New Tesla LDU fault - inconsistent failure to start

[P.S.Mangelsdorf]

I've got a new fault condition with my LDU that has me stumped. Hoping the brain trust here can help sort it out.

So I recently switched from my super janky 2 base phases and 1 sport phase setup to all sport phases. This used my two good sport phases and a good sport phase I bought off ebay. I just swapped out the sport phases, so the rest of the components are exactly what was in the base/hybrid setup prior. When I first tested the car, it worked. Then a week later, when I went to test, it would not start. I opened it up, and powered on with 12V only, and it showed no errors. I put HV to it, and it started fine, no errors, spun the motor. Put the case back on, and went out for a tuning session. Whenever needed, it restarted fine. I parked the car for a couple of weeks, then charged it up to do more tuning. When I went to drive it, it would not start. When it doesn't start, it "starts" but when I put it in gear it is an immediate Overcurrent error, which will only clear with a power cycle. I tried disconnecting HV by unplugging the contactors and starting, then starting again with HV connected, and same issue. I have not reopened it again, as I'm not sure what that would do differently.

I did a diode test per Damien's video. When testing on the main bus bars, everything is normal. When I test from the main bus bars to each phases output is where it gets weird. On all 3 phases, when I test with the positive probe on negative bus bar, negative probe on phase output or when I test negative probe on positive bus bar, positive probe on phase output, it shows the correct reading of approximately 0.23V. However, when I test with the positive probe on positive or negative probe on negative, all 3 phases show an open circuit. I do not get the rise in voltage before dropping to open as it is supposed to.

So, any thoughts as to my problem? Do I have a bad phase? Is it possible the bleed off circuitry is bad? Is it something completely different? I'm at a loss.

[johu]

Which phase shows the error led? Always the same one?

[P.S.Mangelsdorf]

Which phase shows the error led? Always the same one?

I don't know. In this most recent instance, I haven't removed the case (just the bus bar covers), and the previous occurrence, once I removed the case and powered it back up, there were no errors and no error lights. The challenge is that to remove the casing and see the internals requires removing the HV and 12V, and that seemed to change something the first time.

I suppose I could try removing the case, and see what happens. If the errors do go away, leave it with the case off for a few days then try again, as it seems to happen after sitting.

[P.S.Mangelsdorf]

Which phase shows the error led? Always the same one?

So the weirdness compounds - I took the case off, reconnected power, tried to start it up. Still did the immediate OC error BUT no error lights on any phase.

I'm going to try to track down the sensor and discharge circuit from my second LDU and try swapping them in this weekend and see if that helps. I know they are different than the ones in there, maybe substantially enough that I should have done so from the beginning. We'll see.

If anyone has other thoughts, I'm all ears, because this is such a strange failure mode.

[Boxster EV]

I can’t help with your feedback about the IGBT tests, however I had a similar issue to the symptoms you’re describing a few years ago, and it turned out to be gate drive chip going bad. I managed to change the chip over.

It was however eventually evident by the red LED coming on. There’s a thread on it somewhere.

Edit: thread here viewtopic.php?p=41690&hilit=Gate#p41690

[P.S.Mangelsdorf]

I can’t help with your feedback about the IGBT tests, however I had a similar issue to the symptoms you’re describing a few years ago, and it turned out to be gate drive chip going bad. I managed to change the chip over.

It was however eventually evident by the red LED coming on. There’s a thread on it somewhere.

Did you have any indication before the light came on which gate driver was bad?

[Boxster EV]

Did you have any indication before the light came on which gate driver was bad?

I had the drive unit powered up on the bench, and I didn’t notice any physical issues. However, I think tried it in the car and the next time I removed it the chip was fried and light on.

[P.S.Mangelsdorf]

I had the drive unit out of the car and powered up on the bench, and I didn’t notice any physical issues. However, I think tried it in the car and the next time I removed it the chip was fried and light on.

Hmm. What seems so weird to me it that is worked, didn't work, worked well enough for me to push it exceptionally hard, drove home, and then didn't work. I'm planning to go over it with a fine toothed comb to see if anything stupid like a loose connector is causing it.

Did you replace the whole gate driver board or just a chip?

[Boxster EV]

Hmm. What seems so weird to me it that is worked, didn't work, worked well enough for me to push it exceptionally hard, drove home, and then didn't work. I'm planning to go over it with a fine toothed comb to see if anything stupid like a loose connector is causing it.

Did you replace the whole gate driver board or just a chip?

I changed just the chip. In fact, the donor part was one from the haul that was sent to you.

[P.S.Mangelsdorf]

I changed just the chip. In fact, the donor part was one from the haul that was sent to you.

Good to know. I'll have to check whether the chip is the same on the base boards and the Sport boards. Because I have quite the collection of phases now.

[Boxster EV]

Good to know. I'll have to check whether the chip is the same on the base boards and the Sport boards. Because I have quite the collection of phases now.

