Vehicle: 1995 Mitsubishi FTO GPX Work Performed By: Owner Date: September 2001 on Article by: RichardH

Have you ever wondered whether an after market sports exhaust system really gives you a faster car? Or is it just the extra decibels giving the impression your car has turned into a speed machine? How about that pod filter recently slapped on in place of the factory induction? Did it provide torque gains across the entire rev range or not?

How about an even simpler question... Does a bottle of octane booster in your fuel tank really give you more power, or just leave a big hole in your pocket when all is said and done?

If you head off to a dyno every time you want an answer, not only does the cost add up, but you'll find the results tend to differ each time - even if nothing has changed on the car! Ambient temperature, relative humidity, the state of the dyno equipment being used, even the temperature of the fuel in the tank... It all makes for a tricky (and increasingly expensive) situation.

I decided to find an alternative.

I purchased a software package known as Home Dyno. It was capable of taking as input a recording of the engine activity in foot-to-the-floor acceleration through the rev range. When provided with other details about the vehicle in question, it produced a torque and power curve graph, along with raw numeric output.

But would it really work? And what was the best way to "record" the engine activity?

My first attempt was to follow the advice from the Home Dyno instructions and install an inductive sensor on one of the coil leads. This would result in a "spike" every time the coil fired. The FTO MIVEC engine has three coil packs, and uses a "wasted spark" system in which each coil pack services two spark plugs. The software was able to be configured to expect this. Unfortunately, the electrical interference was far to noisy to result in a clean series of voltage spikes. Many locations were tried, but none were clean enough. Attempts to tidy up the results in a sound file editor were also unfruitful. Another plan of attack was called for.

With all the engine management sensors in this car, there must surely be a suitable candidate for providing exactly the kind of output I was after? Out came the FTO Electrical Wiring Workshop Manual, and the sensors to the Engine ECU were reviewed.

DISCLAIMER TIME: If you feel the need to play with your own ECU wiring, make sure you read and understand the workshop manual. Double-check everything! You are doing so at your own risk! Now I have that off my chest, we can continue...

The crank angle sensor was just the thing! It would produce multiple spikes per engine revolution, and could be recorded directly, instead of using an inductive sensor. This would result in low noise levels and a clearly defined spike pattern in the sample.

The workshop manual identified the relevant wire into the engine ECU, located in the passenger footwell. I patched into this my own lead, ending in a 3.5mm mono socket. To ensure the output voltage was appropriate, I added a 10K resistor to both the positive and negative wires of the lead connecting to the socket. Without this added resistance, problems with engine tickover occur... If the ECU can't get its crank angle input, it will just refuse to run the engine! The 3.5mm socket was stowed neatly in the footwell recess, and that was that.

Now that the engine side of things had been taken care of, it was time to attend to the recording device.

I don't own a laptop, and was not likely to purchase one any time soon. If I did have one, it would very likely be a simple matter to connect a 3.5mm lead from the crank angle sensor directly to a microphone input on the laptop. As it was, I needed something to record while out on the road, in order that I might bring it back home and replay the output into my home machine's microphone input.

The Home Dyno software had an alternative package that included a digital sound recording unit. This sounded like a good idea, so I hunted through the odd electronics catalog here in Perth.

Sure enough, Jaycar sold an inexpensive digital recording & playback kit that was perfect for the job. The unit uses an NVRAM chip to record up to 30 seconds of low quality sound. By changing a resistor value, the sampling rate can be increased to record for less time, but with better quality. I experimented a bit, and settled on a 12-second duration - enough for a full-throttle run from 2000rpm up to 8000rpm in second gear. This kit was easy to build and modify, with good instructions and a clear PCB layout. Further details can be found at kitsrus.com (notably Data Sheet and Photo).

So now I had the patched ECU Crank Angle Sensor lead, and built a device capable of recording data while out on the road. Time for a test drive...

I found a flat, straight stretch of road not far from home that had no houses, connecting roads or other hazards... the perfect spot for some test runs in second gear. From a rolling start in second, I started up the digital recorder and gave the GPX full throttle. The car accelerated right through the rpm range, up to the point where the rev limiter engaged. Here is a raw sample from a typical run...

Unfortunately, this output was a little too good for the Home Dyno software! For every engine revolution, the crank angle sensor was triggered three times. At the higher rev ranges, the software was simply unable to get enough distinction between each spike. The result was an early termination when processing the file.

Seeing as my day job involves a good deal of coding, it was not diffucult to write a WAV file parser that took the recorded waveform, analysed it and produced a different format output...

The top waveform shows the spikes as recorded from the crank angle sensor, while the bottom waveform is the processed result. The WAV modifying tool discards five out of six spikes, leaving just one spike per 4-stroke cycle. I also gave it the ability to massage the wave shape. Instead of a positive and negative voltage wave with a complex shape, it leaves only a single, instantaneous spike.

After all this, just how well does the Home Dyno software work?

Here is one of the early runs...

As you can see, the software takes into consideration everything from current temperature to vehicle weight, gear ratios, transmission type and final drive ratio.

I did not worry about absolute values at the time (eg. peak torque & power). It was the shape of the torque curve I was interested in. The output was definitely a recognisable "MIVEC" shape, complete with the double peak and the dip at around 5,500rpm. This was enough to convince me that the project was on the right track.

The next phase was to calibrate the software to match real-world dyno results. This would give me confidence in the actual figures when comparing future tests, instead of just looking at the general graph shape.

Mark A and I took our vehicles in to the local dyno tuning shop. The resultant graph of power at the wheels gave me all the information I needed to calibrate the Home Dyno software to match...

Looking at the above graph, I had to move both the position and the inclination of the Home Dyno graph to match the "real world" example.

It was possible to move the graph vertically simply by adding or subtracting an amount from the Test Weight field. Entering a higher-than-actual value did the trick - 1650kg, to be exact.

A similar trick was possible when changing the angle of inclination. In this case, tweaking the Frictional Losses fields moved the top end up and down as required. For anyone interested in using this software with similar settings, I settled on Drag Coefficient of 0.10 and Frontal Area 10.00 Ft^2.

Because I was after "at-wheels" results, the Drivetrain Loss was set to 0.

The result was a good match to the real world graph...

So there you have it! The above graph was taken from a run up a quiet back road, using home-made hardware and some clever software.

This has a lot of potential. A "before" run can be undertaken, followed by any performance modification you care to name. An "after" test can then be done and the results directly compared.

As it has been calibrated with a real world dyno run, I feel the results are accurate, meaningful and a valuable tool for assessing anything performance related.

Now every time I see "power" bolt-ons advertised with promises of "higher torque across the entire rev range - guaranteed!", I know it's possible to verify those claims without spending a fortune at the dyno shop...

In fact, I can feel a test of some Octane Booster fuel additives coming on...

The following may be of assistance to other FTO owners. While I have endeavoured to ensure the accuracy of all information here, I take no responsibility for consequences of using it.

30 Second Voice Recorder
DIY kit from Jaycar. Cat. No. KD6080. Cost: Around $40.

Software
Home Dyno package.