openinverter forum

Hello,

I am having a problem where my GS450h transmission is causing a jerking motion, shaking on the vehicle. Video:
https://photos.app.goo.gl/7tgTdY3crAGy2M1d7

I have installed the transmission on my Scion FRS with a preliminary battery (~100V) and taken it for a quick spin. I am using the Zombieverter. As far as I know, the inverter and transmission are from the exact same vehicle, so the resolver calibration should match the inverter. The transmission is currently hard mounted to the vehicle chassis and I'm wondering if this could cause such vibrations. I have tried different settings for the Zombieverter without much improvement. Only reducing the throttle span (0-10% over the throttle span) helped a bit.

Have you ever seen such a thing?
(sorry if it was asked before, I searched for a while in the forum and couldn't find anything)

when I built my first EV (lead acid, DC motor mounted to tranny) I had a similar issue if I tried to start off in second gear. It never happened when I started out in first gear. My guess is it has something to do with the alignment of your drivetrain? I never figured my issue out and I haven't driven my GS450h project yet, so that's not much help.

While its very difficult to diagnose something like this over a forum my guess would be the low battery voltage as the culprit. I see a similar effect on bench testing with low voltages if I demand high throttle values. think about it this way : while the inverter does not care what voltage you supply it cannot make ever increasing amounts of torque unless the voltage is available to counter the back emf from the motors. Remember also you have two motors geared differently that will exacerbate the problem. Does it occur in reverse also? In reverse only mg2 produces torque.

Also worth ensuring your resolver and comms cables are correctly wired and screened and you bond the gearbox and inverter to vehicle chassis.

Hardmounting turned your entire floor into an acoustic amplifier. First step - rubber isolators.

Second step - you're running at least half the nominal battery voltage you should be....you don't get to randomly pick a voltage that happens to be able to spin the motor on a bench; it needs to be close to nominal (is it 360V? in a Lexus). This not only has back emf implications, but you've effectively quartered the rise time of current in the stator...that will produce heating and massive harmonics which then play through your homebrew speaker membrane. I suggest you find a way to get up to the nominal Lexus battery voltage or you sell the Lexus CVT and go with a brushed DC motor that's happy at 100V.

The rubber isolators are moot if you don't jack up the voltage, if that's not obvious.

Thanks for all the feedback. Just to reiterate, this is a preliminary setup, not a definitive one. I am obviously planning on using a 350V nominal battery and using rubber mounts. The idea of this setup is to iron out any problems beforehand and make sure I have a proper architecture.

I have tried to see if reverse is behaving differently and indeed it does. When in reverse, the oscillation goes away.

What I will try in the next week:
- Ground the shield wires for the communication between the Zombieverter and the inverter (cable is shielded but shield is not grounded properly)
- Ground the shield wires for the communication between the inverter and the resolvers (cable is shielded but shield is not grounded properly)
- Increase the voltage to >300V
- Upgrade Zombieverter to latest code (I am using a ~4-6 month old version)
- Modify code to disable MG1 while going forward and see if this fixes the issue (not as a definitive solution but to narrow down on the problem)

I will keep you posted.

But if your setup may be creating nonexistent problems, why are you chasing them?

remy_martian wrote: ↑Fri Sep 09, 2022 7:45 pm But if your setup may be creating nonexistent problems, why are you chasing them?

The original idea was to quickly test to see if the setup considered would work. But your question is full of wisedom, I will reconsider spending time on a temporary setup chasing potential "non problems".

Ok so I have tried everything I said I would. Here are the results:

Ground the shield wires for the communication between the Zombieverter and the inverter (cable was shielded but shield was not grounded properly)

No change, still jerky

Ground the shield wires for the communication between the inverter and the resolvers (cable was shielded but shield was not grounded properly)

No change, still jerky

Increase the voltage to >300V

No change, still jerky

Upgrade Zombieverter to latest code (I am using a ~4-6 month old version)

Latest version is not working for me. Oil pump is not running, inverter communication does not seem to work either (i.e. INVudc is reported as 0V).

