Much like the engine crankshaft therefore, which can be excited by the forces of combustion and fail catastrophically when the frequency with which it is excited approaches the component’s natural frequency of vibration, the camshaft can also be subjected to such effects. Rarely does this lead to a torsional failure but, if unchecked, the vibration can have disastrous effects on components such as the valve timing or valve spring control.

If we take the case of an inlet camshaft opening and closing a single valve, the forces involved and hence the torques induced look similar to those outlined in Fig. 1. If you have ever tried to turn a camshaft over by hand you will realise that the forces required (and hence the instantaneous torques) can be surprisingly high, but once moving, the effort is considerably less, the force used to compress the valve spring being recovered when it expands a few degrees later. The total energy expended over a complete rotation may be minimal but the instantaneous forces at any one time to accelerate the valve opening can be amazingly high.

Now, if we consider that 140 cam degrees or so later, after the inlet cam closes, the exhaust cam will open, producing a similar set of forces but displaced by the 140º or so down the shaft. Again, the forces may be high but the overall energy input is low. If we then move along to the next cylinder in this engine (let us say it has four cylinders) we will realise that the inlet cam forces are similar to those of cylinder 1 but displaced by 90º (in the case of an engine firing at 1-3-4-2) and likewise the exhaust cam.

Repeating the exercise for cylinders 3 and 4, if we start to sum up all these instantaneous forces algebraically for the full four cylinders, the instantaneous torques produced at each angle around the camshaft could begin to cancel themselves out along the shaft. As a result, the peak torques will fall and the overall energy used to power the camshaft could be comparatively small. How small will depend on the precise timing of the inlet and exhaust cams to each other and the amount of friction in the system.

Clearly the greater the number of cam lobes along the shaft (say in an inline six – or even, god forbid, eight cylinder) the greater this smoothing effect.

In all cases, however, these instantaneous forces or torques will generate a level of vibration in the shaft which, when analysed or split up using a mathematical technique known as a Fourier analysis, will be represented as a large number of sine waves of different amplitude over a wide range of oscillating frequencies.

If the natural frequency of the shaft – or indeed any component in the system – is the same or close to any of these frequencies, and the forcing amplitude is sufficiently large, then unless it is damped in some way the shaft or component will resonate at that frequency, creating far higher loads than usual. This could be detrimental to either the performance of the unit or its durability. In the extreme case, with a hollow camshafts gun-drilled to remove much of the inner core, these forces could fatigue the component, or if the amplitude of vibration was such that it altered the valve events in relation to the piston position, then piston-to-valve contact could occur. 

This issue is not normally a problem with small, low-speed or short engines, but arises in high-speed, multi-cylinder, multi-cam units where the need for weight reduction can be greatest or the search for outright performance paramount.

Fig. 1 - Cam drive torque at 10,000 rpm engine speed

Written by John Coxon

The primary goal of any camshaft though is to assist the opening and subsequent closing of the valves in the shortest space of time, to ensure the greatest flow of air into the engine in the time available. In direct-acting tappet designs the velocity of the valve is limited by the diameter of the rotating follower, while the acceleration is limited by the forces in the system and ultimately the strength of the spring with which to slow the valve down again as it reaches full lift.

At full lift of course the situation changes, and suddenly the issue is likely to be one of Hertzian stress just below the surface on the camshaft nose. For the return journey back to its seat, for many years the profile would have been a simple reflection of the opening phase, but of late, with the widespread availability of cam design software, the complex mathematics involved is no longer an issue. In the distant past, crunching all the data to develop the lift, velocity and accelerations (and possibly even jerk if you were keen) and getting them to match at each node, would simply have little or no time – let alone the will – to design a different closing profile. The tendency therefore was to repeat the opening flank but in reverse.  Ensuring the valve landed gently back onto its seat was, however, often a problem.

But having designed the cam motion as above, the great concern now is to ensure that, with modern compact combustion chambers and large valve lift and diameters valves, the timing is optimally set to give maximum power without valve-to-piston contact. At this point I am reminded of a conversation with a championship-winning engine builder going back to the days before ‘crate’ engines. In those days, even though the series engine regulations were very restrictive, the camshaft timing relative to the crankshaft was still unregulated. Naturally engine builders do what they invariably do and, having experimented with all the normal engine ‘tweaks’, this builder resorted to altering the camshaft timing, particularly that of the intake cam. The engine already had valve-shaped cut-outs in the piston which, under the engine regulations, could not be altered, but testing had revealed that the engine performance clearly increased with advanced camshaft timing. The question was: did it go on climbing into the region where the valve might hit the piston?

Finally, and saying a short prayer as he did so, the builder advanced the cam to the point when it was just about touching the piston at top dead centre and then, with fingers crossed, ran the engine again through another power curve. The fact that the engine survived is a matter of record, and with it too the marginal increase in top-end performance such a tweak produced. Unfortunately this was at the expense of a slight drop in maximum torque at a lower speed and so – both to sleep at night and preserve his customer’s championship lead at the time – the decision was taken to retard the cam back to its former position.

The interesting thing was that, after stripping the engine, a clear mark remained on the top of the piston where the valve had just about been in contact with it – not quite enough to cause mayhem but in reality close enough to remove the carbon from it, leaving a patch of almost clear aluminium.

Now that’s what I call optimal cam timing!

Fig. 1 - Setting valve lash can sometimes be critical

Written by John Coxon

It was much later in life that the word ‘overlap’ in relation to engines was to enter my vocabulary, and funnily enough, and as you might expect, it was to do with camshafts and timing of the engine valves. As a reminder, valve overlap is that period when the exhaust valve is just about to close and the inlet valve has started to open. In our theoretical engine, the exhaust valve would close instantaneously at top dead centre and the inlet would open at the same time. However, the laws of physics prohibit instantaneous movement (and infinite accelerations!) and anyway, having both valves open a certain amount at the same time has advantages.  

First, the delayed closing of the exhaust valve encourages the last remnants of exhaust gas trapped in the clearance volume, where the piston cannot reach, to exit down the exhaust port. And second, the opening of the intake valve before piston top dead centre allows a fresh charge to enter, helping the exhaust gas to flow out of the exhaust port and replacing it with an additional charge to be burned during the following cycle. Provided little or no fresh charge is lost through the exhaust valve, scavenging the cylinder in this way increases the total volume of intake charge and also removes the potential ‘heatsink’ effect, when heat from the following combustion process is used to heat up the trapped exhaust using heat that would otherwise be turned into power. If ever there was a case of a double whammy, this is it.

The amount of overlap shouldn’t be too great though, and it very much depends on the restrictions to the intake and exhaust flow, both upstream and downstream of the cylinder. At a given engine speed, a four-valve chamber for instance will usually require much less overlap than, say, a two-valve head since scavenging in the former is more efficient. On the other hand, a race engine more used to running up to 6000-7000 rpm may well require an overlap much larger than that of a road-based machine. In such cases, a race engine could demand well over 100º of overlap for maximum power, whereas our roadcar is quite happy with 15-25º.

For comparison, a few years on from my garage-floor lesson, the Coventry Climax FWMV V8 of 1965 revving to just over 10,300 rpm at the time used a constant 89º of overlap to produce its maximum power. Three years later, the Cosworth DFV revving to only 9000 rpm used 116º. In road-going cars of course, maximum power isn’t always the only target – also of primary concern is engine idle ‘quality’, the ability to exhibit a stable engine idle speed using minimum fuel and giving minimal exhaust emissions, which is why most modern road engines use variable cam timing systems to alter this overlap when the engine is running.

In this way the manufactures get to have their cake and eat it. The overlap is minimised for best idle quality but increased again as the engine speeds up to give optimum scavenging when maximum performance is demanded.

Things have come a long way from those early side valve days in my life, but the sight of those two valves sticking up at the same time out of the top of the cylinder block alongside the tops of the pistons is still lodged in my memory.

Fig. 1 - The 1965 Formula One Coventry Climax V8 valve timing

Written by John Coxon

Using simple theoretical analysis it’s easy to see why. In any sliding element, as soon as the boundary friction has been overcome and the friction drops then, as the relative movement necessary to generate the wedge of oil increases, the resistance to movement (equivalent to friction) actually starts to build up again. In the case of rolling contact where, strictly speaking, there is only line contact across the cam-follower mating surface, the absence of relative movement suggests the absence of friction or its inevitable result, wear.

From the outset though, it must be appreciated that rolling contact components are generally much larger than sliding ones and are therefore likely to be much heavier, in general making them useful only in relatively low-revving larger engines as opposed to the much smaller faster-revving units that use sliding technology.  

Apart from the apparent reduced friction, there are of course other issues. For example, a rolling contact increases the contact stresses where the cam meets the follower. Anyone who understands the theory of Hertzian stresses will realise that the maximum stress in the camshaft will occur just slightly below the surface of the cam, and is a function of the radius of both the follower and the instantaneous radius of the cam at the contact point. The larger the roller, the lower the Hertzian stress, but the larger the roller then the heavier it is likely to be and the more difficult it will be to incorporate it within the engine.

The presence of the roller will also change the valve lift curve, simply as a result of the geometries involved, and while it may be possible to generate more rapid valve opening by using negative-radius cam profiles, this further increases the surface stresses, so even higher quality fatigue properties on the cam material are needed to prevent the formation of surface pitting.

While the analysis may also assume little or no slip between the rolling element and the cam, in reality that may not always be case. Slip is almost unavoidable, as a result of elastic deformation of the parts. This increases the surface area over which these forces are applied, reducing the contact pressure and increasing the likelihood of slippage or skidding. Minimising distortion by using materials with a higher Young’s modulus will increase contact pressure but will also increase the levels of Hertzian stress. As ever in the design process, the solution will inevitably fall somewhere between the two.

Whichever way you look at it, roller followers may sound like a good idea at the outset, but like many good ideas the exploitation of the benefits may be down to the matter of detail design.

Fig. 1 - Cam and roller follower

Written by John Coxon

Little wonder then that vehicle marketeers, all too keen to tap into our collective psyche, are now trying hard to rekindle those feelings by offering vehicles which, while modern and therefore refined in all other respects, have reproduced those little characteristics that make us reflect back. One UK car manufacturer recently introduced a new sportscar that ‘pops’ and ‘bangs’ profusely – deliberately so if desired – on a closed throttle. And now camshaft manufacturers are in on the act by designing cams that make you think of race engine idle quality with roadcar tractability, all using modern hydraulic roller technology produced for more mundane reasons. Generally referred to as ‘thumper’ camshafts as a result of the ‘thumpety-thump’ sound generated at idle, their design is a little more considered than you might think.

So what factors did the cam developers take into account, and which characteristics did they investigate to generate the right aural quality for their customers?

First they analysed a typical race cam and looked at how the interaction of the timing and speed of the exhaust opening position contributed to the volume of the sound and tone of the exhaust note. They reasoned that the sound of the exhaust emanates from the point when the exhaust begins to exit the combustion chamber, with the pressure dropping rapidly and combustion still taking place.

Next they looked at the intake valve closing point. This is traditionally the most important valve event in the cycle since it governs the amount of air to be trapped in the cylinder and which can be burned to create power. Early intake closing traps more air at low speed, whereas later intake valve closing is more effective at higher engine speeds. The correct selection of these optimises the power band to produce a lazy tractable unit or one that is a free-revving thoroughbred race engine waiting to be unleashed.

However, the third area they looked at is what power unit engineers refer to as the valve overlap, and is the area where what might have otherwise become an out-and-out race unit had it not been for a slightly different way of thinking. The overlap period in any engine is the time between the exhaust valve fully closing and the intake valve beginning to open. With both valves open, the intake charge can pass through the combustion chamber and help scavenge the exhaust gas, as in a race engine. Too much overlap can therefore cause an engine to run unevenly and misfire at idle speeds, but when not to excess it can give the characteristic loping sound much associated with a race engine. In researching this part of the profile, designers found that the shape of the overlap (and not just the duration) was equally important, and by carefully profiling the closing and opening ramp the right noises could be made.

Thus in opening the exhaust valve at the correct point, reducing the overlap compared to a race engine and closing the intake earlier, the ‘thumpety-thump’ sound could be reclaimed.

And if you ever want to be converted just go out and listen to one.

Fig. 1 - The ‘thumpety-thump’ cam

Written by John Coxon

Historically, engines have been designed using the technology available at the time, and where these days we prefer to use multi-point fuel injection, as recently as 30 years ago the preferred option would have been a single carburettor. So back then, when it came to ‘optimising’ engine performance across all the cylinders, the starting point was far from ideal.

But engine tuners have known this for a long time, and where the compromise of a single carburettor per engine has had to be retained, for example because of the rule book, racers have applied their ingenuity to ensure that the limitations placed on one component are to a certain extend offset by re-engineering another. In NASCAR for instance, where power increases of 20 bhp per year were regularly seen without changes to the rule book, the trick was to alter the cam lift and timing events on each individual cylinder to ensure that the airflow across cylinders was as near as possible the same. That of course could only be achieved in a narrow range of engine conditions, which is fine on the ovals but becomes harder where a wider range of engine conditions has to be endured. And applying this type of thinking to single-carburettor road vehicles was considered out of the question.

Taking their tips from the NASCAR teams, camshaft manufacturers these days are now beginning to offer a similar concept for those with classic street engines. These still retain the traditional four-barrel carburettor, but by doing nothing more than evening out the airflow between cylinders, increases of 5-20 hp are claimed.

Put simply, these manufacturers have essentially divided the traditional V8 into two separate engines. With a central carburettor, classified by intake runner length (see Fig. 1), the outer cylinders constitute one engine while the inner four cylinders form the other. Situated such that the inner cylinder intake runner lengths are roughly equal but different from the outboard cylinder runner lengths, it is inevitable that the airflow distribution between the two groups throughout the engine speed range will be different.

However, if the camshaft lobes on the inner cylinders were designed to be slightly different from those on the outer cylinders then the possibility exists to even out the airflow across the engine. The result of all this is that the inboard cylinders have the same cam profile grinds, whereas the outboard cylinders will have their intake closing event delayed by around 2º along with a slightly increased valve duration. With the exhaust cam profiles being not only different from their respective intake cam profiles but also different between inboard and outboard cylinders, balancing the airflow between the cylinders is the ultimate ambition.

