TL;DR: One of the pulses from the camshaft sensor is half of the amplitude of the other two, feeding the MRE a fake cam sensor signal with all the pulses close to the same amplitude does not produce the problem I'm seeing. Maybe the amplitude difference is tripping up the TLE8888 somehow?
Context: This is a used microrusEFI harvested from an old PNP board and there's some silkscreen damage and the thru hole for VR+ pin 45 has a fly wire connected to the CRNK+ pad of R17 to fix this. This particular ECU has been through lots of abuse including shorting something through the STM32 and killing it, but both the STM32 and the TLE8888 have been replaced since then. The case is also lined with kapton at the moment to rule out the board shorting to it.
Behavior while cranking:
Today I redid the fly wire under a microscope and I don't believe that or the wiring going from the microrusEFI to the bike are the source of the problem. I logged the output of the CRANK pad of the TLE8888 while cranking and this is the result:
- output of CRANK pad while cranking, this is what the MCU sees
- CRANK pin output when actually cranking.png (36.24 KiB) Viewed 61075 times
I also probed the output of the camshaft sensor at both the stock ECU headers of the bike and at the board side of the microrusEFI's connector and got these two waveforms. Notice one of the pulses is only around half the amplitude of the two closely spaced pulses. I think this may somehow be tripping up the TLE8888, and as far as I can tell this isn't due to any of the wiring or any problem with the microrusEFI.
Probes at the microrusEFI:
- Camshaft sensor output measured between vr+ and vr- pins on the board side of the microrusEFI's connector
- VR+-pins zoomed.png (23.89 KiB) Viewed 61075 times
Probes at the stock ECU headers before the adapter harness:
- Camshaft sensor output measured at the connectors for the stock ECU
- Cranking VR sensor measured directly at STOCK ECU headers zoomed.png (40.22 KiB) Viewed 61075 times
Behavior with a simulated camshaft sensor:
Earlier today after inspecting the board and redoing the fly wire I connected the MRE to a power supply and a function generator and fed it a poorly drawn approximation of the camshaft sensor output, which resulted in a good waveform on the CRANK pad unlike what I'm seeing when actually cranking the bike. When I logged this with tunerstudio the trigger error wasn't incrementing over a 10 second datalog, and the triggerchannel1 waveform looked good.
Badly drawn approximation of cam trigger:
- Fake cam sensor output sent to the microrusEFI
- simulated cam sensor output.PNG (30.48 KiB) Viewed 61075 times
CRANK pad measurement with above waveform as an input, Much nicer than when cranking on real hardware:
- TLE8888 CRANK output when simulated cam sensor waveform is connected to VR+ pin 45, as far as I can tell this confirms everything on the microrusEFI side is functional.
- CRANK pin output with simulated VR sensor input.png (34.15 KiB) Viewed 61075 times
Trigger events in tunerstudio with fake cam signal:
- Tunerstudio showing a much nicer looking trigger waveform
- tunerstudio log with fake camshaft sensor input.PNG (60.16 KiB) Viewed 61069 times
Misc info:
The crank and camshaft sensors have a common - pin, but disconnecting the crankshaft sensor leaves the camshaft sensor waveform completely unchanged and I doubt that's the problem.
All of this makes me suspect the issue isn't hardware damage that makes it impossible for the microrusEFI to reliably process the camshaft input and is somehow related to the pulses coming from the camshaft sensor having two different amplitudes. Tomorrow I'll read the TLE8888's datasheet and see if that has any relevant wisdom.
