed is too short a failed gear change will result.

Either way, unnecessary damage to engine or driveline could result. The greater the transmitted torque, the greater this transmission wind-up, and thus the only way to minimise these mis-shifts, is to allow for a longer delay. In closed-loop mode, the sparks will remain cut until the gear has been fully selected, indicated by some form of position indicator - on say, the barrel of a sequential box - after which the ignition can be ramped back in again. With this delay varying according to the driveline wind-up and drive dog arrangement, every shift can be timed to perfection with the minimum of interruption of torque to the wheels.

But, surprisingly, the speed of change may not be very much faster than my manual changes. It's just that they are optimally the same time after time - unlike of course, yours truly, with his manual gearbox, who tires rather easily after the first few laps.

Fig. 1 - Schematic of a pneumatic gear shift system

Fig. 2 - Closed-loop control when the encoder on the gearbox confirms selection of the next gear

Written by John Coxon

In most OEM applications, whether turbocharged or not, the method determining the correct airflow to the engine is the Mass Air Flow (MAF) sensor. Calibrated to deliver a signal to the ECU depending on the mass of air flowing through it, the amount of fuel to be injected can thereafter be easily calculated.

Mounted upstream of the engine airbox, the sensors are conveniently calibrated away from the engine and therefore, as befits a production vehicle, is easily interchangeable, but in the rapidly changing airflow of a high-performance engine the remoteness of the sensing makes them not quite so desirable. However, so long as the airflow range of the unit continues to fall within that of the engine, and provided any excess plenum air is not dumped or fed back in upstream of the sensor, the addition of a turbocharger to the engine will not make any difference to the air measuring system, so the sensor will continue to cope. In our turbocharged engine therefore, part-load when the intake manifold is in vacuum is seen solely as a reduced mass flow of air.

Of the other methods of determining airflow - that of speed density, and what is generally referred to as Alpha-N (or throttle angle versus speed) - only the speed-density approach is applicable to turbocharged engines. Since there is no simple relationship between the throttle angle, engine speed and the required fuelling, turbocharged engines cannot easily use this latter approach.

In the speed-density approach, however, the engine airflow is not measured directly but inferred from the engine speed and intake manifold pressure. And while air intake temperature can have a significant effect on the air mass flow into the engine on a naturally aspirated unit, on a turbocharged engine (when the engine air intake temperature can be considerably higher), the charge air temperature sensor is much more critical if potential engine damage is to be avoided. Unlike the normal Air Intake Temperature (AIT) sensor therefore, which is generally of the shielded variety, the charge air temperature sensor on a turbo unit needs to be highly exposed to the airflow and respond quickly to the rapid changes in the temperatures typical of a turbocharged unit.

But the biggest change to the turbo unit is in the manifold pressure sensor. Whereas in a naturally aspirated engine this can measure either the manifold depression - the pressure below that of the atmospheric pressure or its inverse - or the absolute manifold pressure, in a super- or turbocharged engine only an absolute pressure sensor will suffice. Absolute pressure sensors having an output of 0-5V come in a range of sizes, but to give best resolution of the signal the range of the sensor should be only slightly greater than the expected maximum manifold boost pressure.

From the knowledge of engine speed and absolute manifold pressures, the VE (Volumetric Efficiency) tables can be established and populate the various 'look-up' tables from which the fuelling pulse width can be derived. Further modified by any of the engine coolant temperature algorithms, charge air temperature algorithms or indeed the lambda sensor (monitoring the air:fuel ratio) the pulse width distributed to the injectors can then be determined. While the process of fuelling a turbocharged engine varies little from many of that of a naturally aspirated unit, too much boost pressure or even too little fuel can shorten what could otherwise have been a wonderful relationship.

But with fuel consumption an ever-pressing priority, efficiency is the new watchword.

Fig. 1 - Turbo installation in the new for 2011 BTCC-spec 2.0 litre Ford Focus

Written by John Coxon

By its very function, the engine wiring harness will reside in some of the most demanding environments within the vehicle. Subject to extremes of temperature and vibration, and occasionally doused in aggressive fluids, it not only has to be robust and flexible in equal measure but not too heavy either, since like any other component in a racecar its weight should be at a minimum. I recently measured the weight of a simple aftermarket harness and was surprised to find that the wiring alone, without any of the connectors, amounted to more than 1.5 kg or almost 3.5 lb! In my case, careful re-routing around the engine bay will hopefully bring this figure down but this would again appear to be a typical example of how components that many of us generally ignore at the design stage can easily catch us out later on.

When it comes to wiring, flexibility and robustness are the opposites of the same issue. Whereas a single solid strand of wire is more efficient in terms of its current-carrying capacity, having less resistance for a given current than a multi-strand wire of equivalent outside diameter, the likelihood of the wire breaking as a result of repeated bending (and therefore the difficulty in routing around tight corners in the engine compartment) is such that multi-strand wires are always used. Since the next convenient packaging arrangement after the single wire is seven (one inner surrounded by six others) followed by 19 (one inner, six intermediate surrounded by 12 more on the outside), most engine aftermarket harness wires will be configured in this last way.

Other than weight, flexibility or robustness, another factor in selecting the wire size is the resistance to the flow of current and hence the drop in potential difference across it. For short runs (less than 15 ft) a multi-strand copper wire with a cross-sectional area of about 0.8 mm2 - about 18 AWG (American Wire Gauge) - can take 6-10 A continuously from a 12 V source without overheating or causing an excessive drop in the voltage across it. At 0.5 mm2 (about 20 AWG) this drops to 4-6 A; at 0.2 mm2 (about 22 AWG) it falls to 2-3 A.

Since most sensor inputs and outputs from the ECU will be measured in milliamps rather than amps, it would be safe to say that the 0.2 mm2 wire would be around the lowest weight option. For power devices - injectors, fuel pump, ignition and 12 V supply - the larger wire with a greater cross-section might be best. In the end, however, the size you use will depend very much on the connectors adopted, and since many aftermarket systems use automotive OE-type connectors, the ultimate choice is therefore likely to be either 0.8 or 0.5 mm2. When other, smaller aerospace connectors are used, the potential to minimise the weight of the loom is much greater.