Modify code to disable MG1 while going forward and see if this fixes the issue (not as a definitive solution but to narrow down on the problem)

I went back to version b9c8f19. I changed the code to run on MG2 only. This fixed the problem (forward and reverse). I then changed the code to run on MG1 only. This also worked well (forward and reverse).

So it seems that only using one electric motor in the transmission solves the problem I was seeing. I am noticing that MG1 torque is set to be 5/4 of MG2 torque in the VCU code and the Zombieverter. I am sure this was well tough of but could this ratio be the culprit?

Anyhow, I will also be taking remy_martian advice and stop chasing this problem for now until I get a more robust setup.

Would it be possible to share the .bin file for the modified code to run MG2? Also doing some zombie troubleshooting

Thank you for testing. In terms of the release code not working can you give any more details? This was extensively bench tested over the past few weeks AND tested in a car before release. I didn't just dump it out there and cross my fingers:) Not accusing anyone of anything just want to get to the bottom of a potential deep bug if its not showing for me.

nkiernan wrote: ↑Fri Sep 16, 2022 12:59 pm Would it be possible to share the .bin file for the modified code to run MG2? Also doing some zombie troubleshooting

Yes, file is attached (b9c8f19 with MG1 disabled)

Jack Bauer wrote: ↑Fri Sep 16, 2022 2:14 pm Thank you for testing. In terms of the release code not working can you give any more details?

Unfortunately, I cannot give much feedback other than b9c8f19 is running fine for me and latest was not. I was able to get access to the web interface and was able to read values from it. However, oil pump PWM could not be modified (and pump was not running at all) and inverter seemed unresponsive (no high pitch noise when going into forward or reverse and no information retrieved from inverter through synchronous serial IF - like INVudc is reported as 0V although HV is present).

Let me know if I can do anything to get more information that could help.

gfournier wrote: ↑Sun Sep 25, 2022 6:19 pm Yes, file is attached (b9c8f19 with MG1 disabled)

Thank you

This is highly concerning for me in a number of ways. As I stated previously and demonstrated in video, the release code works on bench and was tested in a vehicle. Now I'm told it just sits there doing nothing. So either A) there is an as yet undiscovered bug in the release firmware which I am unable to find with my test setups so far or B) ?

As seems to be the way on this forum those willing to help are the least able and those able to help the least willing. No matter. I'm installing a 450h into my E39 presently and the problem or problems will be found, rectified and published by me.

Possibly I fall into the 'least able' to help category, so last week I ordered a second built and bench tested Zombie and looking forward to getting it soon. Knowing and being able to prove my GS450H setup is functioning (runs with the GS450H VCU, but InvUDC = 0V and no spinning with my current Zombie), I would expect the new Zombie to work as intended (if my issue relates to something I did with current Zombie hardware for example). If not, it narrows the search. Will report back ASAP

All I can say is that the latest firmware release works for forward and reverse on my GS450h with about 260 volts. Maybe light throttle control is not to smooth but I am sure that is minor settings. Also oil pump is fine the pwm is adjustable and at 30 percent pwm fairly quiet.

Jack Bauer wrote: ↑Mon Sep 26, 2022 7:58 am As seems to be the way on this forum those willing to help are the least able and those able to help the least willing. No matter. I'm installing a 450h into my E39 presently and the problem or problems will be found, rectified and published by me.

I do have a setup on hand that is not working so this is a good starting point. I am also willing to help. And although I am new to this code, I have my share of experience in C and C++. If you have any pointers that I could look at to get me started in debugging, I'd be willing to give a hand.

I don't know if you have a setup to trace code step-by-step in a debugger but I do not. I can compile and upload using the web interface. How are you setup for debugging?

I haven't experienced this myself but one of the reasons Toyota/Lexus hybrids become jerky in electric mode (according to John Kelly of Weber) is if one or more of the stator windings has shorted or partially shorted. Just to eliminate this possibility you could try checking the resistance of each leg...just a thought.

Hi
Did you get the jerking solved in the end?
Regards

Hi,
I don't know if you have tried reducing the throttmax setting in the software. If I set this at say 30 the car is more driveable for me anyway. Latest version of the software is the one I am using.
Good luck