How much is too much?
I don't know about you but I am continually amazed at the power of modern electronic components. Invariably integrated with some form of digital microprocessor, the speeds and in many cases capacity to store information is something truly to marvel.
Moore's Law, formulated nearly 50 years ago by Intel co-founder Gordon Moore, said at the time that the number of transistors that could be replaced by an integrated circuit would double every two years. Later, stating that this would go on only until 2015, according to the pundits he has so far been proved correct, such that processing speed and memory capacity have grown to the level when even the most elementary of children's toys now have had far more capability than the systems that sent the astronauts of the Apollo mission to the Moon.
The same is true for even the humblest of data acquisition systems, whether they're used to store engine, transmission or chassis data. But with a typical race engine how much data do we really need and, once logged, do we even have the time to trawl through it and fully understand it all?
In a typical engine, data loggers work by capturing and storing the streams of data that can either be subsequently downloaded or relayed to a display for the benefit of the driver. Instructed by the internal clock the microprocessor reads a voltage or frequency output from the various engine sensors and stores it in the memory in digital form.
In the case of a simple application these sensors may be measuring as inputs - engine speed, fuel, air and coolant temperatures as well as manifold pressure, barometric pressure and throttle position. Outputs such as injector pulse width and ignition timing will also be routinely logged. And even at this basic level, the output of the wideband lambda sensor measuring the air:fuel ratio and less obvious information such battery voltage may also be logged. So far, and unless I've miscounted, we are up to 11 sensors - and I haven't really started yet!
Monitoring individual injector pulse widths on a sequentially injected engine, and perhaps deriving the error between that and the figure intended, boosts the numbers still further - not forgetting to look at what is happening on the ignition timing side or any other engine diagnostics. The point I am trying to make is that once it becomes easy to produce the data, we just seem to want more.
Early aftermarket systems, when they even considered adding internal data logging, would boast 256 kB of memory - enough to record quite a useful amount of what I would call 'simple' and usable engine data, provided logging rates were kept sensible. This 'simple' logger today can offer up to 8 MB of data storage which, when logging eight channels at 250 samples per second, would give you 55 minutes of stored data. Running this logger at a more sensible 25 samples per second ups this to 503 minutes.
Having already proved that eight channels isn't enough, if we increase the logging to 16 channels - eight at about 3 Hz measuring things like temperatures that don't change very much, and eight at a still very useful 10 Hz - then the total logging time jumps to 1021 minutes, or almost 17 hours.
Now I don't know about you but by the time I'd trawled through 17 hours of 16 channels' worth of data, I think I might just have given up the will to live.
Fig. 1 - 256kB of memory
Written by John Coxon
Link to original article