ZombieVerter VCU Support

dadiowe
Posts: 109
Joined: Fri Sep 17, 2021 6:11 pm
Has thanked: 41 times
Been thanked: 19 times

Re: ZombieVerter VCU Support

Post by dadiowe »

Jack Bauer wrote: Mon Nov 14, 2022 1:12 pm Just pushed an update to the repo :

https://github.com/damienmaguire/Stm32- ... s/tag/1.01

-Update the reported version to 1.01.
-Add use of PWMs as digital o/p. PWM1 use as digital charger enable / disable. PWM2 as digital "Run" mode indication.
-Verify GS450h driving in vehicle

The big ticket item on this release is the Lexus GS450H system control is now verified working in my E39 conversion at full voltage and road speeds. A video will be publicly available in a few days to prove this.
Great job dDmien just in time. I recieved your suggested chip from China to repair my self fried Zombieverter on Monday, unbelievable service as I ordered on the 8th.

I now have a repowered Zombie. Just asking if you could publish your parameters for the GS450H just to make sure we are testing the same settings.

I did a road test but unfortunately lost my plot of torque and speed just after one of my gearbox mounts let go. It looked to me that light throttle was great but giving a little more caused a sudden drop in torque to nothing. Its almost like the inverter said you want how much? no way!

To sum up the plot looked like 2 fairly smooth curves until I pressed the throttle hard. As you have a driveable car and I have the latest software, 1.0.1A, that maybe my parameters that I have set are not right.

Once I repair the mount I will try and get a better plot.
User avatar
Jack Bauer
Posts: 3563
Joined: Wed Dec 12, 2018 5:24 pm
Location: Ireland
Has thanked: 1 time
Been thanked: 87 times
Contact:

Re: ZombieVerter VCU Support

Post by Jack Bauer »

I have created an open task to better focus the community efforts :
viewtopic.php?p=48640#p48640
I'm going to need a hacksaw
et0
Posts: 127
Joined: Sun Oct 13, 2019 8:06 pm
Location: Scotland
Has thanked: 14 times
Been thanked: 5 times

Re: ZombieVerter VCU Support

Post by et0 »

Hi Damien,
I will try to debug this today, obviously it’s working for you so I want to start with checking if there’s any difference in code or settings: would you mind sharing the settings you’re using, and confirm that you’re using the same .hex file that you released?

Also it looked like when you first tried the GS450H code, you experienced the problem? After that you posted a video with it fixed, and now can’t reproduce it.

Do you know roughly what the change was that affected it?

Sorry for the annoying questions but it’s to avoid getting stuck for a long time on something that might already be solved.
EvSteeve
Posts: 34
Joined: Tue Sep 07, 2021 10:01 am
Has thanked: 2 times
Been thanked: 1 time

Re: ZombieVerter VCU Support

Post by EvSteeve »

Jack Bauer wrote: Thu Oct 27, 2022 1:48 pm Brilliant. Another bug for me to find.
Any progress on debugging the latest firmware?

Regards
et0
Posts: 127
Joined: Sun Oct 13, 2019 8:06 pm
Location: Scotland
Has thanked: 14 times
Been thanked: 5 times

Re: ZombieVerter VCU Support

Post by et0 »

Test with the current DM_WORK branch
Screenshot 2022-11-16 103416.png

Torque is actually going completely to zero. I'll try removing that "torque cut" code and see it makes a difference.

With the old code (from last year) put back
Screenshot 2022-11-16 104131.png
User avatar
VWJoe
Posts: 55
Joined: Wed Feb 16, 2022 8:39 am
Has thanked: 7 times
Been thanked: 9 times

Re: ZombieVerter VCU Support

Post by VWJoe »

Uploaded the new version 1.01 (Sept 22)

Just started to put items together, but had trouble initiating the IVT Shunt. It's connected CANBUS 2 (pins 27,28). Found I had to change the CANBUS to CANBUS 1 in the interface to get the VCU to talk to the Shunt. Check my wiring which is correct to the diagram on Wiki, so think the CAN1 and CAN2 in the interface are the wrong way round.
User avatar
Jack Bauer
Posts: 3563
Joined: Wed Dec 12, 2018 5:24 pm
Location: Ireland
Has thanked: 1 time
Been thanked: 87 times
Contact:

Re: ZombieVerter VCU Support

Post by Jack Bauer »

et0 wrote: Wed Nov 16, 2022 9:01 am Hi Damien,
I will try to debug this today, obviously it’s working for you so I want to start with checking if there’s any difference in code or settings: would you mind sharing the settings you’re using, and confirm that you’re using the same .hex file that you released?

