Chrome v2 - Limp mode at 3000 rpm?

[dcreel]

Hi Everyone

I've been struggling with a problem for at least a year now, and I've finally given up on diagnosing it myself. I've also received a notice that it's time for a technical inspection in mid-December. The technical inspection is pretty serious here in Switzerland, and I don't think I will make it through in its current state.

It sounds exactly like this: http://www.3sgto.org/turbo-engines-s...e-stumped.html
That car randomly fixed itself, so that thread didn't help me other than to at least know I am not the only one

I would really appreciate it if you guys could offer some advice. Details below.

Car: 1995 VR4
Relevant Mods: Chrome v2, ECU from Brett, 750cc injectors, Pampena short block with forged internals, Aeromotive FPR
Any other mods shouldn't really be relevant to this problem.

Problem:
Driving with low load and not much throttle, in any gear, as soon as it hits ~3000 rpms the ECU drops the timing advance to like 5 deg. This causes it to run terribly, of course, and it gets stuck in this error mode until I rev it up to ~5000 rpms in neutral with full throttle. Sometimes, it is so bad that I can't get it over 3000 rpms anymore at all. In that case, I am stuck with < 3000 because it drives pretty normal below that. If I disconnect the battery for 10 mins, it clears out this error mode and works fine until the next occurrence.

It does not happen with WOT, and also not with the MAF disconnected. I think it only happens in closed-loop mode.
Today I made a datalog that clearly shows the problem happening. I have basically all fields in the log that could be somehow relevant. See attachments for log file and graph.

Some things that are definitely going on:
- The knock sensor gets into a fault condition and outputs a constant knock sum value. This is configurable in Chrome, so I have changed it to 0 for the constant value and it doesn't help. The knock voltage looks ok, however. Also knock base and knock sum until that point are fine. I have played with knock settings to make it less sensitive to noise, but didn't have any luck.
- The octane flag goes from 100% to 0%. I have tried to disable the octane/knock logic in chrome, but I think that caused other problems. Could have been my tune at the time, though. I will try that again.
- The timing advance drops to something between 0 and 8 degrees or so.


Some things I have replaced already:
Fuel pump
Fuel filter
Fuel pressure regulator
Diaphragm in FPR
Injectors
Injector wiring harnesses
Changed my FP hotwire setup to also include the FP resistor like stock, in order to prevent overrun.
FP relays
Air filter
*Special* boost control solenoid, but that is disabled until very high load. Also tried stock and just running disconnected.
BOV
Intake piping
Knock sensor
Knock sensor wiring (close to the sensor at least)
Fixed most boost leaks
Idle stepper motor
Throttle position sensor
Throttle body seals and adjustment screw
Spark plugs
Spark plug wires
Measure coil resistances, looks ok
PTU is fairly new, but I still have the old one. I will try that also.

At this point, I am thinking of swapping my stock injectors back in and trying the stock ECU. If that doesn't work, I still have my MegaSquirt I guess...

Thanks for any help you can offer!
DavidC (aka VBBlckBelt)

Edit: terrible image quality on the graph for some reason. Here it is on imgur: http://imgur.com/QBogw5R
776cc_weird_problem.jpgDatalog plus tune.zip

[donniekak]

I believe it's a wiring or hardware issue with the ecu. I remember in some of my early testing with flash ecu's(way before chrome) I had this happen. It only happened with some of the sohc ecu's that I converted. I suggest contacting the manufacturer of your ecu and having them test it on a test bench.

[dcreel]

Thanks for the advice, donniekak. I can pretty much confirm your guess now.

- I put my stock injectors back in and copied the stock fuel settings into my current tune, and the problem was still there.
- I then put my original ECU back in, problem gone!
- I stopped at a gas station and put my Chrome ECU back in and started from the plain Chrome v2 binary. I turned off the peripheries for secondary O2 and whatnot, as in my other tune, and the problem was still there.
- I stopped at a rest stop and put the original ECU back in, problem gone again!

I believe it must be something with the knock sensor hardware, because I datalogged again while using the Chrome ECU and I had knock sums of 9, indicating knock sensor failure. Is there some known problem with knock sensors and the clone ECUs?

Anyway, I am stuck with my original equipment for now, but I will probably pass the inspection. What a relief!

My Chrome ECU is from THE ORIGINAL VR4 Clone? ECU | Custom Clone, Clone² and Universal Clone ECUs for 3000GTs
I will try to contact him and see what can be done.
I believe it has not worked properly from the beginning, but I always assumed it was some other problem since I was changing other things at the same time.
If I had known I would have this much trouble, I probably would have just stuck with MegaSquirt :-/

[donniekak]

That is exactly what was going on with some of the ecu's I tested, at first I thought it was the conversion harness I made, but other ecu's worked fine. I never looked into it much as this was before chrome and it was easier to get 98-99 ecu's.

[Unlogic]

Keep us posted on how this goes. Hopefully you'll be able to replace your Chrome ECU with a working one from Brett without any hassle.

[DK77]

Definitely a problem with the knock sensor circuit. I had the same problems, even posted in the thread you linked. Could be in the ECU or in the wiring between the sensor and the ECU. I had the problem with multiple ECUs so my issue had to be in the wiring. I took it apart and tested everything and it was all good, so I put it back together and it has worked since then. I must have had a loose connection somewhere (my most likely guess would be where the sensor wiring plugs into the main harness near the plenum). It's possible its your ECU so if you can get it checked, do it.

[Greg E]

Could also be an issue with the wireharness adapter too.

[dcreel]

Thanks everyone. I suspect it's not something under the hood, because I can swap out only the ecu and it changes the result every time.

What's weird is I do get a signal on the knock sensor. Seems realistic to me as well, but I don't have much experience with diagnosing the oem knock sensors.

I will take a look at the harness adapter this week sometime. I did check the O2 sensor orientations there, but I didn't think about checking the knock sensor.

[dcreel]

I haven't received a response from his gmail address. Does someone know another way to contact him? Please pm me if so.

I would normally wait a bit longer, but I am visiting the US soon and it would be most convenient to take care of this during that visit if possible.

[mjannusch]

Mine did the same thing, bad knock sensor circuit in the ECU. Have to send it in for repair.