There are many factors to be considered when designing an engine wiring harness. Setting aside other issues such as insulation, screening and the type of sheathing used, optimising it all to produce a neat workmanlike solution is an art in itself and seldom appreciated by the wider motorsport fraternity.

Fig. 1 - A typical aftermarket wiring harness
Fig. 2 - Wire strand options

Written by John Coxon

Moore's Law, formulated nearly 50 years ago by Intel co-founder Gordon Moore, said at the time that the number of transistors that could be replaced by an integrated circuit would double every two years. Later, stating that this would go on only until 2015, according to the pundits he has so far been proved correct, such that processing speed and memory capacity have grown to the level when even the most elementary of children's toys now have had far more capability than the systems that sent the astronauts of the Apollo mission to the Moon.

The same is true for even the humblest of data acquisition systems, whether they're used to store engine, transmission or chassis data. But with a typical race engine how much data do we really need and, once logged, do we even have the time to trawl through it and fully understand it all?

In a typical engine, data loggers work by capturing and storing the streams of data that can either be subsequently downloaded or relayed to a display for the benefit of the driver. Instructed by the internal clock the microprocessor reads a voltage or frequency output from the various engine sensors and stores it in the memory in digital form.

In the case of a simple application these sensors may be measuring as inputs - engine speed, fuel, air and coolant temperatures as well as manifold pressure, barometric pressure and throttle position. Outputs such as injector pulse width and ignition timing will also be routinely logged. And even at this basic level, the output of the wideband lambda sensor measuring the air:fuel ratio and less obvious information such battery voltage may also be logged. So far, and unless I've miscounted, we are up to 11 sensors - and I haven't really started yet!

Monitoring individual injector pulse widths on a sequentially injected engine, and perhaps deriving the error between that and the figure intended, boosts the numbers still further - not forgetting to look at what is happening on the ignition timing side or any other engine diagnostics. The point I am trying to make is that once it becomes easy to produce the data, we just seem to want more.

Early aftermarket systems, when they even considered adding internal data logging, would boast 256 kB of memory - enough to record quite a useful amount of what I would call 'simple' and usable engine data, provided logging rates were kept sensible. This 'simple' logger today can offer up to 8 MB of data storage which, when logging eight channels at 250 samples per second, would give you 55 minutes of stored data. Running this logger at a more sensible 25 samples per second ups this to 503 minutes.

Having already proved that eight channels isn't enough, if we increase the logging to 16 channels - eight at about 3 Hz measuring things like temperatures that don't change very much, and eight at a still very useful 10 Hz - then the total logging time jumps to 1021 minutes, or almost 17 hours.

Now I don't know about you but by the time I'd trawled through 17 hours of 16 channels' worth of data, I think I might just have given up the will to live.

Fig. 1 - 256kB of memory

Written by John Coxon

I always enjoy some of the on-board coverage looking over the driver's shoulder. In particular, I marvel at the way that 750-plus bhp can be tamed by our heroes with such apparent ease, changing up and down the gearbox in no more time than it takes to wiggle your fingers.

I think the recent introduction of engine speed graphics allied to the sound emphasises just how much progress has been made in the gear change technology, in that every change is crisp and precise and, from an engineering point of view, deeply satisfying. I only wish we could compare it with on-board footage of similar gear changes 30 or maybe even 40 years ago, when shifting was done manually and the skill left more to the driver than perhaps the engineer.

Using sequential shift transmission technology, the speed with which the engine unloads and re-introduces the load back into gearbox is now the critical factor. One way to do this, and one offered by many aftermarket engine management system suppliers, is a simple ignition cut. Closing the throttle generally takes far too long, and cutting the fuelling introduces fuel wall-wetting issues of adsorption and de-adsorption in the intake system and creates lean 'spikes' generally indicating a loss of fuel control.

Cutting the sparks for a definite period (measured in milliseconds) is probably the simplest way but the method of reinstating it afterwards can be tricky. In many systems the ignition will be progressively ramped back in at the same ignition timing but with the order in which cylinders are brought back to life dependent on the engine configuration and firing order. In some circumstances when the grip available is low, the engine torque output may be reduced by retarding the ignition timing as well. This will then ramped back up to its previous value as soon as possible thereafter. But without any alteration to the fuelling, a simple ignition cut, no matter how short, will still leave unburned fuel passing through the engine. This will subsequently burn in the exhaust or even pass straight through the engine. Either way no useful power is generated and the fuel is to all intents wasted.

During the change gear process itself it must be remembered that all we are trying to do is to unload the gear teeth for just sufficient time for the dogs to disengage one gear and reengage the next. Carefully timed, retarding the ignition only without any ignition 'cut' and ramping up quickly back to its original value is now considered the best way to do this. Making sure that the fuel is burned in the cylinder, no matter how late in the cycle, eliminates any bore wash concerns and done quickly minimises any lost performance from the engine.

Above all, it gives the armchair engineer that deeply satisfying seamless gear change sound.

Fig. 1 - Formula One seamless shift gearbox by Xtrac

Written by John Coxon

In a traditional fuel-injected system the in-tank scavenge pumps transfer the fuel to a central collector 'pot' from where it is picked by the main pressure pump and delivered to the engine fuel rail. In order to keep the fuel flowing through the pump (and hence cool it) the excess to that required by the engine is dumped by the pressure relief valve and returned to the fuel tank to cool.

With a pump drawing something like 7-8 A of current, or even more, this is highly inefficient. In the latest fuel systems the fuel pump will deliver only that required by the engine at the time, at the correct fuel pressure.