Also it looked like when you first tried the GS450H code, you experienced the problem? After that you posted a video with it fixed, and now can’t reproduce it.

Do you know roughly what the change was that affected it?

Sorry for the annoying questions but it’s to avoid getting stuck for a long time on something that might already be solved.
I am using the very same code. My problem was down to the E39 stability control applying the rear brakes to stop wheel spin. Pulling the abs fuse "solved" this. Now that the car is driving on all 4 wheels its not a problem.
I'm going to need a hacksaw
User avatar
Jack Bauer
Posts: 3563
Joined: Wed Dec 12, 2018 5:24 pm
Location: Ireland
Has thanked: 1 time
Been thanked: 87 times
Contact:

Re: ZombieVerter VCU Support

Post by Jack Bauer »

et0 wrote: Wed Nov 16, 2022 10:44 am Test with the current DM_WORK branch

Screenshot 2022-11-16 103416.png


Torque is actually going completely to zero. I'll try removing that "torque cut" code and see it makes a difference.

With the old code (from last year) put back

Screenshot 2022-11-16 104131.png
Now that is an actual software bug. My torque never went to zero as you will see in the plot I posted previously. So something is telling it to drop torque. from the top of my head :udcmin being hit, rpm limiter,brake in signal.
I'm going to need a hacksaw
User avatar
Jack Bauer
Posts: 3563
Joined: Wed Dec 12, 2018 5:24 pm
Location: Ireland
Has thanked: 1 time
Been thanked: 87 times
Contact:

Re: ZombieVerter VCU Support

Post by Jack Bauer »

EvSteeve wrote: Wed Nov 16, 2022 10:34 am Any progress on debugging the latest firmware?

Regards
Really!!!
I'm going to need a hacksaw
User avatar
Jack Bauer
Posts: 3563
Joined: Wed Dec 12, 2018 5:24 pm
Location: Ireland
Has thanked: 1 time
Been thanked: 87 times
Contact:

Re: ZombieVerter VCU Support

Post by Jack Bauer »

VWJoe wrote: Wed Nov 16, 2022 11:18 am Uploaded the new version 1.01 (Sept 22)

Just started to put items together, but had trouble initiating the IVT Shunt. It's connected CANBUS 2 (pins 27,28). Found I had to change the CANBUS to CANBUS 1 in the interface to get the VCU to talk to the Shunt. Check my wiring which is correct to the diagram on Wiki, so think the CAN1 and CAN2 in the interface are the wrong way round.
Can we correct the wiki on this please.
I'm going to need a hacksaw
et0
Posts: 127
Joined: Sun Oct 13, 2019 8:06 pm
Location: Scotland
Has thanked: 14 times
Been thanked: 5 times

Re: ZombieVerter VCU Support

Post by et0 »

Jack Bauer wrote: Wed Nov 16, 2022 11:24 am Now that is an actual software bug. My torque never went to zero as you will see in the plot I posted previously. So something is telling it to drop torque. from the top of my head :udcmin being hit, rpm limiter,brake in signal.
Right, potnom is going to zero as well.

In the utils, processthrottle function, the limits are applied.
I've tried commenting these out in turn so that finalSpnt is just the raw value without any processing. However when I comment out UDClimitcommand, I just get immediate full power when putting it into drive. Don't understand that.

edit: commenting out IDClimitommand does seem to get rid of the problem. Have you changed the ISA setup by any chance?
et0
Posts: 127
Joined: Sun Oct 13, 2019 8:06 pm
Location: Scotland
Has thanked: 14 times
Been thanked: 5 times

Re: ZombieVerter VCU Support

Post by et0 »

Jack Bauer wrote: Wed Nov 16, 2022 11:24 am Can we correct the wiki on this please.
Wait, what's the request? CAN2, pins 27-28 goes to the shunt on the schematic (my car is wired like that), has that changed now? Is that the cause of this problem?
nkiernan
Posts: 505
Joined: Mon Feb 24, 2020 8:59 pm
Location: Ireland
Has thanked: 276 times
Been thanked: 64 times

Re: ZombieVerter VCU Support

Post by nkiernan »

Update on my previous 'GS450H Inverter not running' posts!

