openinverter forum

Just asking,

Does anyone have a Zombieverter, GS450h combination driving on the road?

If you have, how have you managed to get round the problems listed by "crasbe" in his post of 14th December 2022?
viewtopic.php?p=49901#p49901

Is it worth using a Lexus GS450h VCU, which I understand is out and running cars in the community, as a stop gap solution to getting my BMW Z4 on the road for final "fettling" whilst the bugs are "ironed out" of the Zombieverter VCU.

I do understand that the Zombieverter is in development and that the people who are working on it do so without financial gain and I am trying to ask fair questions without causing offence.

I would just like reasonable answers to 3 questions.

Haven't seen much detail about road going builds with this setup yet other than Damien's video

https://www.youtube.com/watch?v=aG3x5pmqF8Y

If you're that far on with your Zombie setup, is it not possible to dial down some parameters in the web interface to reduce any effect of issues and allow for tame testing to get started? It might allow you to help find/test/iron out some of the issues or at least give feedback to the forum here so others can join in as it sounds you're a little further ahead than some of us.

I have both VCU types and had a simple T connection set up so could easily swap between either type. If you don't have a GS450H VCU spare, it would seem a shame to have to go buy one just to do some provisional testing. Timeline on future Zombie releases is unknown though.

Wondering the same thing myself, having been away from the project for a long time to work on other things, it seems that the same problems are outstanding:
No usable current limit, and some bug causing violent hard cut and re-application of torque
No regen
Hard cut for RPM limit as well.

It's accurate?

I now consider trying to solve these myself but from experience the Zombie software is (for me) difficult to understand and debug. And with so few others using or developing it, there isn't a "critical mass" to help make progress when stuck.
So I'm asking honestly if falling back to the arduino based GS450H (even though abandoned) could be be easier to work with in terms of software modifications and debugging, and for being more mature in the first place.

Interesting question! The Zombieverter is a marvellous project, very ambitious and wide-ranging, but clearly still in development.
I was intrigued to note that a clever chap got a 210 gearbox running with yet another option - using a Prius Gen 3 inverter with the dual-motor OpenInverter board.
Each method has different challenges, I guess.

et0 wrote: ↑Tue Jan 31, 2023 12:51 pm I now consider trying to solve these myself but from experience the Zombie software is (for me) difficult to understand and debug. And with so few others using or developing it, there isn't a "critical mass" to help make progress when stuck.
So I'm asking honestly if falling back to the arduino based GS450H (even though abandoned) could be be easier to work with in terms of software modifications and debugging, and for being more mature in the first place.

My thought would be, if there are people at the stage of being able to test the Zombie in conversions, it would be great to see them try push on and help create the 'critical mass'. The arduino based VCU is surely easier to modify for most, but if some have the ability to work on the Zombie code it would help the overall development and possibly get more involved when there are things happening?

I will be in a position in the very near future to run my car again. Just sorting out the DC/DC side so that I don't run out of low volts when driving. Once that is done I would be happy to test any software changes that have been made in a "real world" situation. If anyone would like to contact me regarding testing I would more than happy to help move this project forward.

I'm happy to test in the e65 superyacht but neither myself.of Damien have noticed either if those issues, I would assume it relates to a bad shunt value being supplied by the ivt shunt or interference or a battery setpoint that is being adhered to. If you have any specific things you want me to test let me know

Thanks for the offer paaa.

The age old problem of installation or software issues. It is very hard to say in a lot of cases which is the issue.

As far as bugs are concerned I can only quote "crasbe" as saying on 14th December 2022,

"The bug in the idc limit handling is known but as of now not yet fixed and it's known that the torque limiting due to overcurrent/overcurrent/overspeed can be too hard".

I assume that crasbe is of sufficient level in the development of the "zombieverter" to make that statement. Please can you confirm if crasbe is correct or not?

The only thing I noticed with the land yacht is that it has a 24Kwh power pack and mine is 52Kwh, perhaps the extra power available is the issue!
also the throtmax setting was at 10% on the parameters that Damien posted late last year. I could stop the Torque cut by setting my throtmax to less than 16% but the car was very sluggish.

Regarding the torque dropping off to 0, I think it is the fact that when the power comes back in it is back to max again so the transmission is going from load to no load back to load again.
Attachments
image.png
image.png (35.21 KiB) Viewed 3 times

Just wanted to add my thoughts here:

I believe this issue exists, but after reading the code I don't believe it's a bug as such. Most likely at high current levels, interference is causing abnormal values to be read from the current sensor, which is triggers the idcmax protection.

I see two solutions here. Firstly, some sanity checking on the current sensor values. I believe The IVT can report up to 2 million amps. We can can therefore simply ignore the majority of those possible values as nonsense.

Secondly, I am in the process of writing a smoother derating algorithm for openinverter. If it works, I assume this code can be pulled into zombie.

In the meantime, as an end user, you're going to have to make the decision yourself whether DC current derating is necessary for your setup, and if not, you can simply remove it for now. With manual tuning, it's not really hard to configure torque limits so that DC current always stays within your desired limit.

https://github.com/damienmaguire/Stm32- ... /tag/1.11A Damien cut a new release with fix for DC current limit.

