Dead SDU? or not?

[GVL]

Dear forum,

I am building a Porsche 911 classic, with a Tesla SDU using openinverter board V7, MEB batteries, Orion2 BMS, 6.6 Kw charger (OBC) and Low Voltage Power Management Unit. A few months ago I tested this setup on the bench and I got the motor to spin without any issue. This was done just with a few wires running from and to the SDU, Orion2 BMS and OBC.

The parameters used for this bench motor spin, were based on those shared by catphish, but adapted to my battery voltage. This were just a couple of brief spins, for less than 5 seconds, since there was no cooling present on the unit. It was done using the same number of batteries and voltage that is present today. Please find the bench parameters bellow:

Start CAN 1.json

(1.18 KiB) Downloaded 42 times

Now with a full harness, cooling system and all the components installed the car the SDU does not turn and I cannot figure out why. The car is wired in a way that the Negative contactor closes before the SDU is even switched on, then the SDU is switched on by turning the car key, so precharge and positive contactor close, then the start command is sent by CAN, then the gear lever is pressed to select drive direction. All these steps happen without any error until the direction is selected. At the time an OVERCURRENT error shows up.

I have selected default forward, but the OVERCURRENT error now happens when Start is done.

To exclude throttle issues at first I had the throttle configured by CAN, but now wired direct to the SDU and issue remains. Potnom seems to be working fine either way.

Start CAN 1.json

(1.18 KiB) Downloaded 42 times

For quite a while, I have been reading and learning through your problems and issues around the EV quest. Now it is time to present my own. Thank you already for the shared knowledge on your previous post, otherwise it would have not been possible to arrive here.

Please let me know your thoughts and questions.

Thanks

Gaspar

[tom91]

Take the inverter off the motor and measure between phases and HV connections to check if there is a short. If there is a short an IGBT has blown and the drive unit is dead.

The throwing of a current fault on selecting direction usually indicates this.

[cloudy]

Suspect Tom is right - but just in case - I had an issue caused by the HVIL input

For anyone trying to fix similar - I found a tombstoned capacitor (C50) on the HVIL line into the nand gate - resoldering that appears to have fixed the issue. I guess HVIL was triggering - which it seems would also cause an OVERCURRENT error on the tesla board.

[GVL]

Took the inverter out and the phases do not appear to be short.

Checked both Current sensor voltage, with HVIL switch closed both are the same:

First measurement:

5V: 5.7V
OP:2.85V
REF:0V
0V:0V

Second measurement: around 4 min later

5V: 6V
OP:3 V
REF:0V
0V:0V

Something is wrong with the 5 volt supply, any thoughts?

Thanks

[johu]

Is it from the shop?
Could be as simple as the feedback resistors. R56 should measure 56k and R58 10k

[GVL]

Measured:

R56 - 9.5k ohm

R58 - 7.9k ohm

I bought it from an EV shop, not sure they bought it from you.

Number: 2881179A_Y79-2201110

[cloudy]

Sounds like your multimeter might not be up to par, those measurements are well out of expected tolerance...

[Pete9008]

Or it's been measured while still connected up to the source?

Best to isolate the input while measuring otherwise the source impedance will mess up the measurement.

[GVL]

Just checked again both Current sensor voltage, with HVIL switch closed both are the same:

5V: 5.0 V
OP: 2.5V
REF:0V
0V:0V

Voltage after R56 resistance also read 5 volts.

I don´t know what happened yesterday to be getting those odd measurements. Today I double checked with two multimeters, both with similar readings.

Apparently these resistances are correct for this board.

Any ideas? Should I put it back on and give it another go, despite any change done or any issue found?

Thanks

[GVL]

Now measured voltages on IC5, with HV connected, HVIL switch closed, same result before and pressing start.

IC5
Pin1: 4.96V - Gate drive fault signal A - ok
Pin2: 4.96V - Gate drive fault signal B - ok
Pin3: 3.75V - comparator output - ok
Pin4: 4.96V - Gate drive fault signal C - ok
Pin5: 0V - Gate drive UVLO fault signal - ????
Pin6: 4.96V HVIL - ok

UVLO should be ok after connecting HV cables?

[cloudy]

Not sure what level the fault signal should be but please be super careful if you have HV connected and live with the inverter open. Should ideally be using HV rated gloves. Guessing you are just testing interlock without voltage but just in case!

[GVL]

I have put it back together, but this time it would not even pre-charge. Udc was 1.44 volts. I cannot understand, it was 318 yesterday and pre-charge was working.

I have done the IGTB test using a 12 volt limited power supply just to be sure with one 12 volt bulb

1. positive to DC+, negative to one phase after another
2. negative to DC-, positive to one phase after another

No flow or light on the negative to DC-, but on the positive to DC+ the bulb would light up on one of the phases.

So yes, the SDU has blown one IGTB, just I do not know why and I need to figure it out, otherwise it will happen again.

I will provide more data on this matter, so you can help me understand the reason behind this failure.

Thanks

[johu]

The shorted high side indeed does not sounds good, though it shouldn't inhibit precharge. Is the precharge resistor still ok?
Could there be a bad solder joint on the gate driver connector?

[catphish]

I do not know why and I need to figure it out, otherwise it will happen again.

I believe there are three things that cause this to happen:

1) These units are usually salvaged from unrepairable crash damaged Teslas. In a serious crash, the Tesla will open the contactors, causing a voltage spike that is very likely to cause IGBT damage.
2) Many people test on the bench using a low voltage battery and an undersized fuse. The inevitable consequence is that the drive unit attempts to draw 100A, blows the fuse, and again causes a voltage spike that can damage the IGBTs.
3) Configuring "tripmode" to something other than"Keep DC switch closed". Again, this causes the contactors to be opened when there is a fault, same result as the two scenarios above.

Of the three cars I've worked on with Tesla drive units, two of them have had the drive unit fail in this way, working fine on the bench, but then failing when HV is applied. In both cases. the drive unit was replaced like-for-like with no wiring changes, and the replacement worked perfectly.

It seems that the damage is easy to cause, but doesn't show up until HV is connected.

[GVL]

The shorted high side indeed does not sounds good, though it shouldn't inhibit precharge. Is the precharge resistor still ok?
Could there be a bad solder joint on the gate driver connector?

The prechage resistor is the Tesla OEM one, the 23 ohms one, it is still giving this resistance when tested with a multimeter.

The gate driver, you mean connector 5, correct? The board is now out, all soldering looks well.

I agree with you, no precharge and no udc measurement is strange.

[GVL]

Just a quick update, the SDU was in fact dead. Again not sure what happened, since I had it running months before with the same set up.

Installed only another Inverter with the openinverter board on the first SDU and it did not work. So installed the second complete SDU with it´s Inverter and the same openinverter board and again it did not work. I was getting an Overcurrent error, so decided to close the HVIL loop. This fixed it.