As mentioned before, I had an older GS450H VCU that ran the GS450H inverter/drive setup as expected. I have two Zombieverters (one a built and bench tested unit) that would go into run mode but I couldn't get any inverter whine, Invstat stayed Off and trans wouldn't turn. As the GS450H kit worked with the old VCU, the assumption was the trans and inverter were good (lesson 1!). And the Zombie's were proven good, so the issue logically had to be with the wiring.

Well I wired and rewired many times, checked and double checked many times, but no joy. Very kindly Damien offered to take a closer look as the issue didn't make sense, so I sent the two Zombies and the inverter his way. Luckily the Land Yacht was in the shed and available to swap these parts in to test.

What Damien found:
- Both VCU's passed the bench test with the GS300H trans
- Both VCU's ran the GS450H system in the E39 with no issues
- Swap in the non-working inverter and no whine and no spin
- Scoped the clk signal on the good inverter and it looked nice and clean compared to a broken signal on the non-working inverter
- Scoped clk from the mcu side of the can transciever and all good
- Scope on output side and has breaks
- Measure approx 60R between pairs
- Measure resistance between sets. On non-working inverter get about 120k between clk and all other pairs, and infinite on good inverter

Conclusion, for some reason non-working inverter is leaking clk signal onto the comms wires and vice versa.

But the non-working inverter works on the older VCU, so Damien had a look through the code. The older vcu does not care so much about the return data on MTH. Its just for information, BUT the Zombie uses it to decide if comms have been established between inverter and VCU and wont enter the next step unless it sees valid data. On the non-working inverter it never gets there so no whine etc.

The bottom section of the inverter was removed, and so the real source of the problem becomes apparent. Photos below show the corrosion around the header pins on the PCB and on the opposite side of the PCB.

Interestingly (not really), I got my hands on a second inverter and appear to have the exact same issue again...what are the chances! On the hunt for a third inverter or replacement PCB again, third time lucky!


Inv008.JPG
Inv009.JPG
Inv010.JPG
Inv011.JPG
dadiowe
Posts: 109
Joined: Fri Sep 17, 2021 6:11 pm
Has thanked: 41 times
Been thanked: 19 times

Re: ZombieVerter VCU Support

Post by dadiowe »

Jack Bauer wrote: Wed Nov 16, 2022 11:24 am Now that is an actual software bug. My torque never went to zero as you will see in the plot I posted previously. So something is telling it to drop torque. from the top of my head :udcmin being hit, rpm limiter,brake in signal.
Just got my system running on axle stands and I am getting the Torque drop as well.
Any chance of Damien posting his parameter settings, please.
User avatar
AdrianB
Posts: 63
Joined: Sun Jan 24, 2021 10:34 pm
Location: Wales. UK.
Has thanked: 20 times
Been thanked: 20 times

Re: ZombieVerter VCU Support

Post by AdrianB »

et0 wrote: Wed Nov 16, 2022 1:01 pm Wait, what's the request? CAN2, pins 27-28 goes to the shunt on the schematic (my car is wired like that), has that changed now? Is that the cause of this problem?
Also in second video on the Wiki (see https://m.youtube.com/watch?t=660s&v=MU ... e=youtu.be) it says CAN2, pins 27-28 connect to the shunt.
User avatar
mjc506
Posts: 343
Joined: Wed Sep 09, 2020 9:36 pm
Location: Wales, United Kingdom
Has thanked: 30 times
Been thanked: 28 times

Re: ZombieVerter VCU Support

Post by mjc506 »

nkiernan wrote: Wed Nov 16, 2022 3:13 pm Update on my previous 'GS450H Inverter not running' posts!...
Nice, good debugging to find that (far from obvious!) problem :)

EvSteeve wrote: Wed Nov 16, 2022 10:34 am Any progress on debugging the latest firmware?

