Audi 5-cylinder engines with triple trigger

[quattrodaveuk]

I'm searching for a 'scope... I can't believe in a office/lab full of electronic and software engineers (I'm the token mechanical engineer they keep in a cage in the corner) and no one has a scope I can borrow. Well obviously SOMEONE has the companies Fluke portable, but I doubt they're going to lend it to me

If you need a hand I have a scope, have a built working divider board (with a lot of input from Eric) and I'm not too far from you... Let me know how get on....

Dave

[mickeymarrows]

Thanks for the offer Dave!

[mickeymarrows]

Well I've got the 'scope but despite the lab manager giving me a 30 minute tutorial on how to use it I just cannot get a signal out of the vr sensor. I figured that all I needed to do was unplug the vr sensor, connect the probe to it, crank the engine and voila! But I was wrong.. I get nothing..

I've tried to connect the probe to either of the signal wires, with and without the earth connected..

[jsmcortina]

Do Audi 6 cylinders count here?

I have one of Jeans divider boards but never had any success, but probably because 135 can't be divided by 6, but 2 revolutions is 270, which divided by 6 is 45...

MS2 or MS3?

If you wanted to try to get it working on MS3, I could work with you on it.

James

[mickeymarrows]

MS2 extra. I just cannot afford to upgrade to MS3 due to just becoming a dad, in fact this Audi will probably have to be sold.. But only once its running right which I've given up trying to do with the factory Hitachi.. Hence my reattempt to get this tri tach working with MS2..

[quattrodaveuk]

I seem to remember when testing my analogue scope gave a nice picture @ .1v/division & 1ms/division. Both settings will have to altered as engine speed increases but I think .1v/division & 1ms/division was good for cranking & idling. If you have a digital scope then it should auto range.....

One other thing, if your testing the VR sensors I found it easier to pull them out and hold a cordless drill with an offset bar in front of the sensor, it allows you to run for far longer with out flattening the battery or damaging the starter motor..... I used a sheet of plywood and using some cable ties mounted the drill & sensor to it then using another cable tie on the drill trigger I was able to test, use the scope & use the laptop whilst having a clean steady signal.....

Dave

[mickeymarrows]

I know the VR sensor works OK as the engine runs "fine" on the Hitachi ECU, what I need to ascertain is the signal that that VR Sensor produces.. then I can work out how to swap the Hitachi for Megasquirt...

[mickeymarrows]

Should have bought a scope years ago..

This is all three signals, top is ring gear VR, next is the camshaft Hall sensor and most importantly the lowest is the crank VR.



Pretty self explanatory.. Although it looks as if there are two notches in the counterweight.
It explains why I couldn't get Jeans divider board to work..

[elaw]

Wow, that's a challenging trigger setup!

In addition to the crank VR weirdness, it looks like the ring-gear signal is very weak and noisy. You might want to check that.

It seems to me the only hope for getting that to work would be to 1) build a circuit to "and" the crank and cam signals (you can see one of the crank pulses is very nicely "bracketed" by the cam pulse) and 2) see if you could talk Jean into making a customized version of his divider code that works with one crank pulse every 270 ring-gear pulses instead of every 135. Assuming that modified code used the same division ratio as the current code, what you'd end up with is a signal that appears as if it's coming from a 90-1 tooth wheel at cam speed. To keep the Megasquirt code happy, it might be best if he made it look like a 90-2 tooth wheel instead (with a high-tooth-count wheel it's easier for the code to determine the missing-tooth interval if two teeth are missing rather than one).

[racingmini_mtl]

Changing the number of pulse would normally not be a big deal but the counter in the code is an 8-bit counter and that's limited to 255. Changing the counter to a 16-bit value would mean that there would be a need to check if the code can still cope with the max RPM because the CPU is an 8-bit CPU so working with 16-bit variables might be significant. Or it might be a non-issue.

And the single missing tooth for a cam speed signal would not be more difficult to detect than for the crank speed signal since the number of crank degrees is exactly the same in both cases. So the 90-1 signal should not be an issue.

And I agree that there is a need to do something about that ring-gear signal. I doubt you would get something usable out of that but it would be interesting to see what the signal out of the LM1815 looks like.

Jean

[mickeymarrows]

Based on the Hitachi schematic that Eric posted in page 1 I'll scope the output from the LM2903 tomorrow and post up the results. Same for the flywheel signal...

I'm sure if the Hitachi can read that flywheel signal I'm sure MS will be able to..

[elaw]

Well that's it then, you're just going to have to grind every other tooth off the ring gear!

Jean, what about modding the code to insert a "missing tooth" not only "n" teeth (1 or 2?) after the pin signal but also n+135 teeth after? Basically create a fake pin signal 135 teeth after the real one, so you always get one or the other within 135 teeth? Then the output signal would look like a 45-1 on the crank, just like with the old setup.

By the way on the quality of the ring-gear signal, it looks from the menu displayed on the 'scope that it can present either a 1-megohm or 50-ohm load to the probe? If you had it set to 50 ohms it would explain the low signal level - those VR sensors have an impedance of about 1K and won't drive a 50-ohm load well at all. That's assuming you're using a x1 probe, but the 50-ohm load would also mess things up if you were using a x10 probe due to the resistance in the probe itself.

[racingmini_mtl]

I guess it would be possible to reset the counter after 135 teeth. I'd just have to make sure that doesn't screw up something else. And it would be hardcoded for that set up but I don't think it's a real constraint.

Jean

[mickeymarrows]

Its 5am here (feeding baby) so still too early for me to start the car, I'll need to wait a few hours

Good call on the probe Eric, I'll see what I can find. Jean, if you could add a second trigger to the code 135 pulses later that would be awesome..

[mickeymarrows]

Turns out I had a duff probe . The RPM signal is exactly as per the Audi manual



All I need now is for Jean to feel benevolent enough to write some code for me then I'll get on with buying a 4001 CMOS chip and building the circuit

[mickeymarrows]

I cannot find my divider board right now so I thought I'd feed the crank signal into a ms V3 vr circuit.. I don't think it liked it

[elaw]

Well, the number and timing of the crank pulses will still be an issue, but that 'scope display doesn't look right to me either. I'm wondering if either the trigger threshold is set too low, or the VR circuit is misadjusted?

I'm thinking in particular about 1-1/2 horizontal divisions from the left side of the display, and again at about 8-1/2 divisions, where the VR output appears to change states for no apparent reason. It looks like the trigger threshold is set too close to 0 volts and/or the hysteresis is too low, and it's triggering on noise.

[racingmini_mtl]

The problem with that VR signal is that neither the rising nor the falling transitions are consistently fast ones. So no matter which one you take, you will have inconsistent zero-crossings. You may be able to do something with the MS board VR circuit where you can adjust the threshold to something other than a zero-crossing but that won't work well with an LM1815 or a MAX9926.

Jean

[mickeymarrows]

I wouldn't take that trace too seriously chaps, I'm not convinced the circuit can cope with the high impedence VR sensors. As you can see they produce a 20v spike even at idle.. I just thought I'd post it up for interest..

The changes in state at 1.5 and 8.5 divisions aren't in fact random, they are exactly the same time after the edge of the counterweight. Adjusting the trimmers means I can remove the notches so that the trace only shows the counterweight, but its still a little confused

[jsmcortina]

The mainboard VR circuit likely isn't fast enough to cope with the 135 flywheel teeth.
(That's why the MS3 code expects the "reset pin" on that input and a circuit of your own construction to feed the high-speed signal into PT4.)

James