Control will be via a closed-loop control system taking a signal from a pressure sensor on the fuel rail and directing it towards the engine ECU. This will send a PWM (pulse-width modulated) signal to the fuel pump to switch it on and off. In this way the pump will only be required to move the precise amount of fuel necessary, thus saving on current consumption and heat transferred into the fuel.

Furthermore, on hot re-starts where the fuel could possibly boil in the rail, creating vapour lock, the rail pressure can instantly be increased, subduing this effect and getting the engine started again. Increasing the rail pressure could also be used to extend the range of the injector units for, say, turbocharged engines at full load, while by the same means reducing it at part load.

While all this sounds fine, the only issue - and quite a major one at that - is surely one of the pressure fluctuations inside the fuel rail and the effect this could have on fuel delivery, especially if fuel rail volumes are small. One method currently suggested by some is to use some kind of accumulator system attached to the fuel rail, similar to those used to prevent oil surge. Contrived to dampen out the oscillations and general perturbations inside the rail generated by the sudden all-on/all-off demands of the injectors, the tuning of such systems could be quite tricky and highly dependent on things like the volume of the system as well as its shape.

But irrespective of this little inconvenience, simply look at the advantages over and above those already stated. The immediate benefit could be higher fuelling precision. Knowing the exact fuel pressure and transient condition at the time of injection means that the exact amount of fuel can be injected at the correct time, and when compared with earlier pulse widths could give an early warning of anything going wrong. Engine diagnostics, I believe they call it.

Not to be confused with returnless fuel systems, where the pump works constantly and the pressure regulator is actually installed in the tank, demand-controlled systems are likely to be seen more often on our circuits in the future.

Fig. 1 - F1 Controller

Written by John Coxon

But there are occasions when the one injector is not enough, say when the turndown ratio of any single injector is simply not large enough to cover the anticipated fuelling requirements of the engine. A typical example of this could be a supercharged/turbocharged unit with a fuel demand very similar to that of a naturally aspirated unit at low speeds and light throttle openings, whereas at high speed and rated boost this fuel demand could outstrip the flow capability of the injector.

Since injectors tend to have a 'dead time' of about 1-1.5 milliseconds below which no fuel will flow, and are therefore far from linear at their lower operating flow rates, increasing the size of the injector to protect the engine at high speeds and loads will almost certainly give poor control in the low-speed part of the map. When a sensible compromise cannot be achieved using a single injector unit per cylinder, the added complexity of a twin-injector design will need to be tackled.

An obvious solution is to package the injectors side by side in roughly the same relative position to the intake valve as in a single-injector arrangement. While this might appear to be an elegant solution using the same fuel rail and minimising the Bill of Materials, the layout fails to take advantage of the possibilities of improved performance that could be obtained by mounting the secondary injector ring upstream of the first.

In particular, while the primary injector can be targeted at the back of the intake valves, the secondary set can be moved to the entry of the intake runner injecting centrally just in front of the intake bellmouth. Injecting at this latter position only when the engine is running at higher engine speed and large throttle openings ensures better vaporisation of the fuel and is accepted by many as the position for best overall engine power. At lower speeds, the primary grouping only should be used, as this will give better low-speed torque and transient throttle response - operating the secondary system at anything other than high intake air velocities might cause the fuel to drop out of the air stream, causing hesitation and possible flat spots.

The critical calibration zone, however, is in the transition zone when either the secondary (larger) group takes over or when they combine with the primary group at the higher intake air mass flow rate. Each time injectors are switched on after a period of inactivity, a variable enrichment strategy may need to be followed to compensate for the inevitable wall wetting on the inside of the manifold.

Furthermore, to prevent any possible oscillation between groups at the changeover point when the secondary system cuts in or out, the secondary system should always switch on somewhere at least 50 or even 100 rpm above the point where they switch off. By slashing the injector pulse width or cutting it immediately the secondary group tips in, and by compensating for the wall wetting effect, eventually (and after a lot of trial and error), a perfectly seamless transient calibration can be obtained.

It may take some time to refine but the result could even give a small overall increase in peak performance as well.

Fig. 1 - Injector flow characteristics
Fig. 2 - Variation of injector dead time with applied voltage

Written by John Coxon

Consisting of a throttle pedal device requesting a torque demand from the engine, the engine ECU calculates the ignition and fuelling necessary and requests the appropriate amount of air from the engine throttle. With no physical cable connecting the throttle pedal to the unit on the engine, communication is achieved solely via electronics and digital signals. Systems similar to these were first used without mechanical back-up in the F-16 jet fighter in 1974 and then the Airbus A320 commercial airliner 14 years later.

While the complications associated with flight control make systems of this type ideal for large or sophisticated aircraft, for most motorsports applications drive-by-wire throttle systems are mainly an unnecessary complication. But with the increasing amount of 'spec' or crate engines derived directly from OE manufacturers, having electronic throttle bodies, aftermarket ECM suppliers are having to incorporate drive-by-wire routines of their own to keep pace with the market.

Having few technical advantages for most forms of motor sports, the lack of a direct mechanical link places a greater emphasis on the reliability of the data being processed to ensure that the signal given to the throttle is reliable 100% of the time. With a mechanical system this is guaranteed, but when independent electronic control is used without a mechanical back-up, the system has to have a certain degree of redundancy and, were it to fail at any time, to fail safe.

To minimise if not totally eliminate the concern, both the throttle demand pedal and that at the engine need two entirely independent monitoring systems. Consisting of separate throttle angle sensors - increasingly of the non-contacting (and therefore much more reliable) Hall-effect type - two are required for the throttle pedal and two on each of the engine throttles. For an inline engine, that amounts to four channels simply looking at the throttle position but for a vee engine, assuming only one throttle per bank, this rises to six!

Each sensor on the same spindle will be calibrated separately, such that when the throttle position is calculated its exact position can be identified with confidence. For example, one sensor might be calibrated on a 0.5-4.5 V scale, ascending when opening; the other might offer the same voltage but descending while opening. The two different methods should therefore come up with the same throttle angle.