Regards
I sincerely hope that's a joke... :roll:
User avatar
VWJoe
Posts: 55
Joined: Wed Feb 16, 2022 8:39 am
Has thanked: 7 times
Been thanked: 9 times

Re: ZombieVerter VCU Support

Post by VWJoe »

AdrianB wrote: Wed Nov 16, 2022 4:00 pm Also in second video on the Wiki (see https://m.youtube.com/watch?t=660s&v=MU ... e=youtu.be) it says CAN2, pins 27-28 connect to the shunt.
My shunt is connected to CAN2 (pin 27,28) as per the video and wiki, but I've had to select CAN1 in V1.01 to get in to communicate on pins 27,28 (CAN2). Not a problem once I worked out want was going on. Maybe the CAN numbers have been swapped in the latest version.
User avatar
Jack Bauer
Posts: 3563
Joined: Wed Dec 12, 2018 5:24 pm
Location: Ireland
Has thanked: 1 time
Been thanked: 87 times
Contact:

Re: ZombieVerter VCU Support

Post by Jack Bauer »

Parameter file attached from the E39.
Attachments
E39_450h_params.json
(1.08 KiB) Downloaded 62 times
I'm going to need a hacksaw
dadiowe
Posts: 109
Joined: Fri Sep 17, 2021 6:11 pm
Has thanked: 41 times
Been thanked: 19 times

Re: ZombieVerter VCU Support

Post by dadiowe »

VWJoe wrote: Wed Nov 16, 2022 7:25 pm My shunt is connected to CAN2 (pin 27,28) as per the video and wiki, but I've had to select CAN1 in V1.01 to get in to communicate on pins 27,28 (CAN2). Not a problem once I worked out want was going on. Maybe the CAN numbers have been swapped in the latest version.
I have just checked my Can wiring to the ISA Shunt and the vehicle Can connections are Can1, pins 43 and 44, go to Vehicle Can and Can2, pins 27 and 28, go to my ISA shunt. I have Can1 ISA and Can2 Vehicle set in the parameters. I have downloaded my parameters file and it is saying Shunt is Can0 and Vehicle is Can1.

Don't know if that helps or not just thought I would post my findings.
et0
Posts: 127
Joined: Sun Oct 13, 2019 8:06 pm
Location: Scotland
Has thanked: 14 times
Been thanked: 5 times

Re: ZombieVerter VCU Support

Post by et0 »

I agree, the CAN interfaces as set/shown on the GUI are backwards.
I have the shunt wired to CAN2 as it always has been, and with this latest software, the GUI shows that it's connected to CAN1, however it clearly is getting data from the shunt on CAN2.

So it needs correcting in the SW but I think it's a red herring as far as the "torque drop" problem goes.

As I said yesterday, getting rid of the throttle::IdcLimitCommand(finalSpnt, Param::Get(Param::idc)); call removes the symptom, and the torque application is smooth again.

There is nothing obviously wrong with the Idc value from the shunt when plotted though.

This is the function in question:

Code: Select all

void Throttle::IdcLimitCommand(float& finalSpnt, float idc)
{
   static float idcFiltered = 0;

   idcFiltered = IIRFILTERF(idcFiltered, idc, 4);

   if (finalSpnt >= 0)
   {
      float idcerr = idcmax - idcFiltered;
      float res = idcerr * 5;

      res = MAX(0, res);
      finalSpnt = MIN(res, finalSpnt);
   }
   else
   {
      float idcerr = idcmin - idcFiltered;
      float res = idcerr * 5;

      res = MIN(0, res);
      finalSpnt = MAX(res, finalSpnt);
   }
}
I'll try to debug it further. It's not easy not being able to just print out a parameter and seeing what happens though.. And I did get unexpected results from an experimental change yesterday, so it might not be a straightforward problem.
User avatar
Jack Bauer
Posts: 3563
Joined: Wed Dec 12, 2018 5:24 pm
Location: Ireland
Has thanked: 1 time
Been thanked: 87 times
Contact:

Re: ZombieVerter VCU Support

Post by Jack Bauer »

Interesting. That routine should only kick in if idc>idcmax or idc<idcmin. Assuming you have set those values correctly then my guess would be some number is wrapping around itself possibly due to the change from fixed point to floats as that routine has not been modified itself. as a simple test, set idc at a very low value say 5amps and see if it works as it should then increase until it fails. might give us a pointer.
I'm going to need a hacksaw
et0
Posts: 127
Joined: Sun Oct 13, 2019 8:06 pm
Location: Scotland
Has thanked: 14 times
Been thanked: 5 times

Re: ZombieVerter VCU Support

Post by et0 »

You mean like

Throttle::IdcLimitCommand(finalSpnt, 5);

?


Never mind, I came at it another way.
-> Changed the code back to original
-> checked the problem still exists
-> changed the idcmax from 300 to 3000

Problem is gone (on the ramp at least).
So it seems to be a scaling issue, maybe just a factor of 10.


