Yamaha XJR engine - Legend Cars

[steph_tsf]

Hello Dear developers,

My understanding of MicroSquirt is that it is targeting the market for motorbikes … and Legend Cars also (racecars) but … after carefully examining all MicroSquirt schematics (the inductive pickup amplifier) and after carefully examining the C code (especially the Timer0_ISR), my conclusion is that you cannot a) use a hall effect detector on the crankshaft and b) if you stick to the inductive sensor, you won’t be able to detect the short-short-short-long Yamaha toothwheel pattern.

You can find some info about the Yamaha toothwheel here :

http://www.factorypro.com/products/rotors.html

The rotor has a “T” mark at the exact beginning of the long tooth. Which comes just in front of the inductive pickup centerline when piston 1 is at TDC.
And using a stroboscope, at 7500 RPM, you get the confirmation of this as your stroboscope is aligned with the “T” mark and inductive pickup centerline when you dial 33 degrees of advance.

For this particular wheel, that we are using on Legend Cars (racecars) powered by a Yamaha 1300cc XJR engine, there is an inductive pickup sending the signal to the ECU. But because of the very long tooth, we feel that it is needed to reliably detect a) the positive going signal and b) the negative going signal – otherwise, you run the risk of losing the “long tooth” specificity. Currently, the MicroSquirt inductive pickup interface is only able to detect the positive going pulse, isn’t ? You see the problem ? My first idea was then to use the Timer1 input that is normally dedicated to a camshaft sensor. There, with the 4K7 pull-up, it is possible to hook a hall detector, hence enabling a precise detection of the start of the tooth and the end of the tooth. But … I doubt the current software can cope with that.

It would be marvelous if you compile the info I am providing to you, and then posting “how to” use MicroSquirt on a Legend car (Yamaha 1300cc XJR) WITHOUT attempting changing the toothwheel AND WITHOUT attempting changing the variable reluctance crankshaft sensor.

Now, we have Murphy’s law. On the Legend Car ECU, we found that one wire of the sensor (green/white) is nearly grounded inside the ECU (green wire) , delivering only a very small signal that one never would use as main signal. So the only reliable signal is the one that comes from the other sensor wire (red/white) delivering a +10V positive half-sine signal to the ECU (red wire) when the tooth is hitting the sensor, and a -10V negative half-sine when the tooth is leaving the sensor. So with the current MicroSquirt hardware put in parallel on the ECU (read in a while why I will put it in parallel) , I think you will get equidistant pulses, that’s all. You won’t be able to see that particular long tooth. So you don’t actually know if your crankshaft is TDC, TDC+90, TDC+180, or TDC+270. Annoying, isn’t ?
So, IMO, what you need is to also monitor the precise occurrence of the high-amplitude negative pulse. This is not trivial as you don’t have a negative supply rail. What I would do is to use AC coupling (in order not to ruin the genuine ECU polarization) then bias the signal and compare it to a DC level.

What I have seen also is that you don’t use any form of hysteresis on the pickup signal detector. Is that on purpose ? Don’t you think a small degree of hysteresis (like 1V) might raise the signal/noise ratio and avoid most false triggers especially on “missing” or “long” teeth ? My experience is that if you improperly load the pickup sensor (with more than 33K resistance and some capacitance in the cable or at the input), at high RPM, you may get significant zero crossing overshoot on the long tooth, hence getting the risk of wrongly detecting a “negative” just after the occurrence of the “positive”. And the same, but opposite, when the long tooth is leaving the sensor. Do you follow me ? Remember we are dealing with an inductive pickup having to deal with long flat surfaces instead of short pulses (32 or 60 teeth) or very short pulses (triangular teeth).

I might become a beta tester starting from October 2007 if you find it useful. Because at the moment we are racing under stringent “unmodified cars” regulations … but that may change after this season – I want to promote the idea of a “hot” Legend Car class where engine management can be optimized at will.

Then, what I do visualize, is that MicroSquirt will become the “de facto standard”. Isn't this a fine initative ?