Suddenly loosing VR signal

[gboezio]

Hi, first I must say that I had great results with MS3, but I'm having a bad VR issue that I'm not sure how to start to tackle this.
It's like the problem I had before with my 8-0 trigger wheel 7k on decel and bam no crank signal. First I thought that MS3 freezed, but I could catch it on the tooth log since it does less randomly, but I couldn't catch it on the scope beeing too slow. It did not skip a pulse since 16-2 unless low rpm and raced 6 two hours races + practices, then during a dyno session it started to loose sync on decel, wasn't bad, but it happened, then I have changed the one wire CLT sensor for a GM two wires, I tend to rule it out because it started before.

What is it : MS3 DIY 3.0 board, 450CC single with stock VR sensor
There is a cam GT101 hall that reads the cam for phase

There is a lot of vibrations on this race quad, but ECU box is weatherproof and mounted on rubber bushings


What I know :
It happened before, so I thought it was noise from ignition coils that I have removed, at this point I went from 8-0 to 16-1 to 16-2, trimmed the pots and compensated the little RPM offset.
Then the trigger got usable and work reliably except at rapid changes at very low RPM, but it was a crank speed issue, I got 20+ hours of hardcore race trough all loads, RPM, vibrations.
Grounding had been taken seriously.
During dyno testing the last pulls started to show sync losses during decel at high RPM and the engine died a few times
Then I changed the CLT sensor that became useless, pretty straightforward.
Then the problem of the first races came back exactly like before, it may not be the same cause, but the symptoms are very similar.
Sometimes It will die until I'm to a complete stop sweeping trough the whole lower rev range where it works flawlessly, but logs shows a lot of shorter loss sync.

What can behave like this, a failing component, VR pots shaken by vibration (in this case I'd expect a recovery when the revs goes down), sometimes does.
VR sensor going out, broken solder, noise.
I just need ideas where to start troubleshooting

Ms system have drawn attention, two people came to me interested and the dyno guy was impressed as well and asked me prices and info, I need to get this straight. I'm thinking to build another for ice racing, I have a good recipe. I'm thinking to go 24-2 on it even 36-2 if I could fit it on a 4 inch flywheel, but it seem too tight.
When I get this to work, I will do a write up for the single cylinder people, the trigger is very difficult and I think it needs more resolution due to the light crank mass/speed variation

[gboezio]

Can't catch any fault on the scope, but signal is present at MS input when it dies out completely

Edit : I read how to scope the output or the VR conditionner, can't catch a fault but the signal is present to the CPU all the time during a failure

https://youtu.be/yJPhy_8mGMk

Edit two : I have removed the box and ran the motor holding it in my hands, same issue, pretty much ruling out the vibration issue.

[Matt Cramer]

The reading on the frames with lost sync is there but for some reason isn't displaying properly. Can you try a tooth log? I find these a bit easier to read with missing tooth wheels, and they're less likely to have rendering glitches.

[gboezio]

Thanks matt, will do. But I did fix the whole issue rolling back to a restore tune when it ran good, but there are just minor adjustments, I'm still comparing the differences to point what setting triggered this

[gboezio]

Turned out to be my race fuel timing tables MSQ was sent with pump fuel tables, they hit 50 deg over decel at 10k RPM, as soon as I swapped the table, the computer lost sync over 16 events, backed up to 44, still happened, backed to 40 vanished.
I don't understand how timing advance can kill the engine trough the whole table including less advance where it should pick back up but it's frozen for some reason

Edit : here is a 46-50 deg decel log (52 sync loss and killed the engine), pulled it to 40 deg max and made the other (0 sync loss)
seem that it was a rendering glitch as you pointed out, but it did fit where the engine lost power and coasted down in revs without re synch for seconds
I'm pointing this out because, maybe part of the problem overlaps there, like in the CPU output to spark and output to the composite log.
Ditch the other MSQ

[Matt Cramer]

It is failing to see one of the teeth. This could be a very weird misfire induced noise issue - usually this adds pulses rather than subtracts them, but that's about the only guess I can come up with.

[gboezio]

Yes, it obviously misses a tooth, this may be what triggers this whole thing, or it is caused by it.
While shearching for hints I have realized that the logs were SD card logs, here's a computer log on the dyno.
Can I draw your attention on Status 1 and 2, everything goes normal until hitting 44 to 42 deg on decel, then it looses sync, status changes from full sync to rpm sync, this happens while there is no spark, there is enough fuel to burn there but the O2 sensor is going lean tending to confirm the spark loss, there are random sparks that ignites here and there, it matches the pattern shown on the tooth logger ( this is why I think that the render problem may come from the ECU), usually the engines dies completely even if I would downshift gears. It will sometimes pick back up close to idle or cranking, if it goes cranking it goes back to fully synched if it pick up before it will stay to RPM synched (8). There is also the CEL going on there and limp mode going on and off. Need burn was caused by me advancing the timing.

[Matt Cramer]

The status change indicates it has lost sync. For some reason, it fails to resync with the cam sensor, but this appears to be a symptom and not the cause.

[gboezio]

Well I had the idea to compare VR input to spark only to see that spark output still fire the coil, but to my surprise I caught some 360 crank degree spark reversals, like if the computer still fires in 720 degree spark but at wasted spark timing, it seem to have very little dwell time, it happens a few times.
To add to the mystery, then the cam sensor falls flat, sometimes the engine keeps running, then reverse stroke again and die.
The cam circuit is a GT101 reading the cam lobe for phasing, it runs from 12V, have a dedicated ground I think and has a 5V 1k pullup, it can run all day without skipping a beat, but crashes right after the timing reverse. It's grounded either by the sensor itself or the Opto circuit, or I'm loosing pullup but TPS keep working.
All this triggered by advance.

It starts by random loss of sync, spark firing on exhaust stroke, comes back a few times, cam signal will quit, sometimes engine will run some, then will loose stroke but won't fire twice per cycle, will instead fire on the exhaust stroke killing the motor. Every times I made the problem happen cam will quit, otherwise it can run all day. This is where I'm getting lost, like the sensor gate locks up and conduct to ground and pull down the pullup. This beats my logic. Even if it goes away pulling a few degree, I feel that this must be understood to be avoided.

[gboezio]

Well, it now made sense, I have caught the cam sensor to go to hell many times, it seem like it's happening after a long runtime and most of my testing was made cold.
The rendering glitches even tough seemed to replicate the problem were rendering glitches.
The ECU works too well so I wasn't aware that the problem of the first race wasn't fixed at all, while the cam sensor was failing the ECM counted too well on the crank for me to know it happened, so by fixing the first trigger problem went away but not too far.
During dyno advancing the timing caused (I think)some backlash on the crank and main drive making me loose a tooth, call it mechanical noise.
Then while the cam sensor was out, it lost track of the engine phase and reversed stroke. I would be nice to revert to wasted spark in those case (Status 1 @ 8bit, not in full synch),maybe a good improvement of MS3 for racing.
In the end I have set MAP cam sense and run wasted spark, backed up the lower ignition table to avoid the decel backlash.
I won't recommend the GT101, it seem to be best for other applications like collecting dust on a counter