Sometimes referred to as ‘four-pattern camshafts’, is this not yet another example of where racing improves on the street car world?

Fig. 1 - One half of a single-carburettor V8 engine showing the differences between inner and outer intake runner lengths

Written by John Coxon

A simple enough question for the uninformed, you might say, but one that still left me a little bemused and confused – rather like asking, “Why do the pedals on a bicycle go round if a fish has no legs?” I decided this required more than a cursory reply, so I gathered my thoughts and prepared to explain.

“To start off with,” I began, “the accuracy of a grind refers to the dimensions of the camshaft profile actually ground compared with the values required from the design process. If the maximum lift of, say, your intake valve was calculated to be 10.000 mm on a direct-acting tappet then our camshaft manufacturing colleague was saying that their machine would reproduce that grind to a maximum of 10.003 or a minimum of 9.997 mm away from the camshaft baseline. Furthermore, all the profile lift dimensions to either side should be assumed in the band ±0.003 mm as well.”

At this point I had to admit that camshaft profile designs were often supplied to six, even eight decimal places, so three decimal places – while impressive from a machining viewpoint, compared with the computed profile – might not particularly impress the designer. In addition, stating the ‘accuracy’ of the profile grind in this way doesn’t necessarily describe the ‘quality’ of the grind since the smoothness of the grind isn’t mentioned. While the grind could lie in a band ±0.003 mm either side of the desired profile, it could in theory still be a little like a hacksaw blade [Fig. 1], momentarily introducing high levels of acceleration and even higher rates of jerk, which in practice could result in many valvetrain issues. Although only really just a passing comment on the quality of their machines, what our camshaft friend should have stated was something on the smoothness of their cams; that would have meant so much more.

As to the question about tappet clearances, like my comment about fish and bicycles, tappet clearances and the accuracy of the cam profile are unrelated. Tappet clearances allow for differential thermal expansion of the valves and cylinder head so that the valve will always seat under all likely thermal conditions in the engine. And we adjust the tappet clearance using shims or graded tappets to take into account manufacturing variations so that initial contact between the cam profile and the rest of the valvetrain occurs at more or less mid-ramp at the beginning and end of the valve lift. But I guess you knew that anyway.

Fig. 1 -The theoretical hacksaw-like profile of a camshaft grind 

Written by John Coxon

When the shaft is revolving at a speed such that the frequency of the repeated applied torques or one of its harmonics approaches the natural frequency of the shaft then resonance occurs. At this condition, if the repeated torsional loading is the same as this natural frequency then this strain will never have time to dissipate, and the amplitude of the vibration may increase until catastrophic failure occurs. If you consider this in its academic form – that of a simple spring-mass system, with the addition of a damper or in the case of our crankshaft a torsional damper ‘tuned’ to the dynamics of the system – then catastrophe can be avoided.

If that is true for the crankshaft then the same can be said of the camshaft. In the past, however, camshafts have been very robust devices. More often than not made from solid chilled cast iron, in relation to the forces transmitted in opening and closing the intake and exhaust valves they were very stiff. Lately though, the increased loading given to them, and the unrelenting impetus to reduce engine weight, has also introduced new issues into the design. One of these is the problem of torsional vibration.

The first and most obvious cause of this is that of increased camshaft loading. When camshafts were mounted in the cylinder block close to the crankshaft, the reciprocating motion to open and close the valves was provided by a simple pushrod leading up to the overhead valve gear. The natural limit to the forces induced in the system was therefore the stiffness of this pushrod. Under the high loads demanded by rapid valve opening, the pushrod could ‘bow out’, leading to a loss of valve motion if it was insufficiently robust. Moving the camshafts to the cylinder head and operating the valves more directly improved the stiffness of the valvetrain and therefore the accuracy of the valve opening and closing; later though, by adding the further requirement to deliver high-pressure fuel required by direct injection systems using camshaft-operated high-pressure fuel pumps – particularly those of common rail diesel engines – these forces rose even more.

At the same time though, in the bid to reduce overall engine weight (and not necessarily the rotational inertia of the cam, as many might think) camshafts that are bored, cast or assembled to be hollow will be intrinsically less stiff and will therefore resonate at a lower frequency. If this lower frequency falls into the operational speed range of the applied valvetrain or pump forces then torsional vibration issues may be introduced.

If all that wasn’t enough, the introduction of roller cam followers has made matters worse. Flat tappets are renowned for their higher levels of friction, and use that friction to dampen the effects of vibration. In reducing the friction in the valvetrain by as much as 70-80%, the use of roller cam followers releases the energy that would otherwise be absorbed in friction back into the camshaft strain oscillations, and reduces this damping effect. So while the valvetrain is more efficient, this increased efficiency could introduce camshaft torsional vibration concerns as a consequence.

While not always something we would traditionally consider during the initial stages of engine design, camshaft torsion vibration – the result of a low-friction, lightweight valvetrain – is something we may need to take more interest in in the future.

Fig. 1 - Camshaft

Written by John Coxon

To demonstrate this last point, on the one hand a 2 litre road engine running under typical urban driving conditions (around 2500-3000 rpm) could consume anywhere between 400 and 625 W, or about 30-38% of the total friction losses depending on the grade of oil in use. On the other hand, a Formula One engine at maximum power speed will absorb nearer 10 kW, or 16% of the total friction power. With such a wide discrepancy in percentages and power necessary to drive these valvetrains one can only conclude that the factors involved are complex. And this is generally accepted to be the case since valvetrains operate mainly in the mixed or boundary layer region of the Stribeck curve.

For race engines, however, the friction generated in the valvetrain has to be balanced against the expectation of the increase in engine performance. The rapid opening of intake and exhaust valves will clearly have an impact on the torque required to turn the cam, even though much of the energy used in opening will be delivered back from the spring in the closing of the valve. Higher forces, according to classical theory, will result in greater fluid film pressure and consequently thinner oil film thickness, to the point where intersection of the asperities of the two rubbing surfaces will engage further, leading to higher levels of friction.

Even if moving away from classical theory and considering the elastohydrodynamic effects, distortion of the mating surfaces under the increased loads generated will increase the surface area in contact and at the same time, since frictional force = µ (the coefficient of friction) multiplied by area. The result, whichever way you look at it, will increase friction, but unlike our road-going design engineers where wide-open throttle performance is secondary to other things like fuel efficiency, in race engines higher cam friction can be tolerated.

You might say that converting the rotational motion of something like a camshaft into the reciprocating action of an intake or exhaust valve is bound to introduce high levels of friction, since the process of starting and stopping the valve has to move through the boundary lubrication zone and into the mixed portion. But where engine speeds are relatively low and valvetrain friction is seen as critical, it may be possible to replace the sliding contact surface of the traditional cam and follower by one where only rolling contact exists, using some kind of roller bearing follower. And since rolling contact, with no sliding, is the most efficient way of transferring motion it is no wonder that all new engines designed for road applications may well conform to this technology.

Road engines and race engines – the valves need to do the same thing but the design compromises differ widely.

Fig. 1 - Camshaft 

Written by John Coxon

Cam phasers, to adjust intake or exhaust timing (or both) are one way to limit the damage, but to tackle the real problem, many engineers believe that the camshaft has to be discarded completely. So by taking direct control of the valves, and timing their opening and closing events as well as the amount of lift – by hydraulic, pneumatic or electromagnetic means – the design of the internal combustion engine could be on the cusp of a whole new lease of life. As ever though, it is all down to cost and complexity.

Hydraulic systems seem to be the least expensive and perhaps the most practical way so far of doing this. Although generally not totally camless, hydraulic systems tend to use an actuator activated by a single camshaft – usually the exhaust camshaft – and an integrated fast-acting two-way solenoid valve operated by valve control software. In contrast to conventional mechanical valvetrains, where the cam contour is transferred to the engine valve via a rigid element (such as a tappet or finger follower), hydraulic systems use engine oil in a high-pressure chamber as an intermediary.

When the solenoid valve is closed, the system operates as an hydraulic ram. However, when open, the cam and valve are effectively disconnected, allowing a wide number of lift/timing combinations within the cam profile. Although not totally camless, the traditional throttle is all but redundant, and many of the benefits of independent valve lift and timing can be demonstrated.

The most popular way to dispense with the cam, according to the current way of thinking, would seem to go electromagnetic. Such systems are generally based on linear spring-mass actuators [Fig. 1] and use the principle of the potential energy exchange between two springs and magnetic coils to control the position of the armature to which the valves are attached. For such a system to be acceptable in use, key characteristics should offer the quickest opening and closing times for high-speed operation, have an acceptable impact velocity when closing and, above all, low electrical power consumption. 

A typical such system will consist of an upper and lower magnetic core which together will define the lift of the valve. Actuated by the valve control unit under closed-loop control from a signal generated by the engine ECU – much like a modern-day fuel injector – the current necessary to energise the coils is modulated using voltage pulse width profiles. Permanent magnets are used to improve the response characteristics of the system, in particular to produce better control during the opening and closing ramps where the impact velocities at the valve seat and as well as those between armature and valve head have to be controlled – in fact, just like the opening and closing ramps of a traditional camshaft!

To improve the opening velocity, parts need to be lightweight commensurate with adequate durability, while the spring rates of the two springs needs to be high. Other electromagnetic systems can use hydraulic ‘snubbers’ to control the valve seating velocity, but the resulting action is much the same. Clearly, even these designs are not totally ideal, as they have fixed valve lifts, but while these systems are not yet market-ready, given sufficient impetus in the way of governmental fiscal penalties and fuel economy or CO₂ targets for passenger cars, that day will come soon.

Fig. 1 - First-generation electromagnetic system

Written by John Coxon

So while there are now systems in many modern engines to advance and retard both inlet and exhaust camshaft phasing to improve the combustion, to fully optimise it requires dispensing with the camshaft altogether. For only when opening and closing the intake and exhaust valves, and adjusting their lift and timing to suit the optimum in-cylinder conditions more or less independently of the crankshaft phasing, can we get the best possible thermal efficiency.

If instead of the throttle the engine is controlled by adjusting the lift and timing of the intake valve events, the opportunity can be taken to reduce cylinder pumping losses, and the traditional camshaft as we know it can be discarded. Including a throttle plate in normal engines – convenient as it is – increases the amount of negative work on the piston when the piston is pumping against higher intake manifold pressures and partly closed throttle at part-load conditions. Since the engine air consumption and hence power is effectively controlled by the opening and closing of the valves, rather like in compression ignition (diesel) units, there is little need for a throttle, except perhaps when the engine is at idle.

Furthermore, high cycle efficiencies are best obtained with combustion occurring early in the expansion stroke. Independent control of the lift and timing of the intake valve can initiate large amounts of mixture turbulence to assist mixing, and the fast burn rates thus generated will minimise the heat loss through the walls of the cylinder. At lower engine speeds, delaying intake valve opening can also increase charge velocity, creating better mixing with leaner mixtures which could make engines much more flexible in use, as they would need fewer gear ratios for maximum acceleration.

Higher cycle efficiencies can still also be generated by increasing the expansion ratio. The internal combustion engine converts energy into power from the expansion of the exhaust gas. The greater the expansion, the greater the work (and hence power) produced. Limitations with conventional valvetrain technologies require the exhaust to begin opening well before bottom dead centre. While this ‘blow-down’ effect is good for midrange and high-speed performance, at lower speeds the effect can be negative as well as wasteful on fuel. Delaying exhaust valve opening until nearer bottom dead centre at low engine speeds could produce greater expansion and hence work done on the piston. 

If all of the above were not enough, a camless engine can surely improve engine response – arguably the biggest advantage of all. Modern multi-cylinder electronic engine management systems have the ability to alter ignition timing and the fuel injected on a cylinder-by-cylinder basis. Regretfully though the air needed to burn that fuel can’t be controlled that rapidly. From a driver input it takes time to adjust the throttle angle. More time is lost while the manifold/intake runner fills up to the pressure demanded, and so it takes many engine cycles until the air charge is equivalent to that demanded by the driver. Consequently the ability of the fuelling and ignition to respond on a cylinder-by-cylinder is mostly lost.

With a camless engine the need to move the mechanical throttle is largely lost, and so delays associated with it no longer exist. In fact there is no longer a requirement to have a plenum, except perhaps to collect cool air and direct it towards the intake. Irrespective of where the throttle plate is fitted, therefore, in the un-throttled mode of the camless engine all these time losses disappear. Consequently, during rapid changes in driver demand, the air charge would change instantaneously for each engine cycle, making full use of the qualities of modern management systems and giving even better engine response.

Could it be that the camshaft as we traditionally know it is on the way out? Next time we’ll go into the current options for camless technology.

Fig. 1 - Comparative valve opening at low engine speeds

Written by John Coxon

We know for instance that early power unit engineers understood, as far as they could at the time, the importance of valve timing. They knew for instance that, for best power, cams needed to be of long duration, while for driveability, shorter-duration cams were far better. Furthermore, for equal valve area and lifts, the inlet cam duration should be slightly shorter than for the exhaust cam. Clearly this is not always the case but it’s a good starting point. Thus in the case of the 7.6 litre four-valve Peugeot Grand Prix engine of 1913 the direct-acting inlet cam duration was 223° crank angle, while that of the exhaust was 248°. In developing such an engine, solid engineering principles will have been used together with fastidious experimentation, but in all of this nowhere is it recorded how the actual cam profile was developed.

From various company archives that still exist, pre-1930 camshaft profile data for manufacture tended to be recorded in terms of the radius of curvature of an arc and the centres around which this was based. This was as much for ease of manufacture as anything else. With an opening arc, a closing arc and the bit in between describing the radius of the ‘nose’ of the cam, these were described as ‘three arc’ cams, the arcs actually referring to the base circle, the flank and the nose. Comparatively easy to grind, the cams were designed such that the tappet velocities at the intersections were continuous, to produce a simple harmonic motion of the valve. Unfortunately, although the velocities were the same, this produced instantaneous changes in acceleration that would lead to extremely high levels of ‘jerk’ (the rate of change of acceleration), creating unwanted harmonics in the valvetrain and leading to valve surge or spring failure – or at the very least a noisy and inefficient valvetrain.

