DIY LTCC or similar system for LT1s

[vilefly]

nope, didn't miss a thing. It sounds like a game plan to me. I did wonder why they only synched using cylinders 1,4,6,7, but I think you may have answered that question. Resting position.
The TTL drive sounds like a winner to me, providing a current limiting transistor is used between the two, just in case. 15-20mA sounds safe to me.
I've been buying some Arduino Pro Mini controllers on ebay at about $5 each to tinker with. The very flexible compiler will let me use the same code on a different controller. Kinda nice, when building expendable projects where risk is high. Made some animated 8x8 led matrix eyes for my kid's halloween costume. Taught me a lot. Still, I borrowed heavily from the example code, but I am hooked.
The Pro Mini is the size of a medium postage stamp, and it just might be able to pull this trick off. Much more powerful than I expected.

[spfautsch]

I'm thinking a simple class d bjt amp powered by the ECM's coil trigger output for each coil. Or FETs if you want to get fancy (possibly higher parts count)? There are also some widely available stepper motor driver DIP ICs that may fit the bill.

My thinking is the uC employs a watchdog timer that triggers relearn. So we haven't seen an input change in a while, watchdog timer fires and says assume the engine isn't running anymore so set a bit in memory and go to sleep (power consumption can be an issue with the arduinos).

Let's assume that the stopping position is 20-30 degrees shy of tdc of a compression stroke. We're either going to be near 1, 4, 6, 7, or 8, 3, 5, 2. Let's also make a huge leap that this is not in a light gap in the wheel (logic low to the uC).

So the user cranks the engine.

The user engages the starter, and the uC picks up a rising edge interrupt that wakes it. We look at the "is it running" bit to see if we need to re-learn the position. We do, so we start counting microseconds until the next falling edge.

The falling edge signal happens and we store the microseconds in variable a.

Now we wait for the next rising edge signal. <thumbs twiddling>

Rising edge signal clicks. Now we count microseconds again until the next falling edge signal.

Falling edge signal happens. Store microseconds in variable b. Compare a to b.

if ( a > b ) {
  c = a / b;
  [ some code I haven't thought through completely]
  x = (next firing cylinder);
} else {
  c = b / a;
  [ some code I haven't thought through completely]
  x = (next firing cylinder);
}

Once the running bit is set the uC sequences ignition from memory based on the rising edge.

So you say what if the engine is parked in a light gap, and the first signal we see is a falling edge. Simple - set another bit that inverts the sequence detection logic [that I haven't thought out completely].

Easy peezy. I'll think about fleshing out the code on my drive to work tomorrow. There will need to be some fuzzy math allowing for variance in angular velocity during cranking. That's why they made the big gaps larger by multiples of 5.

[spfautsch]

After working this out mentally and doing some additional research on Arduino interrupt handling I've learned that accessing time functions inside an ISR appears to be a bit of a tangled subject. This would make it quite a bit "hairier" to detect sequence with only the low-res signal.

This got me wondering how the ECM behaves. If no-one knows I'll confirm myself when I have a chance, but I'm wondering if it will start after:

1) disconnecting the high-res signal
2) removing power from the ECM to clear learned cam position

I know it will start and run without the high res signal, but it was more than a few years back when I worked on replacing my opti sensor and I don't know if I ever disconnected the battery throughout that process. I don't believe I did.

If the ECM is improperly sequencing the injectors I would presume the engine may run, but poorly. If, on the other hand the ECM simply won't detect cam position or fire the injectors without the high res signal that simplifies things immensely. Detecting the position becomes considerably simpler and faster if we need only count high res pulses to determine a trigger slot's opening size.

[Terminal_Crazy]

After working this out mentally and doing some additional research on Arduino interrupt handling I've learned that accessing time functions inside an ISR appears to be a bit of a tangled subject. This would make it quite a bit "hairier" to detect sequence with only the low-res signal.

This got me wondering how the ECM behaves. If no-one knows I'll confirm myself when I have a chance, but I'm wondering if it will start after:

1) disconnecting the high-res signal
2) removing power from the ECM to clear learned cam position

I know it will start and run without the high res signal, but it was more than a few years back when I worked on replacing my opti sensor and I don't know if I ever disconnected the battery throughout that process. I don't believe I did.

If the ECM is improperly sequencing the injectors I would presume the engine may run, but poorly. If, on the other hand the ECM simply won't detect cam position or fire the injectors without the high res signal that simplifies things immensely. Detecting the position becomes considerably simpler and faster if we need only count high res pulses to determine a trigger slot's opening size.

1: Don't know
2: It always runs after you pull PCM fuses?

Can't you watch / count pulses whilst the motor cranks over untill you know the cam/crank position then start firing the injectors.
Presumably monitoring/counting the pulses is only watching we're synchronised with the motor.
Timing time between events so we can measure RPM and time spark & injector firing.

Mitch

[vilefly]

1) it will start, but in simultaneous double-fire mode (1x inject per revolution). Since it has a distributor, all it need to do is fire the coil on the leading edge. The IAC will have to open up to compensate for 0 degrees TDC timing, so it may start and die unless the throttle is held open a little and set down slowly.
I might have to disable the hi-res signal and break out the timing light to confirm if it really is 0 deg base timing.

2) I don't believe there is a memorized cam position, but a programmed one might exist, but not applicable in this fail-safe mode of operation.

Hope my waveform posts on page 8 were helpful.

[spfautsch]

Hope my waveform posts on page 8 were helpful.

There was definitely some useful information in page 8 and 9 that I missed. Re-reading...

1) it will start, but in simultaneous double-fire mode (1x inject per revolution). Since it has a distributor, all it need to do is fire the coil on the leading edge. The IAC will have to open up to compensate for 0 degrees TDC timing, so it may start and die unless the throttle is held open a little and set down slowly.
I might have to disable the hi-res signal and break out the timing light to confirm if it really is 0 deg base timing.

2) I don't believe there is a memorized cam position, but a programmed one might exist, but not applicable in this fail-safe mode of operation.

Are you sure this is actually a fail-safe mode and not an unintended coincidence? I'd love to have confirmation that it will run like this - with batch fueling I can't imagine it would run very well.

The batch fire / double fire explains some of difficulties I had with the startup afr tables after an injector swap.

In consideration of your timing light experiment, do you have a keyed balancer hub? It's one of the items I need to put on my winter to-do list.

[vilefly]

There was definitely some useful information in page 8 and 9 that I missed. Re-reading...



Are you sure this is actually a fail-safe mode and not an unintended coincidence? I'd love to have confirmation that it will run like this - with batch fueling I can't imagine it would run very well.

The batch fire / double fire explains some of difficulties I had with the startup afr tables after an injector swap.

In consideration of your timing light experiment, do you have a keyed balancer hub? It's one of the items I need to put on my winter to-do list.

GM always used simultaneous double-fire (all injectors per 1x rev) when starting on prior years, but to blast the shadow of doubt into oblivion, I will see about getting a waveform to prove it. We are crazy-busy at work, so it may take some time to get. My other injector waveform proves the cranking strategy, though.
My harmonic balancer is marked at 0 deg, as I was testing fitment of my northstar crankwheel project. I will probably use the hi-res waveform to prove exact timing.