Other options could be 0.5-4.5 V for the first sensor 1 and 0.5-2 V for the second. Either way though the two input voltages to the ECU will be able to pinpoint precisely the pedal demand requested and process the signal with a high level of confidence. Should these two differ in their expected output, the system will inevitable fall into a default or 'limp home' condition.

While this may seem complicated enough to the non-specialist it gets worse, much worse. To power the engine throttles opened and closed, the throttle body will receive two pulse width modulated signals - one to open the throttle valve, the other to power it closed. This will be undertaken using DC servo motors using a form of electronic PID (Proportional, Integrated and Derivative) feedback control. The significance of these is explained in this month's Dynamometer offering but with two circuits required for each throttle valve, already the complication is increasing.

Drive-by-wire may be the future of roadcar technology, and its usefulness in other systems such as traction control make it highly attractive in terms of general everyday road safety. But for the weekend racer in years to come it could introduce an awful lot of unnecessary grief.

Fig. 1 - The drive-by-wire throttle circuit

Written by John Coxon

In the case of a modern race engine, we are all too familiar with the idea of electronics together with a little help from computing to deliver a superior, more fuel-efficient system with greater levels of reliability. But in many of the aftermarket so-called engine management systems found in motorsports, the degree of this control is somewhat limited.

Promoted because of their user friendliness, these aftermarket systems expose the calibrator to a number of look-up tables and software-toggled settings, which enable the inexperienced to trim the device to the demands of the engine. The look-up tables will invariably consist of fuel maps, ignition timing maps, lambda target maps and temperature correction features that allow the user to fine-tune it the engine to the limits of the system. The inputs and outputs of these will invariably be linked to individual ECU pins, which may or may not be programmable to perform other functions.

Whichever way you look at it, these devices are designed to work in a certain way, and the firmware inside the ECU with its various control algorithms will be fixed and unalterable. The firmware might contain a number of tables or parameters that may be modified to take things such as turbochargers, even multiple injectors per cylinder, but the way in which these work will be more or less fixed according to the installed firmware.

This may be totally acceptable to the amateur, whose primary target might just be to get the engine running in the first place, but to the professional, the control offered by these systems may be somewhat lacking.

At the other end of the spectrum, to introduce a new or revised control algorithm will invariably involve much simulation and testing with third-party software tools. At the prototyping stage, third-party hardware platforms may need to be used with custom software, allowing the simulation to be linked to the real world.

This process will require compilers, assemblers, linkers and, no doubt, even more software. And by this time the team will have expanded to include control engineers, firmware engineers and hardware engineers with enough knowledge and experience across a number of skill sets to deliver a quality product.

Needless to say, ECU development at this level can be very expensive, even assuming the appropriate software licence can be obtained. Is it no wonder therefore that with some vehicle manufacturers, the powertrain electronic management department can often dwarf the engine and transmission mechanical departments combined.

The ideal approach, and one that is now offered by EngineLab in the US, is to enable the engineer to focus totally on the algorithm development and carry out all the work within the production hardware platform. In tightly coupling the firmware to the Host Console application on the host PC, unnecessary complexity can be avoided and once the prototyping is complete, the job is finished.

Referred to as the EngineLab Machine, the architecture is built on a real-time operating system and presents a simple application development interface to the control system designer. Not requiring any further compilers, assemblers or firmware development, the control algorithms are developed directly onto the production-based hardware. Indeed, the equipment as purchased doesn't contain any fixed-function behaviour model, the entire control system being downloaded to the system via the host PC.

Described more as a general control system than a fixed-function engine controller but at aftermarket-type costs, this could begin a trend for future high-level engine control systems.

Written by John Coxon

For those with long memories, gas turbines were all the rage in the late 1960s and early '70s. Undeterred by the experiences of the Rover BRM project of 1962-65, the Howmet GT, STP-Paxton and the Lotus 56 all used gas turbines to various effect, and all were controversial to some degree.

At the time, the automotive industry was looking to the gas turbine as the next Big Thing, but when hopes of fuel efficiency comparable with the reciprocating engines of the period petered out, interest steadily waned. But at a time when fuel efficiency is everything, one wonders why gas turbines in Formula One have been mentioned at all?

On the other hand, the inclusion - or rather, non-exclusion - of direct injection is surely only common sense. I always found it strange that direct injection was banned in Formula One some years ago on the grounds of spiralling engine development costs. Yet it is arguable that DISI (direct injection spark ignition) is the one area where the resources of the manufacturers could have been most successfully applied.

Nowadays, just about every vehicle manufacturer has or is about to introduce a spark-ignition, direct-injected unit, and if Formula One is to resume its place at the pinnacle of engine technology then turbocharged or naturally aspirated, heat-recovered or KERS direct injected is surely an absolute must.

Although the in-cylinder motion and position of the injector are different from those in a port-injected device, from the ECU's standpoint there is little change. The ignition system is still the same, and although the fuel has to be injected in a much shorter time frame (helped by the 200 bar fuel pressure), any high-level engine management system should be more than capable, at least in homogenous combustion mode. For more complex, fuel-saving modes multiple injections per cycle may be necessary.

So for the most part there is little change required at the ECU but one major change, as explained to me recently, was in the problems associated with the gear change.

In a port-injected engine, during the gear change both fuel and spark are cut momentarily and then reinstated according to any number of user-developed algorithms. During this process the manifold and inlet port wall wetted by the fuel in earlier cycles progressively dries out during the non-fuelled cycles and then adsorbs some of the fuel when reinstated.

Despite the best efforts by engineers to compensate for it, this 'imperfect' fuel-air mixture reduces the shock loading and introduces a level of 'cushioning' to the transmission. In direct injection, when the fuel is metered directly into the cylinder, this drying out and re-wetting doesn't take place, and even taking into account any 'ignition cycle recovery' caused by the lack of residual gas in the chamber - and without any change to the fuelling on re-instatement - the extra shock loading introduced could have a major impact on the durability of the drivetrain.