And while the three-arc cam had no ramps to lift the valve initially off its seat, the introduction of these into the profile led eventually to the five-arc cam, and so things developed. If valves were pushrod-operated, flexing of the pushrods (and therefore lost motion in the system) could result, but the effects of high rates of jerk along the rest of the valvetrain would be advantageously constrained in some way. Likewise, in some overhead cam designs the flexing of the rocker shaft could help protect the rest of the valve train. Luckily enough, most engines during this period were not particularly high-revving, the 1913 Peugeot revving to only about 3000 rpm, while 20 years later, Vittorio Jano’s 2.7 litre straight-eight Alfa went up to a little over 5600 rpm.

So if you’ve ever wondered why vintage and classic vehicles are so much more faster than they used to be, maybe it’s not just because of the tyres or of the newer and stronger engine materials used, maybe it’s because with improved valvetrain design methods brought about by a greater understanding of the valve train, the cam profiles now being ground are so much better.

Now I bet many of you didn’t think about that.

Fig. 1 - The three-arc camshaft lift, velocity and acceleration curves

Written by John Coxon

The problem with the camshaft is that other than tinker with the profile you can seemingly do very little with it. Sure, you can drill a hole down the centre and take out a little weight, and even spark-erode some of the zone between the base circle and the cam lobe. I have even heard of Formula One teams designing ceramic, net-shape lobes (for they would need to be) and assembling them on steel tubes, but these were prototypes (and unsuccessful ones, I am told) and never made the light of day. Not too long ago, people were even talking about disposing with the camshaft altogether, preferring to replace it by electrically operated solenoids, and when that didn’t come to fruition, they moved on to using hydraulics to operate the valves. Over time, however, many of these schemes have been abandoned, which is why the vast majority of valvetrains around the world are still mechanical in nature.

But the camshaft as we know it seems to be here for a while longer, and therefore ways of improving it would seem to be the only game in town. Inevitably one of these ways is to reduce the weight.

Hollow cams are of course nothing new. Used purely for saving weight, any reduction in rotating inertia can only be marginal – first because the material is taken from the centre of the cam, where the rotational inertia of the cam is least effective, and second, when running at half that of crankshaft speed, the inertia of the cam, when referred back to the crank, is less in any case. The inertia of a shaft through a gear system is inversely proportional to the square of the speed ratio, so whatever reduction in inertia is made to the cam is proportionately less than the square of the speed ratio when transposed to the engine output shaft and hence the flywheel. So if the saving in inertia is only marginal anyway then the only benefit of hollow cams is principally down to the reduction in engine weight – which of course is always worth having.

And while some cams can be cast hollow, others can undergo somewhat expensive and quite difficult gun drilling operations. But once drilled they can be used to feed oil down to individual bearings or perhaps even the cam phaser on a VVT system. 

Another alternative that is steadily gaining converts in the original equipment world is the ‘assembled camshaft’. Consisting of a hollow steel tube with forged or sintered cam lobes positioned on it, the tube is expanded either hydraulically or by passing a carefully sized mandrel down the centre, and through it thereby anchoring them. Simple, if not easy to do in practice, this is surely only for manufacturing in volume.

But for those with deep pockets but who face restrictions on what they can and can’t do, camshafts for Formula One are simply machined from solid from the finest of steels. Rather better than even aircraft-quality steels, these have minimal inclusions that enable the tubular portion of the shaft to be machined to a minimum of 1.5 mm thick. 

High strength? Certainly. Light in weight? You bet. But cheap? I’ll let you decide.

Fig. 1 - CAD rendition of an assembled camshaft

Written by John Coxon

For most applications I guess it is safest to suggest that, when installing an aftermarket camshaft into a production-based engine, follow the guidelines as issued by the cam supplier. These may request the engine builder to set the peak inlet valve lift to a certain number of crankshaft degrees after top dead centre (TDC), when the piston is at its highest in the bore. In some designs, while it is relatively straightforward to determine the piston TDC and therefore the number of degrees after it, because of the dwell period (when the cam remains fully open) this maximum lift approach is not particularly useful. In such cases it is better to plot the lift versus degree curve towards maximum lift, and the same again after the peak when the valve is closing again. In this way the average of the angles at the same lift will correspond to that of the peak.

In high compression ratio engines when valve-to-piston contact could occur, it might be wisest to assemble the valve using lighter valve springs to establish the valve-to-piston clearance around the inlet valve opening (and exhaust closing) and piston TDC. Using lighter valve springs and a dial gauge on the tappet, the valve can be pushed down to the point of contact at each degree through this critical zone. As well as minimising any potentially catastrophic valve-to-piston contact, this process will establish how much tolerance can be accepted in the timing before serious damage can take place. This is something I believe needs to be established quite early on in any build process should any cam timing optimisation work be contemplated at a later stage.

The timing method above of course assumes that the profile is fully symmetrical, and for many profiles this is still the case. However, in a growing number of engines where, for instance, roller rocker systems have been incorporated, cam lobes are not symmetrical. In cases such as this the cam grinder will no doubt have given you the cam timing at either 1 mm or 0.050 in lift on both the opening and closing flanks. Measuring the actual timing by turning the crank in the direction of rotation, and noting the crank degrees at both 1 mm (or 0.050 in) lift at the opening and closing side of the cam, and comparing them with that required, will soon move you in the right direction.

But assuming you have determined the actual timing as assembled and compared to that required by the cam grinder, invariably there will be a difference. To correct this we have a number of options.

The first and perhaps simplest of these is to use a fully adjustable vernier device. Rotated to the amount necessary and then securely clamped, this is quick and easily understood, taxing the brain as little as possible. However, these vernier devices have been known to slip and are comparatively bulky, so a lot of builders still prefer to use other methods. The other options of ‘offset keyways’ or multi-hole timing wheels are, in my opinion, more robust, lighter, above all cheaper but may require mental dexterity to fit the correct amount of offset or align the correct holes to give the timing required. Once set though, the timing is unlikely to change – unless something else has gone drastically wrong!

Fig. 1 - Timing a Formula Ford 1600 engine

Written by John Coxon

In the design of the camshaft profile there can be many constraints, such as the required maximum lift required to deliver the rated performance. From this are derived all the practicalities of achieving this lift and, crucially, that point relative to the piston motion which best produces the engine torque curve desired. Following on from this we have the maximum velocity of the valve often limited by the diameter of the tappet in direct drive systems, and of course the accelerations and decelerations of the valvetrain controlled by the selection of the valve return spring.

Jerk – the rate of change of acceleration – is another parameter. It’s not perhaps considered as important as those of velocity and acceleration, but high rates of jerk can introduce other problems of a dynamic nature, when abnormal follower motion such as jumping, bouncing or indeed spring surge can result in loss of control of the valve. Attributed to high-frequency harmonic motions within the valvetrain, to minimise this loss of control, profile smoothing techniques are now most commonly used.

If the cam profile is considered as a continuous surface made up from a number of mathematically described polynomial curves, then alongside these constraints the designer also has those of simple continuity. Thus, using a simple polynomial equation, the velocity of the surface of the cam at the end of one of these curves ideally needs to be equal to the beginning of the next, to minimise any instantaneous jump in velocity. Likewise, any sudden change in velocity will create unnecessary jerk, which can excite any of the cam harmonics and lead to dynamic issues in the valvetrain.

The problem with polynomial equations is that they are difficult to manipulate. So cam designers in the past have come up with the concept of ‘splines’: the mathematical equivalent of ‘French’ curves. These are easier to manipulate, and when used in conjunction with a multi-polynomial curve will ensure not only that the end points of each section of the curve will be the same as the beginning of the next section but that the gradients at the intersection will be the same.

Depending on the order of the spline, the various derivatives will be matched at the joins or ‘knots’ where two curves meet. For example, a quadratic spline will have a discontinuity at the second derivative (acceleration) since the first derivatives of each curve will be matched. The velocity will therefore be continuous but the acceleration will be discontinuous. Likewise a cubic spline (third order) will be continuous at the second derivative (acceleration) but discontinuous at jerk, while a fourth-order spline will have continuous jerk derivatives.  

So the surface generated using these techniques will be hugely improved, and the resulting valvetrain motion – unlike Bond’s dry Martini – will be neither shaken nor stirred.

Fig. 1 - All well-designed camshaft profiles will have been through some kind of profile smoothing process

Written by John Coxon

For the vast majority of race engines the direct acting system (or DAMB - Direct Acting Mechanical Bucket), where the cam acts directly on the tappet-valve assembly, is that most frequently found. Where friction is the overriding concern, roller finger followers are the trend, and most passenger car engines limited to 6000-6500 rpm now use this method. But for the DAMB - which is comparatively light in weight but above all stiff - the downside is the limiting of the valve opening velocity by the diameter of the tappet, and of course the high levels of friction. For race engines as well, high engine speeds mean large spring forces, which means greater valvetrain friction particularly at the lower engine speeds when the inertia of the valve is low. Since race engines tend to have larger bores/stroke ratios and hence greater valve area, the valves tend also to be that much larger.

At the cam and tappet contact point, this loss in power is a function of contact load, relative sliding velocity and the friction coefficient. So while the contact load is a function of spring and inertia forces, the sliding velocity is a function of cam profile and engine speed, while the coefficient of friction has many more variables. Contact force, relative sliding velocity, surface textures and the lubricant properties of film thickness, temperature and viscosity are all involved, so an engine with a minimum of valvetrain friction has to juggle all these - and perhaps the easiest of them all to change is the surface, both its type and texture. It is therefore little wonder that of the options available to defeat friction, a change in the surface is one of the simplest and most cost-effective.

For those on a budget, superfinishing the cam would probably be the first choice. Placing the cam into a container of vibrating carborundum stones of various sizes for various times has the effect of removing all the sharp peaks on the surface of the lobes and prevents them from poking through the oil film. The lubrication regime between cam and tappet will move from boundary or mixed into full hydrodynamic, reducing both wear and friction.

At high speeds, however, some of the oil that would otherwise have been rooted in the creeks and valleys of the original surface may simply be centrifuged away. The top-end lubrication should therefore be more carefully controlled and not simply left to the effects of oil splash in the cam box.

Another method is to coat the cam - or, perhaps more practical, the tappets - with one of the many ceramic or diamond-like coatings available. Titanium nitride, chromium nitride, chemical vapour deposition (CVD) or plasma vapour deposition (PVD), all these coatings have higher hardness values than steel tappets. The very thin layers, as little as 1 micron thick, do not normally affect the surface finish, but care must be taken during some PVD processes to ensure that the surface roughness does not increase. Increasing the surface roughness will have the opposite effect to that desired, so if reduced friction is desired then take care in selecting your process.

Fig. 1 - Cam drive torques

Written by John Coxon

One way this engine torque (read bmep) can be manipulated is by altering the camshaft timing - the opening and closing of the intake and exhaust valves to change the gas dynamics. But as we know, bmep is not just about gas dynamics but is the difference between imep, (indicated mean effective pressure) and fmep, friction mean effective pressure. And when it comes to minimising fmep (to maximise your bmep) alongside, perhaps, the friction generated in the ring pack or the aerodynamic windage losses in the sump for example, the torque necessary to power the camshaft must also come under scrutiny. Not quite so glamorous but equally important therefore is the design and development of the valvetrain.

In calculating the power to open a valve, the torque is essentially the product of the eccentricity (the distance from the point of contact of the cam lobe with the follower to the centreline) and the vertical force between the two.

Now, in direct-acting systems the eccentricity of the cam is usually limited to the diameter of the cam bucket since, for reasons of durability, we don't really want the point of contact to run off the side. Invariably there are always exceptions but as a good design principle this is one to hold true, so to minimise the resulting cam drive torques the task is to have either low flank acceleration at high eccentricity or high flank accelerations early in the cam cycle.

The latter approach, however, implies the introduction of high jerk (the rate of change of acceleration) which may or may not introduce other issues. But in opening the intake or exhaust valve against the valve spring, the energy used is mostly recovered when the valve closes again. Thus the torque needed to drive the cam in the opening phase is reflected back into the cam by the valve and spring, producing a torque reversal as perceived back at the cam drive. The difference of course is the friction generated in the system, which is why engine designers go to great lengths to incorporate low-friction DLC-type coatings and the like into their designs. However, these high but short period peak torques and their reversals (sometimes rather more graphically referred to as 'stab' torques) produce not only greater loads into the cam drive but their frequency can also create other issues of resonance well away from the valve.

When designing cams, therefore, not only does the designer need to consider the airflow needed to ingest the desired amount of air, the velocity of the tappet and the accelerations - that is, forces - in the contact zone, but also the degree and magnitude of these 'stab' loads and their possible effect on the rest of the engine.

Fig. 1 - Valvetrain

Written by John Coxon

The limitations of these injectors which both injected and metered the fuel eventually gave way to the common rail system, in which the method of increasing the pressure of the fuel and then injecting it into the cylinder were divorced from each other. Unlike the jerk-pump approach, multiple injections were now possible, and the opportunity to control the combustion more precisely - indeed, to supply multiple injection per cycle - was at last achievable. Advances to the thus named 'common rail' diesel engine resulted inevitably in the ultimate in spark ignition units, the GDI, or gasoline direct injection system.

But while the function of the intake and exhaust camshaft is to open the valves as quickly as possible - to hold them fully open for the required duration and then close them again just as quickly, without valve bounce or loss of control - the requirements placed on the fuel cam of a GDI engine are in some ways significantly different but in others surprisingly similar. So while the intake and exhaust valves are needed to pass as much air as possible, the same goes for the reciprocating fuel pump.

In doing so though, and delivering something like 0.5-1.1 cm3 per stroke at 200-plus bar delivery pressure, the forces on the fuel cam are often considerably greater. Unlike the intake/exhaust cams, the fuel cam works in a different way and it is only on the upstroke when it is moving the fluid. Also, the greater the number of lobes on the cam - whether this be two, three or four - the greater the fluid flow. At the same time, the velocity of the stroke is limited by the forces involved, the diameter of the flat tappet follow and in particular by the limiting minimum oil film thickness between the cam and its follower.

Another issue to consider is the position of the pump. Positioning it at the end of the intake or exhaust cam often introduces packaging concerns and will increase the centre-of-gravity height of the installation, as well as limiting the fuel flow since the cam only runs at half engine speed. Likewise, mounting it lower down or between the cylinder heads of a vee engine and running it at crankshaft speeds can introduce concerns of inadequate lubrication or ensuring the cam follower interface entrainment velocities are not excessive. Running at high crankshaft type speeds will result in oil being centrifuged off the rotating cam lobes, resulting in a lack of lubrication.

