I’m chasing an issue with a stange oscillation in the battery voltage measurement at the ECU. I attach a picture form the log (I also attach a log). This happens only with the engine running.
Basically, the battey voltage read in the ECU oscillates by 800mV with a really long period of more than 5 seconds. I have ruled out the alternator as the same behavior happens without the alternator belt. It therefore looks to be aliasing, altough the super low frequency does not make sense.
All other analog signals look actually quite clean to me. The only thing moving with the same behavior is RPM and lambda, which I somehow got stable by flattening the injector dead time compensation in that voltage range and pushing the idle PID
I have also tried to probe the 12V refererred to POW_GND with an oscilloscope direclty on the ECU pins and the voltage looks stable (second picture, oscilloscope is at 100kSa/s, then the plot is filtered to 50Hz)
I try to add some details that might be useful:
no PMU in the setup
All power GNDs are connected to a single point in the engine block with 2x14-AWG wires. Engine block is then attached to chassis ground.
Analog GND is floating wrt chassis/power GND when ANA GND is disconnected from the ECU
12V supply comes directly from battery isolator through a fuse
12V switched comes from the “PMU/ECU output” of the battery isolator
Noisy stuff that are controlled through the ECU are injectors, ignition (passive coils) and DBW. (VVT is also there but disabling it doesnt help)
Battery voltage is sampled at 500Hz. Filtering is set at 30ms, but the unfitered voltage shows the same behavior
The oscillation behavior slightly changes with engine RPM
Alternator is pure analog, without control on the ECU side
If you have any suggestions I would appreciate! Thanks a lot!
Bypass isolator? Could be some active components in there.
Is scope referenced(grounded) to same spot as ECU?
Is it possible to do some channel math and put probes on each side of isolator to look for deltas? Maybe some of these deltas start looking like the 6s period above?
Thank you a lot for the feedback!
The scope in that log was referenced to one power GND at the ECU. I used one of the two 12V and one power ground to probe the voltage directly at the ECU input pins.
I don’t know however if the battery voltage internal readout is referenced to power GND or analog GND, this might make a difference but I actually didn’t check the voltage referring to analog GND.
I also tried your suggestions by bypassing the battery isolator and unplugging the alternator, but with no difference.
I can leave with it, though I have to be careful and flatten all battery related corrections in that voltage range..so I’d love to resolve the issue at the source
I completely understand and agree!
This is an interesting problem. It’s way too regular over a long period of time to be aliasing (in my opinion). Is there an A/C compressor cycling? An air pump cycling? Fan cycling? O2 Heater cycling? Hmm…
I’ll do some more thinking on other things to try.
You mentioned using the same gnd at the engine block and the battery voltage shows the same oscillation. If every circuit has a dedicated supply (battery +) and a ‘common’ ground this is called a common ground. Noise is a typical symptom of this and voltage drops (Like the dash shutting off during cranking).
IMO you should move Power Gnd’s to the battery (ECU TCU PMU). Also if this is an older alternator that uses a diode trio you may need to add a filter to the alternator because it will never be smooth.
Thank you bravo for the answer.
I totally agree GND is not an absolute zero, but either way you wire it there is going to be noise coming up. Wiring to the engine block is now a must as I would need to change the entire harness, but for exalmple the alternator (which is the main current source providing all the current when the battery is fully charged) is referencing to the engine block itself.
The concern is that there the behavior is weird as it has a really long period but I don’t see any relations with other loads or other things moving that can explain it.
It’s going to be a racecar so no AC nor electric air pump. Fan are off…Regarding O2 heater I checked the log and to me the duty cycle of the haters is not showing strange or cycling behaviors. Still a mystery.
Thanks again for your suggestions!