Since gearboxes are expected to last for at least four races, any move towards direct injection will need to address this issue - as well, no doubt, as many more.

So, gas turbines? Regretfully, no. But direct injection? Definitely yes! Yet surely that goes without saying.

Fig. 1 - Direct injection, surely the future in Formula One?

Written by John Coxon

As opposed to surface ignition (sometimes referred to as pre-ignition), combustion 'knock' is created from the spontaneous ignition of part of the end gas in the combustion chamber in advance of the flame front. Resulting from too much ignition advance for the conditions of temperature and pressure in the combustion chamber, uncontrolled, this phenomenon can be highly destructive and lead to almost instant engine failure because of the extremely rapid and localised increases in pressure in the chamber. When striking the cylinder wall, this can be heard as combustion 'knock'.

The frequency of this knock is engine dependent and more usually measured in kilohertz, while the intensity apparent to the outside world is very much down to the basic structure and stiffness of the cylinder block itself. Over the years, knock sensing technologies have been progressively developed and today rely on piezoelectric technology measuring the structure-born acoustic vibrations. For the best sensitivity this should be mounted where the cylinder block 'pants' the most. Initial surveys might suggest that this should be on the unsupported external wall of the cylinder jacket adjacent to each cylinder but when factoring in the metal boss to which the sensor is attached, this isn't always the best option. Also if only one sensor per bank is used, the position identified has to be a compromise for all the cylinders.

To help ease the problem, where EMS suppliers offer a knock control module as an extension to their ignition controller, the knock sensor tends to be 'gated' for each individual cylinder. This helps to eliminate any spurious engine mechanical noise (especially so in racing engines) and identifies the cylinder in question. Having identified the knocking cylinder, the module will then produce an analogue voltage output according to the knock intensity and the ignition retarded by an amount for that cylinder, as pre-programmed into the ECU. This could be as much as 8 or more degrees of ignition for heavy knock. Once that particular cylinder has stopped knocking, the ignition will be ramped back quickly in stages but the strategy used will depend on the individual engine developer and be jealously guarded. However, most knock control modules will enable detonation to be controlled on a number of levels at a wide range of engine rpm and load (or throttle) settings, but only the most sophisticated will enable the engine to re-learn and adapt the main ignition map to the revised values. This ensures that the engine will run at the optimum ignition advance at all times. For boosted engines some systems might also include a feedback loop to reduce the boost level before retarding the ignition.

When it comes to combustion control, the systems just become more and more complicated. So unless you have a large budget and a willingness to undergo all manner of grief, when the EMS supplier tells you to map clear of 'knock,' it might simply be wise to do so.

Fig. 1 - Combustion 'knock'

Written by John Coxon

The volume of the fuel trapped on the manifold wall is often known as Tau and designated by the Greek letter '?'. In modelling it is assumed that the amount of the 'puddle' or fuel volume remains constant and the air flowing over it carries it away towards the engine. The more the volume of air flowing past or over it, the more the mass of the puddle is reduced. Tau modelling software therefore uses instantaneous airflow calculations to attempt to predict and maintain this value constant by either adding or reducing to the fuel specified in the engine map.

Under steady state running the engine will be calibrated to run at optimum air-fuel ratio, be that slightly rich for maximum power or slightly weak for best fuel economy. During rapid changes, for example when the throttle is suddenly snapped open and the wall film rapidly evaporates, is when compensation is necessary. If this isn't done the engine will 'see' a temporary lean spike resulting in a hesitation or a loss of power followed by a sudden surge back to where it would have been. Likewise when the driver lifts off the throttle at high engine speeds and no power is demanded, fuel may need to be temporarily switched off but to make sure that the fuel 'puddle' is maintained and that a lean spike won't result as soon as the throttle is opened again, a degree of fuel may have to be re-introduced

On more complex systems, the intake models to predict and consequently control the air-fuel ratio more accurately, have to be much more sophisticated. Intake manifold volume, port design and even the positioning of the sensors used, all have to be taken into account to anticipate the airflow to each of the cylinders and during each and every cycle, if perfect engine control is to be achieved.

Fig. 1 - The 'Tau' factor.

Written by John Coxon

In times past the ignition was triggered from a mechanical distributor, cleverly driven from the camshaft. For every two turns of the crankshaft, like the cams, the distributor rotated but once. Firing on cue for each cylinder in turn, the system was effective if somewhat inaccurate, even if it could be occasionally confused by owners setting it 360 crank degrees out. Importantly thought, it knew where it was at all times even when awoken from its slumber.

But then someone brought along electronics and eventually electronic engine management. While the distributor lasted a little while longer in the form of breakerless or transistorised ignition, it wasn't long before it was given its P60 and destined along with the carburettor to the scrap heaps of this world.

But the new system was full of promise and bright ideas and while it could pinpoint precisely where on a single crankshaft cycle it would fire the charge, of the two cycles it couldn't work out which one of the two especially after it had just woken up. On four cylinder engines this wasn't a big issue. Using a system called the 'wasted spark' system and a double-ended coil, two cylinders could be fired at once from the crankshaft at each rotation and while one spark ignited the charge, the other somewhat wastefully and harmlessly fired towards the end of the exhaust stroke on the other cylinder.

At this point riding to the rescue came the variable reluctance sensor. Used to measure the engine speed by supplying a series of electrical pulses, he offered to remove one of its teeth to indicate where he was on the engine cycle and then fitted to the camshaft, he would be able to indicate to the bright new thing where he was in the overall cycle. The problem here was that working from the camshaft, rather like the redundant distributor, chatter and backlash in the cam drive train could introduce inaccuracy and in any case he would only work while the engine was moving. All was rather sad and the poor bright new thing was very upset that he might never be able to demonstrate how good he really was…

But then, suddenly and out of the blue came the Hall Effect Sensor! CKP, as he is known to his friends, needed a simple conducting (and therefore metal) toothed disc and together with his magnetic pickup could produce an electrical output with the passing of each tooth. And like the reluctance sensor, if he removed a tooth and was attached to the crankshaft he could tell where he was most of the time. Furthermore, unlike others, his output signal did not vary with engine speed giving the same waveform independent of how quickly he could move. This meant that the engine timing could be set whilst the engine was not running and pleased his masters so much that he was allowed a friend. With him looking at the crankshaft and his friend perched up near the cam between them the problem was solved and whenever the engine woke up in future it always knew where it was.