Did you see this:

viewtopic.php?p=41690&hilit=Gate#p41690

[Boxster EV]

Just thinking further about this, it does seem as if a loose connection or dry solder joint would be more likely. Seems odd that you can give it the beans one minute but then it won’t work the next.

[P.S.Mangelsdorf]

Quick update: I think your gate driver theory might be correct. I changed out the current sensor and discharge board with it in the car, no improvement. Still no error lights. I pulled the inverter and diode tested again on the bench. All passed (I think the weird diode tests before were me testing back to back too quickly.)

My plan today is to swap in a known good base phase one by one to figure out which phase has the issue, connecting it to power with jumpers. Once I find the bad phase, then I might swap the gate driver board.

Also, it does appear that the Base and Sport gate drivers have the same part number, but are different colors (Sport is red). a bit odd, but might work out for me. Also, I have a possible line on another "for parts" Sport drive unit, but I'd like to avoid spending more money on this.

[P.S.Mangelsdorf]

Well I'm back around to a loose connection. I think. I set up the inverter sitting on a box with jumpers to HV and the motor, and with a base phase subbed in for one of the Sport phases. The plan was to keep switching this around in place of each phase, with hope that it would work in place of one phase, and that would tell me which Sport phase had a bad gate driver.

[WARNING] Dumb hot rodder alert, do not try this at home. (go over to a friend's house)

So I set that all up, powered things up, and it worked! What luck, the first phase I tested was the bad one!?!? Then I remembered its never that easy with this car. I disconnected the base phase, and using the same jumper set up, connected with all the sport phases, and it worked! Which means it wasn't a bad gate driver. The only thing I can think is that something is loose. I did let it sit for almost an hour and tried again, just to make sure, and it still worked.

I did notice a capacitor on a gate that wasn't completely solid, and some of the connections to the brain board seemed ok, but a little loose. What I decided to do was use some potting compound on the capacitors on all 3 gates, and then use some UV setting superglue to permanently attach the wiring harness to the brain board. If I ever have to swap the brain board, I have a spare wiring harness. I'm going to let that all set overnight, and try reinstalling in the car tomorrow or Tuesday.if it doesn't work again, I'm going to start throwing tools.

[P.S.Mangelsdorf]

Well I'm now completely stumped.

Yesterday evening I got the inverter back in the car. I successfully started it multiple times, and took it on a quick rip around the neighborhood, including out onto a main road where I did a couple of WOT pulls (with the "street" tune that runs 12.1 in the 1/4). All good, drove well, seemed perfectly happy. I parked it in the garage, turned it off, went to bed, and went about my day today. This evening, I went out to test that it would start again. It did not.

Clearly there is something wrong somewhere, but with no error lights I have no idea where. I'm pretty sure its not the current sensor or discharge board, as changing those didn't fix it. But I have no idea now what DID fix it, or why it stopped working again.

The last thing I'm going to try before I go hunting for new phases (again) is swapping out my 12V battery for a new one. Now, I don't think this is the issue. I've been starting the DC-DC before starting the inverter, and the first thing I checked was the 12V voltage with and without the DC-DC. The only thought I have is maybe the gate drivers are drawing too much on start up, quicker than the DC-DC can respond, and the 12V battery isn't strong enough to carry it through. It's a hell of a stretch but its the last idea I have.

[Boxster EV]

Feel your pain.

New 12v battery is not a bad shout.

Stating the obvious but can you stick a couple of multimeter probes into the back of the 23pin molex to check you’re getting decent voltage there on start up?

Once a direction is selected and things are clocked, it does consume some power. I’ve never measured it but I can hear the tone of my WP changing.

[P.S.Mangelsdorf]

Well, it's not the 12V. I replaced it with a battery that has even more CCA, and same thing. There's got to be some sort of issue with a phase. Back to ebay I go.

[johu]

Does "won't start" mean it faults out with OVERCURRENT? If so it would be worthwhile checking the 7 fault signals

Maybe even go as far as soldering wires to them and scope if they look fine statically?

[muehlpower]

for my understanding, what triggers an “overcurrent” message. Of the 7 inputs on IC10, 6 come from the error lines of the gate driver. One comes from IC8, and is also transmitted to the CPU. The output of IC10 suppresses the PWM. Is this suppression also output as an “overcurrent” message? If not, the error message can only come from IC8.

