[Shinji]

Ignition timing shifts between cylinders.
I am posting for the first time from Japan. The car is a 1972 Suzuki Kei-car Fronte Coupe, 2 stroke 3 cylinder 360cc engine.
I've had my Speeduino for over 2 years and it runs fine. However, recently I noticed a discrepancy in ignition timing between cylinders.
The situation is below.
No. 1 cylinder timing (crank angle = 0 degrees) .... #1_cylinder timing.jpg
No. 2 cylinder timing (crank angle = 240 degrees) .... #2_cylinder timing.jpg
No. 3 cylinder timing (crank angle = 120 degrees) .... #3_cylinder timing.jpg
*ch1-ignition signal ch-2 crank angle signal (pulley: 6-1)
*Engine speed is at 6000rpm.
*The crank angle signal is an accurate pseudo signal generated by the Pulse Generator.
The time between the falling edge of each ignition signal (ignition here) and the immediately following crank angle signal should be equal. However, the observation results are as follows.
Cylinder #2 lags behind #1 by about 100usec, and cylinder #3 lags behind #1 by about 50usec. This corresponds to an angular delay of about 2 to 4 degrees.
Below are some supplementary information.
・The injector signal does not have such a timing shift.
・The firmware version is a bit old 202108, but the latest version 202305 was the same.
- Attach CurrentTune.msq.
Please let me know if anyone has any information about this inconvenient phenomenon.
Thank you.
Shinji

[jonbill]

That looks an interesting engine.
Do you have a log of it running with this problem?

[Shinji]

Thank you for your reply.
If there's a way to get some logging to parse this, please let me know.
I'll give it a try.

[JHolland]

Up to 100us is quite normal for a Speeduino, it uses a PC style scheduler unlike a conventional ECU. Does your scope have a persistence setting? if so then set that to 100%, that will give you a better view of what your timing looks like.

[JHolland]

What crank sensor pattern are you using?

[Shinji]

Thank you for your reply.
Would you point out that the 100usec variation is within the speeduino's synchronous control error range?
I know the Speeduino has a certain margin of error, but it was around ±30usec. (photo attached)
And this variation between cylinders is determined by a strong will, with a deviation of 50usec to 100usec.
And there is no such large error in the injector signal between cylinders. Therefore, I thought that there might be a problem with the ignition timing generation algorithm of Speeduino.
In addition, if the error of 100usec is within the specification range of Speeduino, I think that the ignition signal cannot be used.
Because 100usec at 10000rpm is equivalent to 6 degrees of crank angle, so such a difference in ignition timing between cylinders is fatal for the engine.
What do you think?

>What crank sensor pattern are you using?
The crank sensor pattern is 6-1.

[JHolland]

That is a better scope capture. 100us is what I have seen previously, you are clearly getting more accurate timing than I was. Have you checked that the input signal is accurate?
If you run it at a different rpm do the delays change?

[PSIG]

@JHolland - I have seen variable input/output delays, but consistent with little jitter or scatter in most setups with little variation moving from cylinder-to-cylinder with a timing light, on ones I've checked. Timing table compensation is simple for this overall effect, as it is relative.

But are you saying C-to-C variation or per-cylinder scatter is a normal or expected effect you have seen?
Should I see this variation on the crank angle with a timing light every 720°/n-cylinders apart (as one example test)?

I agree a v2 code base using 'conventional' interrupt should be investigated as a useful improvement from a variety of perspectives. Unless of-course someone's other code, patent, or other block is preventing that, or otherwise forcing scheduling.

[Shinji]

@J Holland
> Have you checked that the input signal is accurate?
These waveforms are obtained when a square wave generated by a pulse generator is input as the input signal (crank sensor signal), and the signals are much more accurate than the sensor signals on the car.
> If you run it at a different rpm do the delays change?
This delay is invariant with rotation speed. In other words, it is the same at 1000rpm and 100000rpm.

[Shinji]

Replaces the previous post with the figure.
-------------------------------------------------------
I would like to share additional data.
I attach the waveform diagram.
No.1: Overview of crank signal and ignition signal (#1 Cylinder)
No.2: Ignition signal relation diagram for #1 cylinder (crank angle = 0 degree) and #2 cylinder (crank angle = 240 degree)
The falling edge of each signal is the ignition point.
At this time, the rotation speed is 6000 rpm, and the crank angle difference between the cylinders is 240 degrees, so the time difference should be 6.67 ms, but it is actually about 6.77 ms, with a delay of 100 usec. This is a constant lag, not a jitter. (There is a Jitter is about ±30usec as reported earlier.)
Shinji