While the combustion engineer can only marvel at the increase in bmep from his direct injection designs, the cam designer has more than just a few more issues to contend with.

Fig. 1 - Two-lobe DI fuel pump drive

Written by John Coxon

In designing a cam profile, the normal intention is also to supply springs of a sufficient rate to keep the follower in contact with the cam lobe as it slows the valve down towards peak lift. The higher the rotational speed of the cam, the greater the spring force required, to the point where for a given profile, the cam and follower will eventually separate. This condition is generally known as valve 'float', and as a broad rule cam designers tend to make this around a couple of hundred rpm above the maximum speed of the engine.

But that is not the only condition when the prescribed motion of the valves, as dictated by the cam, can be lost. If we strike an empty cup with a spoon then the frequency of the resulting sound will correspond to the natural or resonant frequency of the cup. Likewise, when continually bounded by the rhythmic action of the cam, the valvetrain in its entirety will also have a natural or resonant frequency - indeed, there may be more than one resonant frequency, because there may be many components, each one having its own.

Transferring our attention to the cam, if we then undertake a Fourier analysis of the cam acceleration curve (Fig. 1) then much information can be revealed. For those not too familiar with the finer points of mathematics, Fourier analysis takes any continuous geometrical curve and splits it into the amplitudes and frequencies - or, more exactly, orders of frequencies - of sine waves that make up the curve.

This is very useful when we need to understand the dynamics of our valvetrain, and particularly the valve spring, since it too will have a natural frequency of vibration which, if excited, will affect the motion of the valve. As cam profiles become more aggressive, the higher frequency motions that make up this Fourier sequence can excite the valve spring at engine speeds well below those designed for valve 'float'. Lovers of the YouTube website can see amazing video footage of this and what happens to valves and valve springs when their natural frequencies have been reached.

Generally speaking, valve springs have natural or resonant frequencies in the 350-600 Hz bandwidth, so whenever possible the more significant of these harmonics should be designed to fall outside the resonant frequencies of the chosen valve spring. Twin valve springs and variable rate springs are but more attempts for the designer to 'tune' this spring to the harmonics of the cam profile, and in successfully doing so peace and harmony will result.

I only wish that requesting older siblings to phone their parents on a more frequent basis were as comparatively easy.

Fig. 1 - Fourier analysis of the camshaft acceleration

Written by John Coxon

The fact is that, in designing a cam profile, we have to lift the valve assembly off the valve seat, accelerate it open to its maximum velocity and then slow it down, bringing it to a halt before accelerating it back in the opposite direction, then slowing it down and eventually landing it gently back on its seat. A simple enough process, you might say, but one nevertheless that has challenged and is still challenging designers.

Take a 'simple' polynominal profile for instance. With the profile described by a sixth-order polynomial

and ignoring any ramps, for our boundary conditions when the valve is closed we have y (displacement) = 0, its derivative with respect to angle (velocity) dy/dx = 0, as well as its second derivative (acceleration). At the other boundary when y is a maximum and x = 0 we have velocity = 0 and acceleration = 0.

Differentiating the equation with respect to x to produces equations for velocity, acceleration and jerk (the third derivative) and substituting in the boundary conditions we come up with a series of simultaneous equations, which need to be solved. Once solved, the equations of displacement, velocity, acceleration (and jerk - the third derivative) can be written and the half-cam profile produced which solves all the given boundary conditions. Extending the power of the polynomial beyond six -perhaps as far as 12, 14, 16, even 24 - and varying the individual coefficients will eventually give you a half profile, which will describe (hopefully) more or less what you want.

Needless to say, using a hand-driven calculator with pencil and paper, and quite a lot of trial and error, the process was only for those of a certain disposition and took many days or weeks to perfect - and even then you still only had half a cam profile. It was little wonder therefore, having got this far, that the simplest approach was one of reflecting the profile about the 'y' axis and bringing the valve down in the same way as it went up.

These days all this is done by computer and the various excellent cam design software that is available, such that the role of the cam designer is perhaps no longer so specialised. In opening the valve we might want to do that as quickly as possible to generate pulse activity in the induction system or exhaust pipe. However, it is not always the case that we want to close it as quickly; hence the idea of the asymmetric cam. In some cases we might want to minimise valve bounce by introducing the valve more gently to its seat. Or in other applications it might be necessary to keep the valve open a little longer, introducing more air and swirl into the chamber during the combustion process. Either way it might be beneficial to use a different closing profile to that of the opening one.

What goes up, must come down - but not necessarily in the same way.

Fig. 1 - Calculating profiles using pencil and paper was not for the faint-hearted

Written by John Coxon

With the penalties measured not only in lost kilowatts but in punitive governmental fines in the form of tax, the focus on the mixed lubrication regime that is the valvetrain is now as intense as never before. And when you appreciate that the vast majority of race engines are developed from those originally designed for the road, then the sooner we get to grips with these roller finger follower (RFF) designs the better.

For many a long time the direct acting mechanical bucket (DAMB) valvetrain has been the staple of the performance engine valvetrain. Simple, compact and therefore exceedingly stiff, its limitations are well understood, as indeed is the amount of friction generated by the relative sliding of the tappet in its bore and the face of the cam as it slides across the tappet. Much better designs, at least from a friction perspective, are those of the RFF and roller rocker arm (RRA). RRA designs, although they produce shallower cylinder heads, tend to have an inherent lack of stiffness in the rocker arm, so no wonder more and more manufacturers opt for RFF designs.

Not as stiff as the DAMB but better than the RRA, these rolling contact designs do have major downsides though.

The first of these is an inherent disadvantage of any rolling contact when the contact stress between two curved surfaces is much greater than that if a curved cam meets the flat (or very slightly curved) surface of a bucket tappet or finger. Because of their size, these Hertzian stresses can therefore limit the choice of material, making cast iron undesirable. However, even in using the best of steels, a larger nose radius will be necessary, which in turn for modern high lift and rapid opening/closing will require the use of 're-entrant', concave or kidney-shaped cam profiles to be ground on its flanks.

The process of grinding 'normal' cams uses a comparatively large grinding wheel (between 400 and 600 mm in diameter). The larger this wheel is then the lower its wear rate and the less time required for 'dressing' it during the grinding of the cam. With cycle times measured in seconds, the quicker a cam is ground the less expensive it is since, as we all know, in manufacturing, time is money. But large wheels are no good for re-entrant cams, since large wheels limit the radius of re-entry, effectively limiting the opening/closing speed of the valve. Cubic boron nitride wheels can reduce this diameter to as little as 50 mm (some grinders claim down to 34 mm) but even so the inevitable increased cycle time on the grinding machine increases machining time and adds much to the cost.

So while roller finger followers save friction and therefore fuel, you pay for it all in other ways.

Fig. 1 - Typical re-entrant or kidney-shaped cam profile

Written by John Coxon

To one individual, a weekend racer of some skill, the perfect cam is one that gives ultimate top-end power with maximum torque only a couple of thousand revs lower. Using his ultra-close ratio gearbox and slick, flappy-paddle gear change system, swapping places between peak torque and maximum power, he could lap all day to his heart's content - absolute bliss!

Another individual, a slightly more laid back character, more used to threading his vehicle up the narrow confines of a hillclimb track, disagrees. Gear changes waste time, and any time when the engine throttle is not at its maximum for the track conditions has to be avoided. To him therefore, the perfect cam is one that gives good low-speed torque with the minimum of gear changing, making the car quite docile really and easy to drive. Such a car enables him to concentrate more on the important things like the perfect line or the optimum braking point.

A third person towards the back of the group steps forward, both hands stuffed into his somewhat grimy overall pockets and says he's inclined to disagree with the first two. "No, to me," he says, "the perfect cam is one that is perfectly straight, spinning easily in its bearings, has no greater than 0.001 in variation of lift between its lobes and has a lobe centre angle variation [the angle between two cam lobes] of less than 0.25º peak to peak. Furthermore, a cam that has any measurable run-out with a dial gauge or that exceeds any of these numbers should be viewed as scrap." And even though I count myself an enthusiast before and engineer, I would confidently support this view.

But of course the best of cams, while conforming to the above specifications, are those where the evidence of the manufacturing process cannot be seen - the stress relieving process, for instance. For cams that spin very quickly (as opposed to those that rotate slowly and are heavily loaded), steel to EN40B specification is an excellent choice. A nitriding steel, after rough machining the component should be stress relieved before the nitriding process and subsequent final grind.

Unfortunately some cam manufacturers, to save money, might opt to forgo the stress relieving process and, unsurprisingly, after nitriding the cam might distort. Attempting to straighten the shaft could generate a weakness, while re-machining can break through the thin nitrided layer. Either way, the cam is destined for a shorter life. The best steel cams therefore are those that have been stress relieved before the nitriding process and are, as it says in an advertisement, "reassuringly expensive".

So in a way, although all three of my fellow enthusiasts were correct, a good cam is one that has been not only accurately machined but stress relieved as well.

Fig. 1 - Cam blanks

Written by John Coxon

In early automotive cam profiles when the design was a long and tedious process using trigonometrical tables (or, worse still, graphical methods), having derived the lift, velocity and approximate accelerations involved, no-one was particularly keen to go to the next step. Ploughing your way through endless polynomials and ensuring the boundary conditions were met for all, one could be forgiven for wishing to go to the next level. Nevertheless, many realised that the impulsive forces generated may have had a detrimental effect on their valvetrain but reasoned that with the relatively flexible valvetrains of the period there was little that could be done about it at the time. Even the late Keith Duckworth, designer of the Cosworth DFV, is reputed to have discounted many of the arguments over jerk and opted to go for constant acceleration designs (interspersed with high levels of jerk at their boundaries, presumably) in his early years. It is perhaps ironic that with comparatively flexible valvetrains, the need to control the rate of change of the forces and the excitation this can produce in the lift harmonics is more acute.

In recent times, when valve profile analysis is readily available at the touch of a button - after considerable effort in inputting the data, I might add - the emphasis is all about the 'smoothness' of the design. I well remember the late Professor Blair instructing me in the absolute necessity of smoothing the transition from one lift phase to another in any design intended for a race engine. According to the professor, the most critical time is often the transition from the initial valve ramp into the positive acceleration phase at the start of valve opening. Insufficient care and lack of smoothing in this zone can lead to higher acceleration peaks and substantially greater jerk for what amounts essentially to the same lift profile. Furthermore, even if the valvetrains are considerably stiffer, high jerks or impulsive forces inevitably lead to avoidable cam wear, pitting or scuffing.

It is perhaps interesting that in many aftermarket cams, jerk does not appear to rate highly in the design procedure. With limitations on velocity (from the tappet diameter) and maximum acceleration, so long as jerk is within reasonable bounds then there are generally no grounds for concern. For direct-acting cams, levels like 0.01-0.02 mm/deg3 have been mentioned, but where pushrod arrangements are involved, maximum levels somewhere nearer half of these have been suggested. But as mentioned earlier, when Fourier analysis of the lift envelope is undertaken, undesirable valve motion caused by spring surge and other resonant effects can occur.

So while it may be advisable to ignore the jerk selling you the cam, the jerk that lies within it may not be quite so easy to avoid.

Fig. 1 - A sample camshaft

Written by John Coxon

It's a sobering thought but no matter how fine the surface finish or accurate the machining, inevitable imperfections - however slight - will lead to contact high spots at first run. Unless these are carefully ground down in a controlled way in the initial stages of running, these can lead to localised surface overheating, micro-welding and scuffing of the cam lobe surface. Eventually, catastrophic failure later on in the life of the engine could occur. The initial starting of the engine and bedding-in of the cam and its associated components is therefore of critical concern.

Cleanliness and the absence of any foreign matter is surely a given, and in order to ensure surface protection during assembly and timing, cam lobes and journals should be well oiled. Many builders go so far as to use specialised assembly lubricants containing an increased amount of anti-wear additives which, being heavily polarised, physically stick to the surface of the cam during the build. As a general rule it is unwise to introduce additional and unknown additives into the engine since some additives may not be compatible with those in the oil, and will produce precisely the opposite effect of that intended. However, if sourced from the same lubricant manufacture, any potential conflict should be minimised.

Before first fire it is always a good idea to build up the oil pressure, not only in the main oil gallery but also those leading to the valvetrain, and ensure that the lobes are well oiled. For many aftermarket dry-sump oil pumps this is a simple matter of removing the drive belt and priming the oil system by hand. With other systems or engines it may be necessary to crank the engine on the starter motor before firing until full oil pressure is achieved.

Once you are happy that there is enough oil where it needs to be, the engine can be fired, while at the same time taking care not to allow the engine to rev. As a rule, slow running kills camshafts so idling at any point during the break-in stage is to be avoided, but initially around 2000-2500 rpm should be fine. If particularly aggressive cam profiles are in use, it may be better to use softer valve springs during the early stages. If using dual valve springs, simply omitting the inner spring during the build, to be refitted later, will have the desired effect. When running just the single (or softer) spring, remember to limit the speed of the engine to avoid the outset of early valve bounce.

Pre-heated oil often works better than that straight from the can. Not only is it less viscous (and will therefore flow much quicker) but many of the anti-wear additives are sensitive to temperature, and don't actually start to become active until around 60 C. During the running-in stage some people also prefer to use diesel-type lubricating oils. These will generally have a greater detergent/dispersant content than those intended for gasoline use, to cope with the increased amount of soot and particulates expected in the combustion gas. Using diesel oils will therefore be better at trapping any wear metal generated and deliver it to the filter or hold it in suspension where it can do minimal harm. Once fully bedded-in, the diesel oil should be replaced with the race oil and double flushed to minimise any cross-contamination.

As oil quality has improved over the years, some specialist oil manufacturers are beginning to offer specific break-in oils for their customers. Formulated to allow controlled bedding-in of components and with a higher than normal detergency, these must not be used for any serious wide-open throttle work. Furthermore, by keeping to the same supplier, any potential unintended additive interaction can also be avoided. These specialist race oils will have a greater level of anti-wear additive and less in the way of detergents than most other API or ACEA formulations.

Taking care of your cam during its early life will produce dividends later on.