And they all lived happily ever after.

Fig. 1 - Hall effect sensor.

Written by John Coxon

But in the world of the calibration of electronic fuel injection systems we have our own Big Bang - that part of the engine start-up procedure when the engine is cold and fuel is injected for the first time. While not quite so critical for racing engines with port fuel injection, this initial period of operation is increasingly significant for road going vehicles and as technology moves forward, with all types of direct injection systems.

In any engine control system there are essentially four phases of engine operation: cranking, warm up, open and closed loop control. To these we can also add acceleration, deceleration and engine idle but let’s not get too carried away with all these at this stage. All of these will require fuelling and ignition but because each is also so sufficiently different from the other, different approaches to the control software will be needed for each phase. The first, and judging by the amount of research work undertaken over the years, the most difficult, is engine cranking and first fire. Now at this point it is perhaps worth pointing out that every manufacturer will have its own ideas, and that, depending on if you have to achieve certain hydrocarbon emission targets or not, will make the process more difficult. As these targets change, the ideas and strategies to meet them may also change.

In most, if not all modern port injection systems, the individual fuel injectors will be located such as to spray fuel directly into the intake port and timed to coincide with the intake stroke when the valve is open. Generally known as sequential injection, the system requires to know the precise position of the piston relative to top dead centre and also where it is relative to the valve timing as well. For this a crankshaft and a camshaft angular position sensor (CKP) will be required.

During the engine starting procedure at the point of cranking, the engine will not necessarily know exactly where it is in the cycle of events of the 4-stroke cycle. It will be able to detect the ignition trigger point but even with a cam sensor it will not necessarily know on which of the two cycles, firing or induction, it is on. With the desire to fire and start the engine as quickly as possible, one strategy is to fire all the injectors at once during the first cycle producing what in effect is a ‘Big Bang’. This will ensure that the engine fires almost instantly upon cranking and, say, when the engine speed is greater than 300 rpm, the electronics will know that the engine has completed cranking and moved into the warm-up phase. At this time another software subroutine will take over.

For racing engines with sequential injection and not necessarily the requirement for instant, key-on firing, a different strategy may be used. To build up oil pressure in the dry sump system, cranking could be encouraged for say, a second or so. After this, the engine will have determined when to fire its injectors relative to valve opening and since there aren’t any hydrocarbon emission spikes to worry about, each of the cold cylinders can fire in turn.

There’s much more to getting an engine started than just cranking it.

Written by John Coxon.

The two most important parameters in determining the air trapped within the cylinder at any given engine condition are speed and volumetric efficiency. While the former is relatively simple to measure, the volumetric efficiency is in itself a function of many parameters – things like engine capacity, valve opening and closing events and intake and exhaust system design, to name but a few. For the sake of simplicity, all these can be approximated to a single measurement of the throttle angle and hence the concept of Alpha-N systems came about. Fuel pulse width value was therefore computed purely as a function of engine speed and throttle angle. Although this system works and will often work much better than a poorly set up carburettor, it lacks finesse and not taking into account parameters like the intake manifold pressure, may introduce drivability issues at some stage. But as a basic system, Alpha-N is usually more than adequate for the average club track car where long periods of wide-open throttle are used. However, any engine changes, however slight, will inevitably need a complete dyno re-map.

Most high performance road cars today run mass air flow meters. Based on the hot wire principal, these measure the cooling effect of the stream of engine intake air across a heated platinum wire and hence can calculate the precise amount of air flowing at the time. Changes to the throttle angle simply reduce the amount of air and likewise any minor changes to the engine will not necessarily demand recalibration of the whole system. The downside to MAF meters is that in some applications they may be difficult, nay even impossible to install. Generally incorporated into the intake of an air box and integral with the throttle assembly, these need to be installed where the flow is ideally laminar and non-pulsating because although the meter can measure the flow, it cannot necessarily determine its direction.

An alternative to either of the above is the speed-density approach. A modified version of the Alpha-N system, the air mass flow is calculated based on manifold intake pressure, air temperature and engine rpm. Knowing the air intake pressure and temperature, and with an appreciation of the universal gas laws, the intake air density can be deduced. And if the volumetric efficiency of the engine is determined then at any position of engine speed and manifold pressure the true air mass flow can be calculated. In any speed-density system therefore careful mapping of the volumetric efficiency is essential.

As an alternative to carburettor ‘tuning’, engine management systems may be easier to understand but the task of refining the calibration to one that matches the demand of the engine, is no less exacting.

Written by John Coxon.

As I remember I first came across it in the mid 1980s. Concerned that car doors were being loaded with so many electric components – central locking, electric windows, adjustable rear-view mirrors, heated rear view mirrors, and that the volume / weight of the additional wiring was bringing reliability issues, a wiring system was developed to reduce these wires to just two. The system was known as multiplexing. And over time when engines and gearboxes were controlled by ECUs and the amount of electronics and shear weight of the wiring became an issue, this multiplexing evolved into the standard, developed at Robert Bosch as CAN Bus or Controller Area Network Bus.

Originally designed to handle short messages (up to 8 bytes), support multi-master access (collisions being avoided through a system of priority) with a high degree of reliability, today virtually every passenger car manufacturer incorporates at least one such module in its electrical system. Offering a significant reduction in wiring – reducing weight and cost, and minimising the number of connections giving improved reliability – the CAN bus is effectively a pair of twisted wires (one ‘high’, one ‘low’), which transmit a series of digital messages, either 1s or 0s according to a given protocol and between ECUs. Devices connected by the CAN network will be sensors, actuators and their controlling devices with messages being loaded and downloaded along the way (effectively the bus stops) at various nodes.

