No start due to P3106, ECM -> HV ECU communication failure?

The motor on my 2002 prius stopped while driving, and will not start. The car will not go into READY mode, but still allows slow operation on battery power. When powered up, the car usually makes one attempt to spin and start the motor, but there is no spark.

The trouble code listed is P3106, listed as the HV ECU complaining about no communications from the ECM. I made sure the ECM is getting good power and DC signals according to the service manual, and then replaced it with another used module, but the problem persists. I'm now about to investigate the serial data link and wiring between the HV and the ECM, as well as speculatively replace the HV ECU. Good thing gen 1 computer modules are cheap and readily available on ebay.

Has anyone heard of this specific failure? Does anyone have any hints for investigating the serial data bus? It looks like a pair of unidirectional RS422-like serial links, each over a shielded pair. HTE-/HTE+ and ETH-/ETH+. I was going to put a scope on the ETH pair and see if the ECM is even trying to speak to the HV ECU.

Any other hints?

 

RE: discharged traction battery. Indeed. That was my first challenge. The motor quit at the exact top of a mountain pass, so the battery was nearly discharged to begin with. Limping off the freeway and a few days' worth of start attempts, tinkering, and moving it around the yard left it completely discharged.

Luckily, I have a 36V->360V DC/DC converter that I'd cobbled together as part of a DIY plug-in conversion for this car. It's been on a back burner for a while, but I dusted it off and trickled an amp into the battery for a few hours, and it's good to go. Plus, moving the car around back behind the house after recharging the traction battery now officially counts as my first PHEV trip, eh?

RE: scan tool recommendations and dealership diagnosis. Thank you for the advice. I did tow it to the dealer for analysis, and despite me explaining that I was going to tinker my way through this problem, and specifically requesting a printout of all available info, including sub codes, other ECU trouble codes, and freeze frame info, they only had the gross overview paperwork that reported P3106 when I went to pay and pick up the car. They claimed they could not retrieve the more detailed paperwork or records, but did offer to run a diagnosis again for another $100. I was very frustrated, took my car, and went home.

RE: HV ECU connections- I'll do this. I had exposed the HV ECU in preparation for the replacement, and wiggled and inspected the connectors, but nothing jumped out as bad. I was so sure that it was a bad ECM that I didn't do much more diagnosis while I was waiting for the replacement ECM to arrive from ebay.

The failure happened at the exact peak of a mountain pass, about when I let off the gas pedal. I figured the high-current ignition and injector circuitry in the ECM would have been at unusual peak thermal stress for the 10 minutes of max RPM abuse coming up the pass. Combined with the P3106 code, I was pretty certain the ECM was just dead.

Anyway, I will keep you posted what I find. Thanks for the concise and expert advice!

Jim

 

Solved. WHEW!

I spent a few days trying to figure out why multiple ECM's all failed to communicate. I put a scope on the serial communication lines, and although the HV ECU was chattering away at the ECM, the ETH+/ETH- lines were completely dead. Same thing with multiple replacement ECM's. I poured over the power supply and DC signal lines, looking for bad grounds, bad connections, everything.

Turns out it was a bad fuel tank vapor pressure sensor that was shorting out the ECM's 5V sensor power supply wiring. (Pin VC on the ECM.) This irritated the ECM enough where it wouldn't even boot. Possibly some internal logic or processor in the ECM uses the same power supply, which is shoddy if it's true. Anyway, I disconnected the faulty sensor, and the car sprang to life.

If the sensor Vcc is as critical to other ECU modules as it is to the ECM, checking for a healthy sensor power voltage (Typically pin "VC", specified as 4.5 to 5.5) should be a top step in troubleshooting any suspect ECU.

 

Yes, I disconnected the sensor at the tank. Line E2 is a sensor ground, shared by several sensors. The faulty pressure sensor basically shorts it and VC together, drawing VC down to about .2V.

I'm driving it around disconnected for now, but I assume this will cause it to fail its ODB2 gastank pressure test cycle next time it runs it, so I will need to replace it before my car is due for its emissions inspection.

I hadn't thought of spoofing it. I believe that particular emissions test cycle just sees if the gastank can draw (and hold) a vacuum, so just putting in a resistor that returns the approximate vacuum value on PTNK might work.

Thanks for the spoofing idea. Anyone need 2 classic ECMs and 1 classic HV ECU, by the way?

 

I grounded the PTNK signal to simulate a good tank vacuum, and this stopped most of the DTC's, but when the car ran the OBD2 evap monitor routine, it failed with a P0450, and left the evap monitor marked incomplete. I'll need to either replace the sensor (ie: remove/replace tank- ugh) or spoof the signal (better) before next emissions check.

Upon reading this link: (can't post link, search for "EVAP VSV Prius Hybrid13.pdf")

... I am not surprised grounding the signal did not work. The response expected from the gas tank pressure sensor is fairly sophisticated.

The ECM applies intake vacuum to the tank, and watches for the pressure to drop from atmospheric pressure to 20mmHg within a certain time limit. Then it removes the vacuum and isolates the tank, and watches that the vacuum held in the tank does not rise too quickly.

I will devote an hour or so to trying to mimic this response with a little RC delay circuit driven by the signal from the ECM that seals up the tank for this test. I'll post results under a new thread if this works. It could be a fairly useful general scheme for spoofing an ODB2 leakdown test using $1 in parts.