Fig. 1 - Take care of your camshaft

Written by John Coxon

To many (except perhaps Ducati owners) the word 'desmodromic' will mean very little. A remnant of a bygone age when valve spring technology was still in its infancy, desmodromic valve gear used one cam to open the valve and another, wholly separate from it, to close it again. The valve spring, if you like, was made redundant, and in theory at least this enabled valves to open and close much quicker, or the engine to rev much higher than would otherwise be possible. But for the arrival of the pneumatic valve system, pioneered by Renault in Formula One, 'desmo' systems might now be much more common. However, improved valve system modelling has given us a better understanding of valvetrain dynamics and, together with better valve springs, has effectively sidelined desmodromic technology - at least for the time being.

In traditional valvetrain systems, as the engine speed increases, sooner or later the momentum of the valve and tappet assembly will overcome the ability of the spring to control it. At this point - graphically referred to as 'valve float - at best the valve can land heavily on the closing flank of the cam, producing abnormal wear and valve bounce, or at worst catastrophic engine failure as a result of piston-to-valve contact. Either way, and somewhat of an understatement, the engine's durability will be impaired. In removing the spring and having a separate cam for closing, as well as the one for opening, this limitation of valve float is reduced, and valvetrain mechanisms can be designed to give faster opening and closing velocities.

The use of desmodromic valve gear doesn't, however, do away with springs altogether. In a practical engine, when the valve has to be fully closed in order to develop compression in the cylinder, light 'helper' springs have to be provided. But turning a desmodromic engine over by hand at slow speed reveals that these systems have very low friction at low speed, unlike traditional sprung valves.

Traditional valvetrains have very high torque requirements at low speed, which steadily fall as the boundary layer lubrication slowly morphs into that of hydrodynamic lubrication with increasing engine speed. Desmodromic systems, on the other hand, start by having very low friction at slow speeds, which steadily increases as the engine revs rise. For maximum fuel economy, friction and engine speeds will need to be kept low, and so in the relentless search for ever improving fuel economy, desmodromic valve trains may yet see a resurgence.

Desmodromic or demonic? I'll let the reader decide, but wouldn't it be nice if mechanisms like these were more common?

Fig. 1 - Ducati MotoGP desmodronic valve gear

Written by John Coxon

I was reminded of this some years ago when a recently built prototype engine had scuffed tappet faces after only a few hours' running. Noticing the condition and reassembling to continue the programme, after a further ten hours and stripping once more, rather than seeing even more surface damage, the original scuff marks were slowly beginning to fade.

Recognising the situation and understanding the precise build procedure, the issue was quickly put down to a lack of lubrication at the build stage and an unsympathetic regime when firing for the first time. The lack of lubrication during the initial running caused scuffing on the softer tappet surface when running against hardened cams. Once oil was present after only a few seconds the initial damage was slowly and steadily polished out until, after several hundred hours, no visible damage was evident at all.

That valve tappets no longer have to revolve to even out wear was evident by this exercise, but only when flooded with suitable lubricant to prevent metal-to-metal contact during the initial stages. Should these particular tappets have been allowed to rotate, I doubt any scuffing would have been observed at all. Historically, however, tappets have always been encouraged to rotate in their housing, and along with them (in overhead cam engines at least) the intake and exhaust valves beneath.

Common in many older overhead valve applications with the camshaft in the block, the tappet would be encouraged to spin by having a point of contact offset to the axis of the tappet. Rather than being completely flat, the tappet would be slightly domed with a spherical radius of 40-70 in (100-180 cm) and the cam carefully ground with a taper to match. At the designed point of contact, eccentric to the centre of the tappet bore, the offset would cause the tappet above to rotate and consequently even out any subsequent wear.

On the other hand, in overhead cam engines - particularly those intended for racing - the current trend is to have entirely flat tappets. Much easier and therefore cheaper to make, the cam can also be produced without a taper, which makes grinding a much simpler exercise as well. Mounted offset between the centre line of the tappet and that of the cam lobe, the tendency is for the tappet to spin. But even if not mounted eccentrically, the tendency (as any engineer will testify) is for the whole assembly - tappet, spring and valve - to rotate anyway, as a result of the spring introducing a slight turning force on the valve as it is compressed and the inevitable torsional vibration induced.

But if you have ever looked at any high-speed video of a complete cam/tappet/spring/valve assembly, it is difficult to conceive a situation where they cannot rotate.

Fig. 1 - Evidence of rotation on a flat tappet

Fig. 2 - Cam-in-block arrangement

Written by John Coxon

weeks (or even months) using a calculator and pencil and paper.

But in all this we must not forget that the camshaft is only the switching element of a valve feeding intake gas to the engine and enabling extraction of the spent gas thereafter. Assuming our mechanism can flow sufficiently to fill the cylinder in the time required, the next significant requirement is the timing of these opening and closing events.

That we are trying to open and close the valves in the shortest possible time must surely go without saying. But in order to do so and gain maximum benefit, these actions have to be instigated at times that to some might make little sense - for instance, the opening of the exhaust valve.

Once the fuel-air charge is burning, the pressure in the cylinder will reach a maximum sometime shortly after TDC (top dead centre). After a further 90º or so of crank rotation, most of the energy has been recovered in terms of mechanical torque on the crankshaft, so opening the exhaust valve well before BDC (bottom dead centre) loses little in actual performance.

The period after exhaust valve opening (EVO) is generally referred to as 'blowdown', and although irrecoverable in terms of energy lost, the extraction effect in the exhaust can assist in reducing the pumping losses that would otherwise be incurred in pushing the exhaust gases out on the piston upstroke. Perhaps the least significant of all cam timing events, EVO generally occurs between 50º and 90º BBDC (before bottom dead centre).

If EVO is considered the least important then, of the four intake/exhaust events in the four-stroke cycle, the intake valve closing position is arguably the most important. Intake system dynamic events tend to dominate the characteristics of any naturally aspirated four-stroke engine. In many past discussions on this subject the unsteady flow in the intake system was considered to be the algebraic sum of two phenomena - wave ram and inertial or column ram.

At the time, column ram was considered to be simply the inertia of a column of air flowing down the intake runner, onto which would be superimposed the oscillating pressure pulses resulting from the combined effects of valve opening and piston movement. As a result of much analysis - and as I was continually reminded by the late Prof Gordon Blair - this is not now considered to be the case, and only the pressure pulses travelling up and down the intake system need to be considered.

Thus the wave ram negative pressure pulses produced as the intake valve opens travel up the runner and are reflected back as a positive wave. As this wave reaches the intake valve again, if the valve is on the point of closing, the slight increase in pressure will improve the volumetric efficiency of the cylinder by trapping that little bit of extra air. Should it arrive back a little later as a negative pulse then the reverse is true and the volumetric efficiency will fall. Ensuring that the intake valve closes at the most beneficial time is therefore the secret to good performance.

So while getting the valve open (and closed) in the shortest possible time is important, ensuring this happens at the correct point in the cycle is perhaps more so.

Fig. 1 - Valve events on the four-stroke cycle

Written by John Coxon

I'm used to listening to hot engines, and beneath the rhythmic pulsing of the intake and exhaust, piston slap is a familiar sound. But this particular noise coming from the tappet chest appeared to me at least to be all camshaft-related. Indeed, it sounded very much as if the camshaft profiles were missing one very important parameter - the opening or closing ramps!

A normal camshaft valve-opening profile consists of three stages. The first consists of lifting the valve gracefully off its seat before rapidly accelerating it open in the second stage and then slowing it down again for the third, when it is fully open. Interrelated, the general idea is to open the valve as quickly as possible and benefit from what is called 'maximising the airflow under the curve' - and, apart from ensuring that the spring and camshaft material can cope, other concerns are principally valvetrain dynamic issues and of course, in road transport applications at least, the noise. But while some noise can be generated from the vibration of the valve springs, much of it comes from the contact at the initial lifting of the valve off its seat and placing back thereupon again.

Traditional practice in the design of the ramp is to specify a constant velocity which ensures that the valve is picked up at a known speed once the initial tappet clearance, in the case of mechanical tappets, is taken up. For hydraulic tappets the amount of sink-down has to be taken into account. On the closing side, the seating velocity is critical, particularly to the durability of the exhaust valve when at higher temperatures material properties may suffer.

The height of the ramp therefore has to be enough to ensure seating once differential thermal expansion, base circle run-out, valve cock and - in the case of hydraulic tappets - other sink-down effects have been taken into account. For mechanical tappets this ramp height may be 0.3-0.45 mm for both opening and closing, with a ramp rate, a preference of the designer depending on the application, of somewhere around 0.015-0.030 mm/degree not being uncommon. In some cases, since seating velocity may be critical, the closing ramp rate may be slightly less than that of the inlet allowing the valve to seat carefully without the risk of bounce.

But in the search for more engine performance it may be necessary to open the valve much quicker, and if that is the case then a number of options are available. The first option, and the one most often exploited by engine tuners the world over, is to reduce the tappet clearance progressively until the power falls away. This has the effect of increasing the area of possible airflow underneath the curve with no additional stress into the valvetrain. However, although peak valve opening positions relative to top dead centre will not change, the optimum cam position may do so but only slightly.

Shorter constant-velocity ramps might also help, but assuming the valvetrain has been designed carefully, these will be no longer than required anyway. The only other way is to pick up the valve at a higher velocity using some kind of constant acceleration ramp rather than one of constant velocity. Although the impact force will be much greater, and consequently durability much less, as soon as the valve leaves its seat it will be accelerating much faster than before, hence in effect opening sooner. A challenge to the designer and presumably very difficult to manufacturer, it might also sound almost as bad as my bag of hammers.

Fig. 1 - The opening ramp of a typical valve profile

Written by John Coxon

Invented by Lotus and manufactured under licence by INA Schaeffler KG in Germany, the tappet consists of an outer annulus surrounding a much smaller inner post, such that the mechanism can operate in two different valve lift modes. Taking the place of a traditional direct-acting mechanical bucket, the mechanism is operated by a system of three camshaft lobes per valve, which are packaged in the space previously occupied by only the one. The outer camshafts operate the outer annulus while the inner one, which has to be much lower in lift, acts on the inner central core.

This inner core and its outer annulus can move independently of each other until positively locked together by a spring-loaded plunger or locking pin operating under engine oil pressure. When the outer annulus and inner core are locked together, the outer cam profile is transferred to the valve motion. When unlocked, the central but lower lift cam will control the valve.

Since the oil pressure needs to be switched on and off to the tappet, two separate feeds are needed - one supplying the switching oil pressure (at about 1-1.2 bar pressure), the other supplying the lubrication. In addition to the normal valve spring(s) controlling the valve, the tappet has contained within it another, much lighter spring, ensuring that the outer tappet does not lose contact with the outer cams when the system is unlocked.

A relatively simple and surprisingly reliable component, often no bigger than the diameter of the standard tappet it replaces, in its unlocked mode the central tappet runs against the base circle of the inner cam, much as in a traditional hydraulic tappet. However, in the fully locked mode, the central tappet still runs on the base circle of the central cam until such time as the clearance between the base circle of the outer cam and that of the outer tappet is taken up. At this point the outer cam takes over to give the full valve lift, producing a sort of part-hydraulic, part-mechanical arrangement.

Generally running at much lower lifts - about 3-4 mm as opposed to, say, 9 or 10 mm of the main, outer cams - valves can be operated either in pairs or individually to control either tumble swirl or axial swirl depending on the in-cylinder air motion required. Furthermore, the inner cam can be advanced or retarded (using a variable cam-phasing mechanism) to enhance combustion stability, improve low-speed emissions or give better driveability around town.

A simple system in concept, switching tappets can be applied to many existing overhead-cam engines. Requiring few additional modifications to the cylinder head, correctly engineered they can produce high-performance engines with little in the way of compromise to everyday road use.

Fig. 1 - The switching tappet

Written by John Coxon

In the search for greater efficiency, compression ratios have increased in recent times, resulting in much smaller, more compact combustion chambers. To avoid mechanical mayhem and yet still flow ever-higher amounts of intake air, valve lifts have therefore become higher over a shorter period of time. This would seem to be an unalterable fact of life and applies equally to production road-going engines as their even more highly refined race engine brothers.

Helped by more reliable knock sensor technology, the knock-safe compression ratios of 8:1 in two-valve production engines of earlier years have now morphed into the knock-sensitive 10:1 (or even greater) compression ratios of modern four-valve units. But when it comes to extracting even more power from these units some enthusiasts still seem to think in terms of changing only the camshaft.

A trick used by all engine designers to boost performance at high engine speeds is to design in a certain amount of 'valve overlap' - that period in the engine cycle when both exhaust and inlet valves are open at the same time. To the casual observer this might sound rather contradictory - both filling and emptying the cylinder at the same time - but by carefully sequencing events the extraction effect of the pulsating exhaust can draw into the cylinder a greater amount of fresh charge. The greater the valve overlap the greater the potential for improved performance at higher engine speeds, but since the piston crown is rapidly approaching its maximum height at top dead centre the ever-present threat of mechanical contact between piston and valves has to be taken into account.

In production engines where the combustion chamber will have been finely honed for optimum performance and emissions, unlike earlier times, the valves will come very close to the piston towards the top of its travel. Cut-outs in the piston crown, if they exist at all, will be very small since they can harbour unwanted and unburned hydrocarbons.

Any significant change in the camshaft lift to gain extra airflow will therefore require larger pockets, which will in turn reduce the engine compression ratio. Increasing the duration of the cam will also almost certainly risk the whole profile needing to be advanced, in the case of the exhaust cam, or retarded in that of the inlet. In both cases, such a move is likely to create very little in the way of increased top-end power at the sacrifice of that much lower down in the engine speed range.

Perhaps more a testament to the skills of the power unit engineer over the years, the upgrading of a production engine by replacing the camshaft alone, other than in race formulae where it is the only option, is likely to disappoint.

Fig. 1 - Valve-to-piston crown clearance

Written by John Coxon

To recap, a hydraulic tappet for an overhead cam engine generally consists of an inverted bucket follower body incorporating a plunger, which is guided by a sleeve. Inside the plunger there's a loose spring, a ball check-valve or plate and a thrust pin, which also slides within the plunger.