Now the sensors used for most engine management systems will be of reasonable quality and provide sufficient resolution to ensure that the engine works satisfactorily to within certain performance criteria. The information they supply to the bus however will not be accurately calibrated and anybody using this information for engineering research / development purposes will introduce a level of uncertainty into their measurements as to make them, if not totally useless, then certainly suspect. The only way to get high quality development data is to use good quality transducers calibrated to your requirements. The CAN Bus data, while convenient, is not always up to this standard. As one data logging expert said to me recently, “They think they have accurate data but could end up seriously misleading or confusing themselves.”

Maybe a case of the CAN Bus which can’t?

Written by John Coxon.

Rather than just dribbling out under the action of the depression in the choke and whichever direction the air in the manifold happened to be pulsing at that instant, the invention of the electronically controlled fuel injector was a dream come true. Injecting the fuel at around 3 times that of atmospheric pressure and timing it to coincide with the positive pulse of air going into the engine, that same high-speed video now showed that the fuel droplet size was vastly improved. While engines generally gave little more outright power than before, the timing of the injection pulse enabled much better control of the engine and was to lead the way to a whole new range of possibilities.

Injectors are generally driven by ‘grounding’ the 12-volt electrical signal, and energising the coil within causing them to open, injecting the fuel at high pressure. Early versions were mainly of the ‘low impedance’ type. Typically having resistances between 1 to 4 ohms, once energised, the high currents involved generated lots of heat which was potentially damaging to the transistorised circuitry then in use. To limit this, injector drivers of the ‘peak and hold’ variety were devised. The battery voltage was applied until a pre-determined current was reached. At that point and once the injector was open, the current was then reduced by modulating the 12-volt source on and off very quickly for the remaining period of the injector opening. The high initial currents were designed to give improved injector response times while the subsequent reduction in current protected the switching transistors. When using these types of injectors it was therefore imperative to ensure that both the injector and driver characteristics were carefully matched achieving maximum injector response time but with minimum build-up of heat.

Later on, designs were introduced with a much higher impedance. These so called ‘high impedance’ injectors (usually between 12 –16 ohms) take the 12 volt feed directly without any form of current control and as such the associated drivers are generally referred to as the ‘saturation’ type. So long as the circuitry is designed for the task, a similar outcome can be achieved if a resistor is used in series with a low impedance injector but the advice must always be to check with your EMS supplier. In either case lower currents mean reduced injector response but the heat build up in the electronics should aid reliability.

While the mastery of the gentle art of tuning carburettors may no longer guarantee a job for life, the far wider scope enabled by it’s much younger sibling together with the advancement in modern electronics could possibly now do so.

Written by John Coxon.

But the key word in all of this is reliability, and in particularly, the reliability of the engine and exhaust emission sensors. Placed in some of the most inhospitable surroundings, these electronic components were not only requested to cope with all manner of under bonnet conditions of grime, temperature and vibration but they have to perform 100%, all of the time to levels of reliability hitherto undreamed of.

One such sensor is the engine coolant temperature sensor, or the ECT, as it is more widely known. Used to control the engine during warm-up, ECTs are essentially thermistors (a semi-conductor resistor) inside a hollow sealed brass tube. Having a negative temperature coefficient (NTC) and a large resistance at cold ambient temperatures, the resistance of these thermistors drops significantly as the coolant temperature approaches the more normal working conditions of an engine. They are therefore ideal to control the amount of fuel required during the warm-up phase.

An engine starting from cold will require an enriched fuel air mixture. This is to compensate for the condensation of the liquid fuel on the ‘cold’ surfaces in the intake tract and piston crown/combustion chamber and ensure that a combustible mixture is presented at the spark plug for first fire. During cranking, most ECMs will take a signal from the ECU and running ‘open loop’, inject about 30 to 60% more fuel than normally required according to some pre-determined algorithm established by either the equipment supplier or the engine manufacturer. Adding lots of fuel on initial cranking, this then trails away after a couple of seconds to ensure that, should then engine not fire, the engine isn’t flooded. However, once the engine fires (often defined by the engine RPM being greater than a threshold value - generally around 400 RPM), the engine will move into its after-start enrichment strategy.

Beginning as a user-defined enrichment (typically around 20%) this after-start strategy will slowly ramp down to zero after typically two hundred or so trigger events, finally gliding down into the coolant temperature correction map. Thus for the rest of the time the coolant temperature sensor is in control, eventually deciding when closed loop control of the fuel/air mixture, should you wish, takes over as well as many other temperature related events.

In slightly more sophisticated systems, in addition to the fuelling, coolant temperature corrections can also be made to the ignition map to compensate for the slower flame speed of rich mixtures. Often called the master sensor, it is little wonder that its robustness and reliability are of paramount importance.

From a time when the only cold start control device was the carburettor ‘choke,’ engine fuelling technology has moved a long way.

Written by John Coxon.

Because controlling fuelling, ignition, even valve timing, duration and lift achievable through electronic control units doesn’t necessarily mean that is the best or even the right solution. That is a deliberately controversial statement but it is worth considering what lies behind it and what the alternatives and options are.

At its simplest, an engine needs oxygen and the right amount of fuel in order to breathe. Attach a supercharger or turbocharger, and you cram more oxygen in allowing more fuel to be oxidised. Which is why a 1.5 litre serially supercharged V16 had about the same outputs as a 4.5 litre naturally aspirated engine.

In ignition terms, a spark plug ionises the gas mixture around it, starting a complex combustion process that relies on long chain molecules breaking down.