I see you have your IDCmax set to 5000, so that ties in.
dadiowe
Posts: 109
Joined: Fri Sep 17, 2021 6:11 pm
Has thanked: 41 times
Been thanked: 19 times

Re: ZombieVerter VCU Support

Post by dadiowe »

et0 wrote: Tue Aug 09, 2022 11:30 am Something a bit odd with the contactor control on my setup. After a few seconds, one of them starts to buzz, or even chatter and drop out. I added bulbs to each coil to see what's happening. The behavior improves with the bulbs wired in!

Any idea what's going on here? As far as I know none of these contactors have an economizer built in.

In the video you can see the start sequence, then bulbs dimming in turn (why?), then the buzzing sound from one or more coils). I didn't look with a scope yet.

Thanks

Just been doing some testing on axle stands. First run of the day and I was trying to plot the Torque drop issue after changing a few parameters.

The usual torque drop happened and the torque plot dropped to zero. I took my foot off the accelerator and then tried to re apply a small amount of throttle. It seemed that the torque drop was much worse coming in and out at much lower throttle application.

Suddenly big smokey from the battery area. Powered down everything and found that the Precharge resistor, Arcol 30Ω 100W Aluminium Chassis Mount Resistor HS100 30R F ±1% had self destructed.

I can only conclude that the main contactor had disconnected at the first torque drop and because the Precharge contactor remains powered (not sure why) until the ignition is switched off, it then, was carrying the full load that the inverter was asking for.

Ironically I have just seen Damiens parameters and they show that he has throtmax set at 10, regentravel at 0, regenmax at 0 and throtramp at 1.
Mine were all set at default except throtramp which was 0.09 which I believe is the minimum.

I will have a new precharge resistor tomorrow so I will report back on Damiens settings on my car.

Could I ask if anyone knows why the precharge contactor remains energised. I recall that with the Leaf setup it disconnects after about 5 seconds.
dadiowe
Posts: 109
Joined: Fri Sep 17, 2021 6:11 pm
Has thanked: 41 times
Been thanked: 19 times

Re: ZombieVerter VCU Support

Post by dadiowe »

et0 wrote: Thu Nov 17, 2022 11:55 am You mean like

Throttle::IdcLimitCommand(finalSpnt, 5);

?


Never mind, I came at it another way.
-> Changed the code back to original
-> checked the problem still exists
-> changed the idcmax from 300 to 3000

Problem is gone (on the ramp at least).
So it seems to be a scaling issue, maybe just a factor of 10.

I see you have your IDCmax set to 5000, so that ties in.
I have tried IDCmax at 500 and 5000 no difference on the road for me.
Interesting you mention a factor of 10 as Damiens Throtmax is 10 rather than 100 for me.
et0
Posts: 127
Joined: Sun Oct 13, 2019 8:06 pm
Location: Scotland
Has thanked: 14 times
Been thanked: 5 times

Re: ZombieVerter VCU Support

Post by et0 »

dadiowe wrote: Thu Nov 17, 2022 12:10 pm Just been doing some testing on axle stands. First run of the day and I was trying to plot the Torque drop issue after changing a few parameters.

The usual torque drop happened and the torque plot dropped to zero. I took my foot off the accelerator and then tried to re apply a small amount of throttle. It seemed that the torque drop was much worse coming in and out at much lower throttle application.

Suddenly big smokey from the battery area. Powered down everything and found that the Precharge resistor, Arcol 30Ω 100W Aluminium Chassis Mount Resistor HS100 30R F ±1% had self destructed.

I can only conclude that the main contactor had disconnected at the first torque drop and because the Precharge contactor remains powered (not sure why) until the ignition is switched off, it then, was carrying the full load that the inverter was asking for.

Ironically I have just seen Damiens parameters and they show that he has throtmax set at 10, regentravel at 0, regenmax at 0 and throtramp at 1.
Mine were all set at default except throtramp which was 0.09 which I believe is the minimum.

I will have a new precharge resistor tomorrow so I will report back on Damiens settings on my car.

Could I ask if anyone knows why the precharge contactor remains energised. I recall that with the Leaf setup it disconnects after about 5 seconds.
IIRC it's to do with protecting the inverter against blowing up if the contactor opens unexpectedly.

I have *not* seen a contactor open during my testing, since I fixed the issue from the post above, which was simply that the power drivers in the VCU are not rated to drive the contactor coils directly.
You might have the same issue or perhaps even just a loose wire?
Locked