Oil under pressure from the main engine oil gallery finds its way into the chamber and expands the assembly such that the traditional valve lash clearance normally seen in purely mechanical systems is taken up, and the inverted bucket runs against the cam profile for the whole of its rotation. Open to oil gallery pressure when the bucket is running along the base circle of the cam, when the bucket begins to move the oil inside the chamber is sealed off from its supply and, being almost incompressible, transfers its motion through to the engine poppet valve and spring assembly.

While all this might sound fine, hydraulic tappets of this type have some inherently undesirable characteristics. First, and probably most obvious, they are much heavier than mechanical tappets. Second, and perhaps more significant, they suffer from what is known as 'leak-down' when the oil in the chamber, no matter how little, inevitably escapes.

Third, hydraulic tappets can have a tendency to 'pump up' when engine speeds approach that of 'valve float'. As this happens, the valve may not return back to its seat quite so positively and the engine may lose power momentarily. To combat this effect 'anti-pump' tappets have been devised that generally include a form of internal snap ring to limit the movement of the cup within the tappet and keep control at high speeds.

Given all these disadvantages however, there was nothing I could find that would make me choose hydraulic tappets over mechanical versions until I looked at a third type of hydraulic element more usually described as the variable duration type. In this design, rather than 'leak-down' the tappets are purposely designed to 'bleed down' during use. This carefully controlled leakage inside the tappet allows the mechanism inside the tappet to move slowly as the bucket is rising, the combined effect of which is to limit both the duration of valve lift and the lift itself.

The net effect at slow speed is to simulate a cam timed better to the demands of the engine at this speed with little or no valve overlap. As the speed of the engine increases, this 'bleed-down' effect - which takes time to occur - is reduced, and the cam lift and duration rises back to more normal levels.

Producing better throttle response and greater power at low engine speeds, the system operates in effect as a variable lift and duration camshaft. When engine speeds are limited (to about 7000 rpm) anti-pump technology can also be used.

In theory at least, and in formulae where valve lift and engine speeds are restricted, there could be an argument for using hydraulic tappets. Sounding a little noisy at low engine speeds when the tappet is effectively collapsing, in the right application and carefully calibrated, these might actually work.

As a purist, however, I much prefer the lightweight mechanical systems.

Fig. 1 - Typical hydraulic valve cross-section

Written by John Coxon

Take, for instance, the selection of valve sizes and their ultimate effect on the lift of the camshaft. The theory goes that, for a race engine, the intake and exhaust valves should be of a suitable diameter commensurate with that of the cylinder bore and hence displacement of the engine. Thus, for a given size of engine, the valve diameters and therefore valve spacing will be set at an early stage in the overall process.

Now, assuming we have opted for the comparative simplicity of a direct-acting overhead camshaft with parallel valves, the selection of the valve diameters has by default essentially dictated the diameter of the inverted direct-acting buckets most commonly used in such applications. So we have indirectly, either willingly or unwillingly, limited the maximum valve lift velocity available from the camshaft. For a given duration of opening, this will ultimately limit the maximum lift of the valve.

At this point, many might try to illustrate the situation by using the complex mathematical formulae involved in camshaft design; however, I prefer to imagine the side-on view of the camshaft as it rotates across the follower. When the valve lift velocity is low, the flank of the cam has more curvature, and the point of contact with the tappet moves slowly across its top face.

As the curvature of the cam becomes less pronounced, the cam becomes more 'aggressive' - or more 'intensive', as some prefer to call it - and therefore moves much faster across the bucket face for every degree of rotation. The maximum velocity achievable therefore depends solely on the diameter of the tappet.

When the designer is up against it and needs that last little bit of valve lift, the contact point can move out beyond the edge of the tappet, but this will invariably cause the edge of the bucket to 'dig in' or at least cause lubrication issues. Some cam designers surmount the problem by introducing a tiny amount of curvature to the bucket face. In the past this may have been done to encourage the valve to spin using a slight taper on the lobe to give eccentric contact, but these days truly flat tappets are the norm and valve spin is introduced by positioning the cam lobe centre line slightly eccentric to the centre line of the totally flat tappet.

When higher valve lifts are necessary and durability is an issue, however, this extra valve lift can be obtained using a more pronounced curvature of the tappet, which is pegged in the tappet bore and, although free to move up and down, cannot rotate. To even out valve seat wear, valve rotation can be encouraged in other ways, but the 'domed' tappet face ensures that the contact point with the cam will always stay within the tappet diameter.

Surprisingly reliable and a bit like a 'Get Out of Jail Free' card, the domed tappet can sometimes help designers get out of the rut.

Fig. 1 - The domed tappet

Written by John Coxon

Now there can't be a cam designer or engine builder who hasn't been confronted with this problem at times and

who hasn't wanted a higher lift profile than the one in his possession. The problem, as I am sure you have been quick to realise, is that in certain types of engines - particularly those of an older type, with the camshaft mounted within the cylinder block structure - installation has to be made through the end of the block. The cam bearings have to be carefully oiled and the whole camshaft judiciously pushed through successive bearings before being secured against some form of thrust plate.

The problem is that the circle prescribed by the rotation of the cam lobe has to be less than the diameter of the supporting bearings in order for the cam to pass through. While this is rarely an issue on many types of road-going older engines, when it comes to designing higher lift cams the options are somewhat limited.

On overhead cam engines we get around this problem by using separate cam upper bearing caps, and the constraints on the cam lobe design suffer one less restriction. But when the cam is mounted in the block the only option is to go for a smaller base circle, if possible, or extend the duration of lift to give a less pointy cam nose.

The former will inevitably increase the level of contact stresses while the latter may give far from ideal gas dynamics. Either way, there are compromises to the design, which can be removed after a little thought.

The solution, according to Mr Gustafson, is to manufacture the camshaft in two halves with a separate centre bearing. Manoeuvring each half, front or back into position via the sump and between bearings, central support is offered by a separate bearing inserted just prior to the installation of the second half. Using a keyway to locate all three radially, the system is locked using a single long bolt down the centre. A quick look at the accompanying pictures will explain all. Currently patent pending, the design allows much higher cam lifts than would otherwise be possible.

The design was prototyped on a vintage Model A Ford used for a Hot Rod, and from a standard valve lift of 7.67 mm, a further 5 mm could just about be introduced before other issues crept in. An innovative solution to a common problem on older engines, for higher powers the designer says that a more elegant solution, replacing the keyway with splines, could be incorporated.

It's a simple and neat solution to the problem, but now the task is to make sure the rest of the engine can cope with the increased performance.

Fig. 1 - Exploded diagram of the centrally split cam
Fig. 2 - Manoeuvring the rear half into position.

Written by John Coxon

Adhesive/abrasive wear tend to be lubricant-based issues. Poor lubrication or an incorrectly specified lubricant - it amounts to the same thing - will cause the oil film to break down under extreme loads, producing metal-to-metal contact causing abrasive wear (scuffing or scoring) or in severe cases, adhesive wear. Commonly avoided in competition engines by careful running-in using oils with high anti-wear additives, this subject has been discussed briefly elsewhere in RET-Monitor.

Corrosion, however, can cause other forms of camshaft surface failure. It's perhaps not as common as it once was, as high levels of base additive are now introduced to engine lubricants to 'mop up' much of the strong and weak acids produced as a result of combustion before they can harm the exposed surface of the cam.

If oils were not 'fully formulated' to contend with extended operation, the cam surface could become pitted very quickly, triggering scoring or abrasive wear as soon as the engine started running again. This sometimes makes corrosive wear very difficult to diagnose, but if engines are left lying around unused for weeks on end - as in the case of many weekend racers - other signs of pitting on the cam or follower might give the game away.

The most difficult form of surface failure to diagnose is that which results from surface fatigue. Thankfully very rare and perhaps seen only in engine development workshops, this form of surface failure is associated with the fatigue effect of what is more commonly known as the Hertzian stresses in a cam.

In a typical poppet valve, spring and cam arrangement, and whether the cam is rotating or not, the force of the compressed spring reacts through the tappet to produce a load on the cam surface. In any high-speed valve gear, it can be appreciated that the greatest load at high engine speeds occurs on the opening flank of the cam, while at low speed the highest load is to be found on the nose of the cam.

Classical analysis of the forces in and around the contact zone, (assuming the usual isotropic and elastic properties of the metal) shows that the maximum compressive stress (?z) is produced at the centre of the contact patch between the cam and its follower, perpendicular to the surface. Two other principal stresses (and using the conventional polar coordinate suffixes r and ?) ?r and ?? are also exist. At this point the failure criteria of the maximum shear stress as a result of the combined effect of these is comparatively small. As we travel down into the material of the cam in the direction of the z axis, however, these shear stresses build up to a maximum at a depth below that of the contact area.

These are called Hertzian stresses, and increase steadily to a maximum at what is therefore the weakest point in the material. Should they for any reason exceed the maximum allowable for the material, micro-cracks will inevitably result which, when subjected to repeated cyclic loading, will join together with other similar features and cause the surface material to break away, leaving a hole or pit in the surface of the cam. These stresses are very much a function of the contact loads at the cam-tappet contact point, as well as the instantaneous radius of curvature of the two components as they touch.

To the skilled eye, the pitting produced by failures of this type will be immediately obvious, as will the remedial actions necessary to correct the design.

Fig. 1 - Surface fatigue leading to cam-lobe pitting

Written by John Coxon

While it is true that the vast majority of camshaft wear takes place during cold start and warm-up, when the temperature of the oil is less than 60 C, the durability and performance of the valvetrain can still be a major consideration when the engine is operating at its normal working temperature. At the most basic level it's all about friction and its inevitable consequence - wear.

When two surfaces are running against each other and are not separated by the presence of a lubricant, friction and hence wear will result. The mechanisms of wear - abrasive, adhesive, fatigue and chemical - depend very much on the precise conditions at the mating surfaces at any particular time, but one method of minimising any form of wear it is to increase the surface hardness of at least one of the components. In the commercial world, this tends to be the most valuable, so when it comes to the camshaft and its mating tappet, the camshaft tends to receive most attention.

Thus for high-performance steel camshafts, virtually all of them will be surface hardened using one of the several nitriding methods available. The nitriding process infuses nitrogen in the form of metal nitrides into the surface layer, producing a ceramic that can be up to twice as hard as the core material.

In mass production, where unit cost is important, steel cams may have a surface finish of about Ra 0.4 ?m. Taking a little more time and using a finer grade wheel will bring this surface finish down to about Ra 0.1 ?m or even slightly less.

For most applications, this will suffice. But increasingly, for competition and where valvetrain friction needs to be reduced still further, customers are specifying a superfinished surface.

Usually consisting of a three-stage process, the finish-machined cam is placed in a vibrating container of ceramic chippings, into which a grinding paste is added. For the first stage the paste will be relatively coarse, but after a number of hours it will be substituted by a much finer product and left for a further period. The exact grades of the grit used and precise timings used by cam manufacturers tend to be closely guarded secrets, and to give the cams a degree of corrosion resistance an anti-rust compound is introduced towards the end. When complete, the surface finish of the cam will be close to Ra 0.01.

To give maximum hardness of the mating surfaces, and therefore minimum friction, once they are superfinished the mating surfaces can be coated with diamond-like carbon (DLC). A generic term for the family of coatings consisting of sp2- and sp3-bonded carbon atoms (part-way between graphite and diamond), DLC coatings share the properties of graphite (a solid lubricant) with those of diamond (hardness).

When bonded to a metal surface the material can give a significant reduction in friction compared to a similar phosphate-coated arrangement. (see Fig. 1). This reduction in friction was demonstrated recently to me when the cam of a 20-valve Audi cylinder head, coated with DLC was rotated by hand with ease. Using the same method the original phosphate-coated cam could not be budged.

It must be recognised, however, that the torque required to turn any camshaft, falls considerably once it begins to turn. This is because much of the boundary-layer friction quickly dissolves into mixed and hydrodynamic lubrication as soon as it is rotating. So while this static test may seem to be impressive, it somewhat disguises the fact that the overall reduction may be only 0.2 Nm or so for each cam when the engine is running.

While accepting that any reduction in friction is of value, the major advantage to running DLC seems more likely to be one of improved durability. Although coating both cams and tappets may give the lowest friction of all, much of the benefit can still be obtained using a superfinished cam running against DLC tappets.

Fig. 1 - Cam motoring data

Written by John Coxon

With camshaft lobes needing to be ground to accuracies of +/- 6 microns or better, the backlash in the machine is therefore of critical importance. The result of loose or worn components in the cross-slides - bolts, ball screws, end supports and keyways, to name but a few - a typical cross-slide leadscrew might have anything between 0.003 and 0.005 in (0.076-0.127 mm) of axial play or backlash in the mechanism. And without any measures to counter it, the effect can be alarming.

In explanation, as the leadscrew is rotated, one side of the screw will push the saddle nut (which is attached to the machine saddle) across the machine bed. At the completion of the stroke, as the direction of rotation reverses the leadscrew must rotate through a finite angle before the opposite side of the screw can begin to move the saddle in its reverse direction.

The degree of this backlash is therefore down to the fit of the leadscrew inside the saddle nut. This is true for any machine tool incorporating leadscrew technology moving the tool post or grinding head in or out against a rotating workpiece.

Methods to minimise this leadscrew backlash have been used for many years. Methods include the use of an additional anti-backlash nut on the same screw. This abuts against the opposite side of the thread to that used to move the saddle. Effectively jamming the leadscrew between two nuts, this is highly effective at minimising backlash but creates substantial and unwanted friction in the mechanism.

The approach most commonly used in all but the very latest in CNC machines, however, is the ballscrew. Replacing the simple lead screw profile with one rolled or finely ground to accept ball bearings running between it and a similar thread form in the saddle nut almost totally eliminates this backlash, which is further reduced using a spring-loading mechanism. This mechanism is still used in many of the best cam grinding machines but with the high forces involved it is susceptible to wear and will need checking regularly.

In the latest CNC machines, the little backlash that exists can be accommodated in software by adding or subtracting a fixed amount depending on the direction of motion. Cams ground using this type of machine should easily achieve accuracies of +/- 6 microns or less if the machine is kept in tip-top condition with regular servicing and provided material removal rates are low.

In the world of volume camshaft production though, reduced cycle times inevitably require high material removal rates, much larger machining loads and the greater levels of friction in the guide ways that accompany it. In addition to backlash we therefore have other errors creeping in - errors such as ballscrew windup and an effect referred to as 'hysteresis'.