Yes, saw the update on github and already tried this version 1.11A-

"Fix the non working idcmax param caused by move to float calcs in throttle.".

So it was a bug, and was at least partly related to the line I identified in utils.c, which is now
Throttle::IdcLimitCommand(finalSpnt, ABS(Param::GetFloat(Param::idc)));

And I confirm the torque drops are not there any more (tested with wheels off the ground).

So that's good, I shouldn't have spent all yesterday trying to fix it


Question: how likely is regen to start working - I guess I am asking how wise is it to spend time on it myself?

Damien has covered regen in a video coming out at the weekend. In short, it's not going to coming yet, unless someone else works on it.

I'll post the video here when it's publicly available.

Many thanks to Damien for the fix (and for all the previous work of course)

I guess I was entirely wrong! Thanks Damien for fixing.

I have not had a chance to view the video as yet. But just a quick post to say that I have just driven my car on the road and I can honestly say the difference is unbelievable. It pulls well and there is a proper rev limit control with a gentle drop in power and then back on power. Thanks to everyone on the forum who have helped me get to this point.

Again, great that this regression is finally fixed! Thank you Damien.

I still experience what seems like a hard cut on idcmax - causes hard juddering once you reach the set limit, totally smooth if you stay below it. On GS450H, 1.11.A. Is this to be expected, or am I on the wrong version or wrong setting?

Example behaviour with idcmax set to 25A - throttle held steady as speed increases.

FWIW we finally swapped out to the GS450H Inverter Controller board, and it's been night and day in terms of understanding how to set it up, adding functionality and the development cycle. For my skillset anyway!

Already have working regen, a throttle map, current limit with PID control, safety features, CAN interfacing to other components, a useful serial diagnostic output.. Lots more to add but with just a few days hacking the vehicle starts to behave how it should.

So, thank you again to Damien for that project, and Catphish for the refactored code!

Just to clarify, have you replaced the Zombieverter with the GS450h Inverter controller?

Just wanted to ask if you swapped-out a "bad" Toyota board within the inverter? Or if you swapped the Toyota part for an open inverter board? Or if you swapped the Zombie controller board with???

Cheers

et0 wrote: ↑Thu Mar 30, 2023 12:33 pm FWIW we finally swapped out to the GS450H Inverter Controller board, and it's been night and day in terms of understanding how to set it up, adding functionality and the development cycle. For my skillset anyway!

Already have working regen, a throttle map, current limit with PID control, safety features, CAN interfacing to other components, a useful serial diagnostic output.. Lots more to add but with just a few days hacking the vehicle starts to behave how it should.

So, thank you again to Damien for that project, and Catphish for the refactored code!

dadiowe wrote: ↑Thu Mar 30, 2023 3:15 pm Just to clarify, have you replaced the Zombieverter with the GS450h Inverter controller?

DkubusEV wrote: ↑Thu Mar 30, 2023 10:11 pm Just wanted to ask if you swapped-out a "bad" Toyota board within the inverter? Or if you swapped the Toyota part for an open inverter board? Or if you swapped the Zombie controller board with???

He referenced my code, so I assume he's using the GS450H VCU with custom code based on mine.
https://github.com/damienmaguire/Lexus- ... Controller
https://github.com/catphish/gs450h/tree ... /gs450h-v8

While it isn't nearly as flexible as the Zombie, the older VCU code is very easy to work on because it's just a small Arduino program.

Bummer, I only have the zombie verter controller I bought fully built from Damien. As it seemed like it was supposed to be the "better" (ie "supported") system going forward. So I guess I'll need to be very patient as I have no clue how to code... either that or I buy a old version to use until the zombie catches up???

DkubusEV wrote: ↑Fri Mar 31, 2023 8:56 am Bummer, I only have the zombie verter controller I bought fully built from Damien. As it seemed like it was supposed to be the "better" (ie "supported") system going forward. So I guess I'll need to be very patient as I have no clue how to code... either that or I buy a old version to use until the zombie catches up???

The Zombieverter works fine with the GS450H, I'd usually recommend sticking with it. As far as I know, the only thing it's lacking now is regen. I'm not certain why it was removed, but hopefully it will reappear soon enough.

catphish wrote: ↑Fri Mar 31, 2023 9:03 am The Zombieverter works fine with the GS450H, I'd usually recommend sticking with it. As far as I know, the only thing it's lacking now is regen. I'm not certain why it was removed, but hopefully it will reappear soon enough.

Regen was removed because the throttle code was rewritten. Since at the time the GS450h testbed of Damien was not ready yet and the time he is able to devote to debugging regen is still limited, it is disabled for the time being.

So far nobody stepped forward to write and test an implementation, so for the immediate future it'll stay without regen I guess.

crasbe wrote: ↑Fri Mar 31, 2023 9:10 am Regen was removed because the throttle code was rewritten. Since at the time the GS450h testbed of Damien was not ready yet and the time he is able to devote to debugging regen is still limited, it is disabled for the time being.

So far nobody stepped forward to write and test an implementation, so for the immediate future it'll stay without regen I guess.

Very glad to learn as to why regen was removed, seems very logical, I wish I was able to do coding for such things so I could contribute to the zombie situation but unfortunately I'll need to remain patient on the sidelines. Cheers for the info.