[johu]

If only someone had programmed it that way

Code: Select all

extern "C" void tim1_brk_isr(void)
{
   if (!DigIo::desat_in.Get() && hwRev != HW_REV1 && hwRev != HW_BLUEPILL)
      ErrorMessage::Post(ERR_DESAT);
   else if (!DigIo::emcystop_in.Get() && hwRev != HW_REV3)
      ErrorMessage::Post(ERR_EMCYSTOP);
   else if (!DigIo::mprot_in.Get() && hwRev != HW_BLUEPILL)
      ErrorMessage::Post(ERR_MPROT);
   else //if (ocur || hwRev == HW_REV1)
      ErrorMessage::Post(ERR_OVERCURRENT);

   timer_disable_irq(PWM_TIMER, TIM_DIER_BIE);
   Param::SetInt(Param::opmode, MOD_OFF);
   DigIo::err_out.Set();
   tripped = true;
}

EDIT: fixed it, @Paul do you want to try? https://github.com/jsphuebner/stm32-sin ... 5507451451

[P.S.Mangelsdorf]

Ok, let me summarize the issue and what I know:

When the Fault Occurs
I turn the key, get Wait Start, give the start signal, get Run.
I select a direction, get immediate Overcurrent.
If I put it in neutral and give another start signal, goes back into Run. Direction again gives immediate Overcurrent.
Occasionally, I hear the briefest moment of PWM whine when I select the direction (less than 1 second).
None of the phases show an error light.

Sequence of Events

1. I put together the Sport inverter, installed in the car, tested on jackstands and with a test drive.
2. Parked it in the garage for a week.
3. Went to test again and the Fault occurred.
4. Removed the inverter case, reconnected HV and 12V, tested, fault did not occur.
5. Put casing back on, test drove, tuning session, including a couple times shutting the car off and turning back on.
6. Parked in the garage for a week or so.
7. Went to test again, and the Fault occurred.
8. Removed case, reconnected HV and 12V, fault continued to occur, but none of the phases showed an error light.
9. Tried swapping out the current sensor, still faulting.
10. Tried swapping out the dissipation board, still faulting.
11. Removed the inverter from the car. Performed a diode test on each phase, all pass.
12. Tried using jumpers to swap in a known good phase and test with the inverter on the ground, Fault did not occur.
13. Used the jumpers to connect the Sport inverter, with nothing changed from when it was previously faulting, and the Fault did not occur.
14. Glued a few things in place that might have been loose.
15. Installed inverter in car.
16. Tested on jack stands, and test drove, no faults.
17. Parked for 24 hours.
18. Went to test again, and the fault occurred.

What It is Not (based on the tests I have done)
12V supply voltage or amperage
current sensor
dissipation circuit
OI board (was not doing this with the previous base/sport hybrid inverter)
At least one of the Sport phases (as it was in the base/sport hybrid)
An actual failure of any IGBTs - all phases pass a diode test on the bench.

What I Think It Could Be
Based on the fact the phases pass a diode test, don't show an error light, and do in fact work sometimes, it seems to me there is some gate driver that isn't working 100% correct. I'm not sure why time is a factor, but it seems to be.

Next Steps

• I have already bought a "for parts" sport LDU from Ebay, but I'm not picking it up until next weekend. The idea would be to harvest a good phase or two and swap them into mine, and hopefully eliminate which phase has the bad gate driver?
• I think I will have time this weekend to test a few things. Of course, there is the chance that when I remove the case it magically starts working again. I have a couple of multimeters but I do not currently have an oscilloscope (I know, I know, it's been on the list to buy for a while. Someday I'll get one)
• What points should I test, and what should I be looking for?

[johu]

Based on that I still vote for intermittent connection problem, caused by moisture or wear.
a) cracked solder joint of connector on OI board. Diagnose with microscope or by wiggling every single pin with tweezers. (I don't think a continuity test will discover this because intermittent).
b) Corroded pin or female connector
c) same as a) but on gate driver side (were you able to trace this back to a specific phase board at all?). Less likely than a)
d) ok, maybe something funky such as damaged gate driver. But not really buying this one.

[P.S.Mangelsdorf]

Based on that I still vote for intermittent connection problem, caused by moisture or wear.
a) cracked solder joint of connector on OI board. Diagnose with microscope or by wiggling every single pin with tweezers. (I don't think a continuity test will discover this because intermittent).
b) Corroded pin or female connector
c) same as a) but on gate driver side (were you able to trace this back to a specific phase board at all?). Less likely than a)
d) ok, maybe something funky such as damaged gate driver. But not really buying this one.

Thanks - I will check it over again. The thing that has me the most confused is it a) drives into the garage, then faults the next time when nothing physical should have changed. b) seems to consistently not fault the first time I test and will start back to back, but then will fault the next time I go to drive it if its been a few hours. It seems super odd. Maybe it is the dissipation circuit and I just didn't give it long enough when I tested? And the reason disassembly "fixes" the issue is some errant short/spark across the bus bars dissipates some remaining charge? (I'm not entirely sure how that whole thing works, so I don't know) But that board was in it and working fine previously.

I have not yet been able to narrow down which phase is the issue. I think I have it narrowed down to 2, because the 3rd was in it when it was a base/sport hybrid, but I'll check all of the connections again. And if it still faults when I take the case off I'll try to check which fault signal it is.