Hysteresis is a function of the rigidity of the machine and is much harder to solve than simple backlash. Caused by the slight distortions in the machine's structure under load and how these relate to the workpiece being machined, hysteresis is therefore minimised during the design stage, and all modern machines are designed now to much higher standards to reflect the increased loading they are expected to experience.

To eliminate backlash in modern CNC machines, positioning feedback devices are used with either rotary or linear encoders, often called scales. While the rotary encoder is still prone to backlash (compensated in software) the linear encoder is effectively blind to it, and the only machining errors are as a result of hysteresis in the machine.

The most accurate cam grinders nowadays would appear to be those using linear motors rather than ballscrews, and have hydrostatic slideways with little or no friction and therefore little wear. But older, ballscrew-type machines can still be very accurate if lower feed rates and careful programming to avoid backlash-inducing fore-aft movements are used. As one low-volume cam supplier said to me, "An accuracy of 6 microns is perfectly satisfactory for the cams we make. To go to the 3 microns of a new machine we would need to sell an awful lot more products, and it's not technically needed."

Fig. 1 - Backlash

Written by John Coxon

The idea comes from Danny Williams, of New South Wales, Australia, who has applied some ingenuity and a lot of hard work to come up with a variable duration camshaft that maintains the opening and closing characteristics - velocity, acceleration and jerk - of a traditional engine camshaft, but which changes the phasing at which these occur.

In essence, he has taken an opening flank and a closing flank and joined them together with a 'bridge' that keeps the valve open for a variable period of time. Calling it a 'helical' cam, he not only designed it but built one to test in a Suzuki 250 GSX, in which it has run on the bench only, for about 15 hours.

Now I have to say that it is a lot easier to play with the device than describe its action, but described by Williams as a 'coaxial shaft, combined profile cam', the unit consists of two shafts, one inside another.

The outer shaft carries with it the normal cam opening profile at one end which, as we travel along its length in the axial direction, blends into a lobe of constant radius. The inner shaft moves in and out co-axially and has the closing flank and ramps attached to it (see Figs. 1 and 2).

As the two shafts move relative to each other at one end of the cam, the profile is that of profile A in the drawing here (Fig. 3), while at the other end profile C is followed. At one end the duration is about 240 crank degrees; at the other about 320.

The interesting point about this cam is that, as the inner shaft moves and rotates, the period of opening - the duration - is continuous and directly proportional to the position along the axial width of the main cam. Suitable only for a finger-follower system, in this particular example the drive comes in through the centre from a roller chain and sprocket assembly.

In theory, the valve dynamics are little changed from the standard GSX profile - apart from the extra accelerations and jerks caused by the constant lift portion of the curve - but the main concern would appear to be the apparent complexity of the assembly process and the accuracy of the fit of the sliding parts required during manufacture.

The angle of helix in this example is about 30º, and obviously the wider the main cam or the shorter the change in duration then the less this angle would be. But in trying to make this as compact as possible, the risk must surely be in the mechanism jamming. Williams, however, assures me that this is not the case in practice.

It's a refreshingly simple approach and one that would be good to evaluate on a dyno, despite the difficulties and costs associated with manufacturing it. But what do you think?

Fig. 1 - Closing flank at a mid-position slides along between the end of the nose and the beginning of the base circle
Fig. 2 - Another view
Fig. 3 - Cam profiles(s)

Written by John Coxon

But of course, the 3L has been around for the best part of 20 years. And over that period the company, now called Cinetec-Landis, has progressively updated the specification. But no matter how good the machine is, and machines are regularly returned for updating and repair, the state-of-the-art has moved on - to the Landis LT1.

Smaller, more thermally stable with higher accuracy is what drives all machine technology and cam lobe grinders are no different from any other. Higher accuracy means frictionless hydrostatic slideways and linear motors for manipulating the grinding wheels rather than the older type ball screw. Driving the machines moving elements using a magnetic field with positional feedback from linear encoders or lasers, means no contact and therefore no friction but even more importantly no wear! With no wear, backlash is eliminated and so the machine stays more accurate for longer. Alongside all this, sturdier machines, capable of removing material faster with larger cuts, require stronger, more rigid, spindle systems. Thus the latest spindles are also hydrostatic in nature and have brushless motors assembled into them rather than using a separate motor and coupling. With virtually no friction in any of the moving parts the machines are therefore maintenance-free for the whole of their productive lives, which can make an important contribution to any of the payback calculations needed to justify the purchase in the first place.

Accuracy, speed and payback, but what about the quality of the machining?

Like the machine before it, the control software is configured to maintain a constant metal removal rate irrespective of the variables involved. Thus the grinding wheel speed stays constant, only increasing slowly to take account of the smaller diameter as the wheel wears. Using Cubic Boron Nitride wheels, this it does only very slowly. With a constant velocity of the surface at the point of grinding, to ensure a constant metal removal rate the rotating workpiece has to speed up and slow down as it revolves. As the instantaneous radius of curvature on the cam lobe becomes greater, so does the grinding 'footprint.' To keep the metal removal rate constant therefore the work piece has to slow down. And since to remove about 6 mm off the surface of a cast iron lobe and finish grinding to size takes about 8 seconds, this all happens in the blink of an eye.

Cam grinding has come a long way from the days of the profile grinder and master cam and the LT1 may be better on paper at least, but an icon always remains an icon.

Fig. 1 - Landis 3L - an icon?
Fig. 2 - Replacement LT1
Fig. 3 - Cam grinding

Written by John Coxon

To understand cam friction we need to appreciate the basics of general lubrication and remember our old friend - the Stribeck curve. Plotting frictional force against relative velocity for two parallel plates separated only slightly by a fluid, there are three phases. Phase 1 where relative velocity is low the friction is high. Referred to as boundary lubrication, the asperities, which form the peaks and valleys of the magnified surface effectively engage and lock together resisting movement. As the relative velocity begins to increase, the friction falls away sharply until such point as it reaches a minimum. This phase is often referred to as 'mixed ' lubrication. Once the minimum has been reached friction starts building up again, as the role of the viscosity in the fluid begins to take effect. This last zone is often referred to as hydrodynamic lubrication.

Unlike the piston ring pack and bearings, the regime of lubrication in the camshaft, depending on its speed, is generally reckoned to be a mixture of boundary and mixed for much of the time. Excursions into the hydrodynamic zone being brief, the entrainment velocity of the oil during much of the time is relatively low and the lubrication regime spends much of its time running up and down that part of the curve where friction is significant.

Figure 1 shows the oil film thickness from a typical flat tappet arrangement. Along the back of the cam the oil film thickness is moderate and friction is low. Along the flanks entrainment is high, the oil film builds up and friction is also low. However, towards the nose of the cam the entrainment velocity falls, in this case very close to zero, the asperities lock and friction is high. The resulting frictional torque of this flat tappet arrangement when plotted the cam angle is therefore showed in Figure 2. Altering things like the cam profile, increasing the spring loads and reducing valve train mass as well as improving the surface smoothness of both cam and tappet (using processes such as superfinishing or DLC), will change the onset of boundary lubrication and altar the slope of the mixed regime that follows. However unlike the piston ring pack or the bearings, reducing the viscosity of the oil is most likely to increase the friction in the system through higher boundary forces, which do not offset the lower hydrodynamic effects.

Friction modifiers can of course be added but that moves us into an altogether different discussion.

Fig. 1 - Oil film thickness and cam friction torque schematic.

Written by John Coxon

Thus in the traditional suck-squeeze-bang-blow cycle, the intake valve has to start moving long before top dead centre and will not be fully closed again until well after bottom dead centre. Likewise the exhaust will open well before bottom dead centre and be fully closed again after top dead centre. Often expressed as something like Inlet - 36/64 and Exhaust - 76/24, the numbers refer to the timing in degrees in the case of the inlet valve as - opening 36 degrees before top-dead-centre and closing 64 degrees after bottom-dead-centre. For the exhaust valve in this example the timing would be opening 76 degrees before bottom-dead-centre and closing 24 degrees after top-dead-centre. The duration of the intake valve opening therefore will be 36+180+64 = 280 degrees as indeed is that of the exhaust (76+180+24).

But many people think that the longer the duration, the more powerful the engine. While undoubtedly in many instances this might be the case, it doesn't necessarily always follow. It all depends on the timing of these events relative to the piston position. And the most important of these is the closing of the intake valve.

Having reached full lift when the piston is something like just over half way down the bore, the intake valve will begin to close again. At the bottom-dead-centre when the piston has come to a temporary halt, the valve will still be almost half open (see diagram). At this point when the piston is starting to move back up one would expect the charge to start flowing back through the open intake valve. However, near to bottom dead centre the piston is moving very slowly and the inertia of the incoming charge generated from the pressure pulsations in the intake port will continue to fill the cylinder. When the pulse inertia of the incoming charge is brought to a halt by the velocity of the piston opposing it and preferably when there is also a positive pressure wave just coming through, the valve should be closed trapping all the incoming charge. The higher the engine speed, in general the greater this pulsation effect. Early intake valve closing will therefore make the engine run better at low speed and conversely the later the closing, the better the engine will run at high speed - but like anything in the engine world, only up to a point!

Finding this optimum position is what engine 'tuning' is all about and while computational methods can be a great help, the final solution is best left to the engine dynamometer.

Fig. 1 - Typical intake and exhaust valve motion with intake valve almost half open at Bottom-Dead-Centre (BDC).

Written by John Coxon

In selecting any new performance camshaft the starting point must surely be the catalogue. For within, alongside your particular make and model of vehicle, will be an array of other numbers under the headings of timing – intake and exhaust, and the subject I would like to touch on here today - valve lift.

The choice of valve lift is pretty straight forward – or is it? While it is definitely true that for a given inlet and exhaust timing a higher valve lift should give more airflow, it doesn’t always necessarily mean that a higher lift will increase engine performance. It’s all tied up with the speed of sound. Let me explain:

The piston travelling in the cylinder bore on its intake stroke starts at zero velocity at the top (top dead centre) and finishes at zero velocity at the bottom (BDC). In between, it increases up to a maximum at its mid-way, 90° mark and slows down again to zero at the bottom. During the first part of the intake stroke the cam will be opening the valve allowing air to flow until at its maximum lift (represented by point ‘A’ in the diagram) and close to that of maximum piston velocity, the flow into the cylinder will be at or near its peak. If we replace the camshaft with one of higher lift (represented by point ‘B’ in the diagram) then it doesn’t matter how much more valve lift we have, no more air will flow and the effective restriction will be in the port and not in the position of the valve. More usually, this restriction is immediately behind the valve and known as the ‘throat’. Here designers try to accelerate the airflow and guide it between the valve and its seat, prior to its discharge into the cylinder.

At low engine speeds and in an unthrottled engine therefore, the velocity of the piston governs the airflow. But at high engine speeds, when the piston is travelling down much faster, the air flowing through this throat will approach that of the speed of sound at the particular temperature and pressure prevailing. Although the piston might wish to draw in air at a higher speed, this throat will be become what is known as ‘choked’ and no further increase in the mass of air flowing will take place. As a result no improvement in engine power output will be demonstrated either. An inappropriately high cam lift will consequently not improve performance but certainly will introduce unnecessary stresses and strains into the valve train.

The moral of the story is therefore, if in doubt, always get your cylinder head airflow tested by a reputable organisation and take advice from the cam supplier. After all it is in his interest to make sure that you go away a happy customer.

Written by John Coxon.

Traditional reciprocating engine practice is to open the exhaust valve well before piston bottom dead centre. At this position most of the expansive power of the combustion on the piston crown has been spent and opening the valve at this time enables rapid evacuation of the cylinder. Also, opening the valve sufficiently early enables the exhaust gas to be swept out of the cylinder without creating too much back-pressure during the piston upward stroke and minimises pumping losses. Likewise on the next engine revolution, and with the exhaust valve closed, the inlet valve only closes well after bottom dead when attempts are made to harness the inertia of the intake air and trap as much of it in the cylinder as possible. However, it is around top dead centre on the non-firing stroke when things can get pretty hair-raising and the opportunity for a disaster begins!

First we have the intake valve at the point of opening. Beginning just before top dead centre the valve will be off its seat and moving ready to admit the intake charge in response to the piston downward motion. Somewhere shortly afterwards and slightly after top dead centre the exhaust valve will be coming to rest back on its seat. The time between the intake opening and exhaust valve closing is referred to the world over as the valve overlap, and a number of interesting possibilities could occur because amid all this lurks the piston crown. In a modern wide bore, short stroke engine, to produce a compact combustion chamber with a high compression ratio and with minimal surface to volume ratio, requires a high compression height to the piston at top dead centre such that the piston motion just misses the exhaust valve on its upward stroke and is almost collected by the intake valve on the downward stroke. It is clear therefore that once the valve-piston geometry is designed, even with small cut-outs in the piston crown, only slight adjustment of the valve timing may be possible. In particular, the higher the speed for which engine performance is optimised, the later the exhaust valve closure. Retarding the cam in this way risks the exhaust valve touching the piston on its closing leg. Likewise advancing the inlet cam chasing an even better torque characteristic, risks attacking the piston on its downward stroke.

As a double check when building prototype engines it was therefore always considered best to fit very light valve springs on the initial trial build so that around the piston top-dead-centre non-firing stroke, the valves could be pushed all the way down to contact the piston to determine the actual clearance. Although the minimum running clearance when hot will be slightly different, at least you will be able to build up an idea of where the limits to the valve timing reside.

In designing any camshaft profile therefore, due consideration has to be made not only to the cam duration required and the valve-to-combustion chamber geometry, but also to the practical aspects of the piston-to-valve clearance available.

Written by John Coxon.

Although there may be a touch of dewy-eyed nostalgia in all this, at the time camshaft designers had a hard time of it. Given unlimited valve sizes and lift, the skill was to manage the airflow through the engine chasing the maximum airflow past the valve. Since this was proportional to the area under the lift curve, designers would invariably go for maximum lifts in order to keep the cam opening periods short. Short duration cams generally ensure better engine drivability and are as a rule preferable to one with a lower lift but higher duration.

Ten years ago the limiting feature on those 1999 spec engines was the size of the tappet bucket. High lifts and short durations invariably mean high cam velocities which in turn when using flat tappets give high eccentricities and the very danger that the nose of the cam will fall off the edge of the tappet. The solution to the dilemma was to limit the maximum velocity effectively flattening it off (see diagram) but this was only achieved at the expense of increased positive acceleration and increased mechanical loading in the valve train. Fortunately limiting the engine speed to 8500 rpm did help.