The chemistry relies on hydroxyl radicals: in a common rail diesel the break down relies
on sono-chemistry where cavitation or vacuum voids are produced by the very droplet speeds and the design of the injector nozzles. In a gasoline engine, that breakdown – the pre-combustion chemistry – relies on exposing as many fuel molecules to oxygen as possible and to the ionisation process. This is why a gasoline engine needs more and more advance as the revs rise because the time available for pre-combustion chemistry reduces.

OK, that was a child’s guide to engines – what does it have to do with control systems? So the modern control system is provided with a memory that can be programmed with differing parameters for different performance requirements which is then able to control the variable elements of an engine in order to meet the performance demands. Is there are an alternative?

If you look at the chemistry and the time that the chemistry requires, things become clearer. Probably the most important thing is to deliver a fuel/air mixture with enough light fractions broken off the hydro-carbon molecules to give a sustained oxidisation. To do that takes a specific amount of time that is absolutely independent of engine speed. If the time isn’t long enough, the unburnt hydrocarbons will pollute the exhaust. You can optimise combustion by completing as much of the pre-combustion chemistry as possible and really one way is to time injection so that the fuel droplets can oxidise before reaching the combustion chamber.

An alternative – which is where possible redundancy in ECU’s comes in – is to control the chemistry. In some engines that can be done by recirculating exhaust gases which are rich in hydroxyls, the key bit of chemistry. An alternative is to increase the hydroxyl level of the intake air.

Either way, engine control can be achieved without mapping and, in the latter case especially, with a fixed ignition timing. In other words, listen to what the engine says. Increase the load and combustion chemistry changes. Reduce it, and it changes again.

So if you measure combustion efficiency by sensing increases of decreases in unburnt hydrocarbons, you can achieve more by clever chemistry than you can by complex mapping.

Or that at least is the theory that may become more and more relevant as fuel efficiency is pursued.

Written by David MacDonald.

For the first decades of engine development, the performance characteristics of the engine were set through the original design: valvetrains, carburation, ignition timing and their relationship to engine revs were the tools that were used to optimise engine performance.

It owed as much to art as science in many cases and laid the foundation of reputations won and lost.

The ability to change the engine parameters to suit varying conditions makes the modern i.c. engine what it is. Typically, an ECU will contain an engine ‘map’: that is, set points for ignition and fuel timing that have been determined on a dyno to provide maximum efficiency within the operating contours of the map. In race applications, this degree of control extends to transmission behaviour. During the development of modern race engines, control of the engine’s parameters could be set remotely by engineers reading the telemetry outputs of the engine.

The modern ecu certainly has the capacity to manage far more than ignition and fuel. Valvetrains operating independently of camshafts will allow a gasoline engine to avoid throttling losses and to optimise gas flow velocities. Real time analysis of exhaust gases will allow engine parameters, including intake air chemistry, to be continually adjusted to meet the key factor in good combustion, that of completing necessary pre-combustion chemistry in a timely manner.

One of the tragedies of the current trend to allow rules to be set by technical illiterates who treat motor racing (both four and two-wheeled) as entertainment is that standard ecus are used and the forefront developments of engine work now inevitably move away from racing. A simple example has been traction control: an engine is entirely capable of being its own traction controller. By knowing the maximum wheelspin limited acceleration in each gear, the engine developer can calculate the maximum reduction in time between the ignition/fuel point and that immediately following. By setting that reduction as the maximum, an engine can softly limit wheelspin.

Clever use of ecus allowed good engineers to provide traction control without the crude systems that became outlawed. Instead of applauding the ingenuity, standard ecus were introduced removing the premium on the engine designer’s creativity. The speed and capacity of ecus used in road cars is extraordinary and in many respects in advance of what is genuinely useful. The drawback in many cases is the necessity to collect information in adverse environments and to that extent the ecu relies on advances in sensor technology. Ride-handling compromise control is well within the mapping capability of an ecu, but the limits and much of the expense lies in the provision of the necessary sensors and their incorporation into components.

Written by David Macdonald.

The reason it has been possible at all is down to the ingenuity and resourcefulness of some very bright people. More usually the approach has been the complete replacement of the electronics with aftermarket systems.But, sometimes regulations demand the use of the greater part of the vehicle’s electronics. In Group N rallying, for example, competitors are required to keep the entire loom and the original ECU connectors in their original position. The internal hardware and software are free, however, but in this era of controller area network [CAN] systems, developing technology that can interface successfully is a big challenge.Certainly this is the view of motorsport electronics company GEMS, which has been tackling the challenge through many generations of car electronics. However, the company’s Henry Skinner feels it has met its toughest challenge yet with the latest product from Mitsubishi, the Evo X. There were the usual problems of making the unit survive in its original engine bay location despite the greater heat and electromagnetic noise of the Group N rally car. But also the unit interfaces with Mitsubishi’s most complex installation yet, communicating with the differentials, antilock braking system, dashboard and drive-by-wire throttle and ETACS [extended total access control system] via a CAN network.“It was the most complex plug and play ECU we have done so far,” says Skinner. The challenge, he explains, was understanding the complex interplay of error checking messages, any of which would stop the system working properly if they did not receive the correct responses. As manufacturers never release information on their electronics, ways to keep all the car systems functioning had to be worked out without any prior knowledge of how they worked. “We spent ages trawling through the data, looking at all the different bits and bytes trying to work out what they meant. It was like looking at a book with all the letters mixed up, says Skinner.” Only once all the system’s protocols had been understood could GEMS start applying its own software strategies for engine, diff and brake control to produce the production unit.But despite the size of the achievement, Skinner believes this could be the company’s last plug-and-play replacement ECU for rallying. Apart from the increase in development time making it more expensive to develop, he sees S2000 taking over from Group N in the near future. “We can still supply units for all the existing Mitsubishi and Subaru Group N cars but S2000 allows you to replace the entire electronics system, which is more straight forward for us but much more expensive for the competitor.”

With the increasing complexity of manufacturer electronics and their continued reluctance to share information, complete substitution of systems seems like the way forward for rule makers. But this will always be the more expensive route, just like replacing throttle bodies with carburettors.