When the rules changed in 2000, restricting the cam lift to 12 mm and then 11 mm in 2007, life became altogether easier for the cam designer if probably not quite so for the drivers involved. Although the lift had been restricted, so too in effect had the cam velocity. Stipulating that the tappets would be free, their height and diameter however must be retained. Effectively fixing the diameter to the homologated figure, the rules limited the maximum eccentricity of the cam and hence the velocity. The11 mm lift in recent years has made life so much easier in that the cam is much less likely to reach such high velocities in the first place. Nevertheless, with such a limit greater emphasis is required on opening the valves quickly and keeping them open for a longer period of time. In this way the flow area under the lift curve will only marginally suffer but so too will be the drivability of the engine. Generally reckoned to be delivering something like 285+ bhp, development over the years was probably more down to the detail design of other components rather than just chasing airflow through the engine.

With the move to low boost turbocharged units with intake air restrictors for 2011, depending on the final detail regulations, it certainly looks as though life isn’t to get too onerous for the cam designer. With the air restricted in such a manner there would seem to be little point in going for high lift aggressive cams.

Cam designers should therefore rest easy and look for more challenging opportunities elsewhere.

Written by John Coxon.

But the one area that has changed over the years, however subtle to the layman, is the material. Although predominantly ferrous based, even in Formula One (as required by the current regulations), slight changes to the material specifications have been deemed necessary depending on things like the aggressiveness of the profile, the location of the part and in many cases, the cost economics at the time. Throughout time, camshafts have been made out of various types of iron. Early mass production, pre-WW2 examples were made from hardenable iron. A Grade 17 (now referred to in BS 1452 :1990 as a flake iron Grade 250 - 250 N/mm2 being it’s tensile strength ) cast iron, this 1% chrome iron had a much finer grain structure and graphite distribution than most cast irons of the day. Designed primarily for block-mounted camshafts where an abundance of splash lubrication from the sump was available, these were flame or induction hardened prior to final profile grinding. Not suitable for overhead cam applications because of the lack of adequate lubrication, this material attracted a poor reputation in early OHC designs. Replaced by a similar grade 250 cast iron but this time poured into steel moulds in the shape of the cam lobe, the chilling effect on the molten iron produces a very hard and difficult to machine ‘white’ iron surface. This ‘chilled’ cast iron is still very popular in production engines of all types today but can only be used when production volumes can justify the extra costs of making them.

When the additional costs of these chilled castings cannot be easily amortised, or where a more specialist application is envisaged, steel remains the only other realistic alternative. Plain medium carbon steels like EN8 and 9 when flame or induction hardened were entirely acceptable and certain types of alloy steel (EN351, EN34 and AISI 8620) have been frequently used when used against a chilled cast iron follower. However the most popular material of all and still used widely today in the UK is that of EN40B. Now referred to as BS 970 722M24 and a slightly tighter specification, this chrome-molybdenum steel when supplied in a fully hardened and tempered condition is highly wear resistant with excellent toughness. A nitriding steel (a process by which the surface layers of the material are enriched with nitrogen) using 3% molybdenum, its tensile strength is around 920 N/mm2.

When a slightly higher spec can be afforded the almost identical aircraft quality spec BS S106 is available with tensile strengths quoted from 930 to 1080 N/mm2. Although very similar to 722M24 if not quite identical, this material is claimed to have fewer inclusions and therefore is ultimately stronger. For higher performance still, S132 can be used. Referred to as a chrome-moly-vanadium steel, this offers a tensile strength of between 1320 – 1470 N/mm2 representing high core strength and excellent surface hardenability. And if for some reason you are not totally happy with that, then a specialist steel manufacturer will no doubt will be happy to supply you with some of their deep nitrided GKH steels for use when even higher fatigue loads are expected.

Cast, steel billet or gundrilled, outwardly the camshaft doesn’t appear to have changed very much at all. Inwardly however, very careful and subtle changes have taken place.

Written by John Coxon.

Today, with powerful user-friendly computer software, kinematic cam profile analysis may be readily undertaken using ‘one click’ template driven models, manipulating profile lift data or derivative curves and produce almost instant on-screen designs. With the motion originating from either the cam or the valve end of the mechanism, this is now usually described in either many-segmented polynomials or in the more recent piecewise Bezier curves giving complete control over lift, velocity, acceleration and jerk data satisfying all the necessary boundary conditions.

But it wasn’t always like this. Indeed long before the advent of the computer, cam design was very much in the hands of the specialist who had limited means to analyse their designs let alone fully understand the implications to the valve train.

Most cam design methods start from the acceleration curve and express this against crank or cam angle in graphical form. Acceleration in the valve train is a function of the contact forces - at first resisting the lack of motion of the valve on the opening flank and then again slowing it down on the closing flank. In between, the zone referred to as the ‘nose’ of the cam is the area where the spring is coming into play and although these contact stresses increase with camshaft speed, it is the instantaneous radius of curvature of the cam producing the Hertzian stresses in the material, that creates the limitation here. However the quality of any design is also to a certain extent how smoothly the curves from each section blend into each other reducing ‘jerk’ and its potentially harmful effects.

Early post World War 2 automotive cams were often based on so-called, ‘3 Arc’ designs. Relatively easy to draw and specify for manufacture, acceleration rates rose instantaneously from zero to a high value at the start of lift and then dropped away even more suddenly from positive to negative at the junction of the flank to the nose. Equivalent to a hammer blow, these high ‘jerk’ designs were only tolerated because of the low stiffness inherent in the valve train and the slow speed of the engines at the time. The two sine wave profile came next. Consisting of a positive sine wave acceleration curve followed by another, longer one of lower ‘negative lift’, designs of this type produced large cam lobes for a given lift, which restricted its use but did reduce spring surge. Further developments, introducing the idea of multi-sine waves (up to 6) or ‘harmonics’, enabled more suitable lift curves to be produced enabling spring forces to be reduced and better control of the jerk between the flank and nose.

Over the years these trigonometric designs made way for those relying on polynomial equations. Superior in many ways, they provided smoother action to the valves, producing less vibration, wear and noise, were easier to manufacture and required reduced torque to drive them. Satisfying the equation:

y = C0 + C1x + C2x2 + C3x3 + C4x5 + C5x5 + C6x6 + C7x7 + ……

and inserting the boundary conditions as well as those in the three derivative curves of velocity (dy/dx), acceleration (d2y/dx2) and jerk (d3y/dx3) and then substituting the values back in, eventually produced a series of polynomials for further analysis. At a time when electronic calculators speeded up the process only slowly and trigonometrical tables were de rigour it might take as much as a whole week or more of successive iterations and diligent endeavour to produce a profile at even one-degree intervals. Generally limited to the 6th power of x, by time alone, cam design was a long and time-consuming exercise.

That the overall engine design process has become more complicated, there is no doubt. But with modern software and the ability to produce highly efficient cam profiles with minimum jerk (maximum smoothness) between sections, the process of designing the cam profile itself is no longer the domain of the specialist.

Written by John Coxon.

alter that micro-topography for the past five years, initially focussing upon the cam/finger follower interface before investigating other potential applications. During that period the French engineering group successfully developed surface micro-texturing, which enhanced friction and wear characteristics.

The basic concept of micro-texturing is in effect to pit the surface in a carefully controlled manner. This is done by exploiting laser technology in such a way that surface flatness is not affected and there is no danger of creating stress raisers or cracks. The laser removes material from the surface, creating tiny dots or grooves, according to the specific requirement. Mourier says that, by way of example a dot might be 10 to 200 micron in diameter with a depth of up to 100 microns, but that this surface manipulation has no specific geometry – it is always application specific.

Mourier emphasis that the advantage that can be gained through surface micro-texturing is not due to enhanced oil retention, in the way that a piston ring working face might be micro-pitted to retain oil, for example. “Our surface micro-texturing causes a change in the way the oil flows – we manipulate the oil flow between two surfaces. This creates a lifting effect that actually parts the two components.”

Mourier remarks that if a cam is experimentally run against a tappet with increasing force, there will inevitably come a point at which the oil film breaks down. However, he says that the breakdown of the film can be delayed using the correct surface micro-texturing. “We can modify the characteristics of the lubricant film so that it withstands higher load”.

By the same token, Mourier says that HEF has obtained a measured reduction in friction through micro-texturing the working surface of a piston ring, with no change made to the cylinder bore surface. Mourier warns, however, that if not properly optimised for the specific application, surface micro-texturing can actually spoil rather than enhance performance. “Done correctly it will optimise the performance of a given contact area”.

HEF is currently concentrating on applying this technology to one of the two running companion surfaces but Mourier sees no reason why further gains should not be made from its application to both surfaces. While current applications have been successfully made at the cam/tappet and the ring/bore and the piston pin/con rod interface, other likely gains are to found at the piston skirt/bore and the crankshaft journal/bearing interface. Research work is continuing in many areas of the race engine.

Assisting this effort, Mourier added that the equipment has now been installed by HEF to apply very carefully controlled micro-texturing to a wide range of surfaces, coated and uncoated. The laser-based tool can apply specific micro-texturing to any surface that its beam can ‘see’.

Currently micro-texturing is being applied to uncoated steel parts but aluminium and other materials may follow. It is also being applied to PVD-applied coatings including titanium nitride, chromium-nitride and a range of Diamond Like Carbon (DLC) coating formulations. In the case of thin film advanced coatings, either the coated surface can be micro-textured or the substrate can be treated prior to coating, the optimum approach taken depends upon the component in question and the appropriate micro-texturing form.

Applied correctly, surface micro-texturing has been found to be more effective on a DLC coated than an untreated DLC surface. “Surface micro-texturing is complementary to DLC”, says Mourier. “Applied correctly, it is an advantage over untreated DLC”.

Written by Ian Bamsey.

have little or no inertia and hence the forces involved in the valve train are negligible. Those of us in the real world know that this isn’t the case and that somewhat inconveniently, the forces involved in opening and closing these valves can be substantial. To compensate therefore the process of lifting the valve starts some time well before Top Dead Centre and as such will not be back in place until well after Bottom Dead Centre. Thus the valve has to be lifted off its seat, fully opened and then closed again placing it gently back on its seat smoothly and without introducing any undesirable motion, all in the space of a fraction of a second.

In order to fulfil this function, cam designers split their cam profile design into a number of separate regions, which can be characterised in terms of lift, velocity, acceleration and rate of acceleration (or deceleration – negative acceleration) referred to as jerk. Each region has to seamlessly blend into the next to ensure that no unnecessary forces are introduced that could generate excessive contact stresses or overcome the spring force ensuring the precise control of valve train components and avoiding separation under all running conditions. While older cams may have been designed using long and tedious techniques involving hand calculated, trigonometrical methods, modern cam computer design packages use the theory of spline functions giving almost infinite flexibility in the shape of the acceleration curves that can be produced.

To understand some of the issues faced by the designer the cam profile can be split up into a number of parts or zones each with it separate function. The first of these are the ramps – both opening and closing. The purpose of the opening ramp is to enable the valve gear to be lifted up off the base circle at a known velocity. Mechanical tappets need to have some running clearance to ensure the valves are fully seated for much of the engine cycle and the function of the ramp is to take up this clearance with minimal impulsive force. Likewise the closing ramp which in many cases is longer, needs to drop the valve carefully back on its seat and is particularly important for the exhaust valve, to preserve a critical level of durability at the higher temperatures involved. Most competition cams tend to use constant velocity ramps and when using mechanical tappets the height of this ramp dictates the valve lash setting.

Once the valve has been picked up by the ramp it moves quickly into the opening flank. Arguably the most important part of the design, this aspect controls the acceleration and eventual velocity of the valve train, which then has to be slowed down by the action of the valve spring if separation is to be avoided approaching peak lift, resulting in valve ‘float’. In general the faster the valve is opened for a given valve opening duration the more air will flow. When the valve reaches full lift it will stop moving prior to dropping backwards towards the seat. This period, referred to as the ‘Dwell’, needs to incorporate as large a nose radius as possible to minimise contact stresses when hydrodynamic lubrication is at a minimum. Once over the peak the valve spring continues its task of keeping control over the system until the valve is fully closed.

Although the principle of the camshaft is in itself very simple, the practicalities of implementing such a system are far from it.

Written by John Coxon.

These similarities continue and while we all have our favourite comedian, to compare them would, I am sure you will agree, be extremely difficult; the lack of suitable objective measurement perhaps being the first of the problems. Strangely enough though the same too could apply to camshafts. It is perhaps not the lack of data when comparing cams but the lack of comparative data, which is often the issue.Recently I had reason to compare a number of different camshafts. To do this task properly would naturally entail contacting all the various cam suppliers and request the details of their grind; principally lift above base circle at, say, 2 degree intervals. But as ever I doubt if manufacturers would want to give that information anyway, it’s late at night and I need the information more or less straight away. All I can find from the various sources is the published lift and the advertised duration. If I’m lucky the documentation might refer to the height above the base circle at which the duration is measured but more often than not, vital information such as this is left for me to guess.Now I know that when I actually come to purchase the cam, it will more than likely have been machined on a state-of-the-art Landis 3L grinding machine with a profile accuracy of 60 microns or so to an angular accuracy of 0.020 degrees but at this time of the night all that seems irrelevant when I would be happy with some comparative data to within a couple of degrees that I could rely on. When the information has been volunteered, quoted durations of ‘ramp-to-ramp’, from ‘lash to lash’, or at 1 mm from base circle have all been seen recently but since little or no significant flow takes place until at least 1 mm lift this would seem to be at least an ‘honest’ place to start.

From what I can see in the US, things seem to be slightly more organised where 0.050 inches from the base circle (1.27 mm) would seem more widely accepted and therefore comparisons perhaps more easily made. At this height the tappet will most definitely be on the opening or closing flank and therefore small changes will have a significant effect. The cam manufacturers totally understand the frustration when a 236 degree (at 0.050”) duration cam expands to 270 plus at ‘Ramp-to-Ramp’. But so long as Jo Public insists on ‘long’ duration cams that are still drivable on the public road, then little here is likely to change.In the mean time as our comics might agree, the laugh most definitely is on us.