2.20 with Rear outlander Topic is solved

[LRBen]

Finally got around to updating to 2.20 on the MG over the weekend and found that there is a bit of judder compared to 2.17. I've used 2.20 on a leaf setup recently and that was smooth as always.
I've done a little back to back comparisons this afternoon so I will throw them up here and update as we figure it out.

Here are a couple videos of the behavior. It is most prominent at low speed and take off. But the little torque drops can be felt at high speeds as well.
Forgive the state of the interior. It's a workhorse and I treat it poorly. Also the display isn't setup to show current yet so ignore that.
2.20
https://photos.app.goo.gl/fkT9V1tCdFYqifd47

2.17
https://photos.app.goo.gl/NxqozniekHhwvFiE8

Attached are the params. I've also testing a modified 2.20 firmware where I reversed reverted the Outlander heartbeat changes so it was being sent through the Rear outlander modules as per 2.17. This didn't make any difference to the behavior.

So far I have only graphed pot nom which looked the same on both versions, no signs of that dropping. Will try to take some can logs soon.

Edit: Overview from thread. So from 2.20 throtramp needs to be turned down lower than 10 with Outlander rear motors to avoid any juddering. Also the charger should now be powered cycled between each run/charge mode to ensure the DC-DC works. But this is best practice anyway.

[tom91]

Please try a few things like:
1. turn off regen to see if it reduces or increases the judder
2. move the UDC limits to see if it reduces or increases the judder
3. move thermal limit to see if it reduces or increases the judder

[LRBen]

Done some more experiments and played with the settings, regen, udc and thermal limits made no difference.
But throtramp did (credit to RStevens for suggesting I try this), it was at 10, the higher it went the worse it gone. I've got that down to 2 now and it's very smooth again. I guess it was hitting some limit that gets noticed from 2.20 onwards?

As an additional find I found that the DC-DC was not running when in run mode on 2.20. It works in charge mode, went back to 2.17, works in both modes. On my setup this is activated by the dcsw_out pin as I'm using Sbox for contactor control. However it did work when using the modified 2.20 firmware I made that moves the heartbeat message back to the Outlander module when in run mode. This modification is only when in run mode, it still uses the heartbeat module for charge mode.

[tom91]

DC-DC was not running when in run mode

Okay this is weird, please provide logs or a GitHub variant of what you are running. Possibly there is a slight issue with the CAN message. Strange thing is Jamie, BigPie, seems to not have this issue.

[LRBen]

Okay this is weird, please provide logs or a GitHub variant of what you are running. Possibly there is a slight issue with the CAN message. Strange thing is Jamie, BigPie, seems to not have this issue.

https://github.com/SomersetEV/Stm32-vcu ... rtbeat-mod

I will do some back and forths between the firmware versions and see if it happens each time. Just waiting on the laptop to charge back up.

[arber333]

Hm... Is this something that happens overtime when VCU is kept powered?

I dont know about Zombie 2.20 but i experimented with aot93 Teensy 4.1 VCU.
When i kept Teensy powered while car was stationary i noticed CAN commands seemed to fell out of sync.
When throttle would be applied car would jump forward and lurch backwards. I tried to change to another routine to ammend this, going from FWD through NEUTRAL to REV, but it wouldnt help.
I couldnt pinpoint this behaviour to anything particular in firmware...

I resolved this when performing direct reset on the VCU or i simply removed power(fuse).
In the end i decided to only keep VCU powered when car is running or charging.
I dont know if it would help, but here it is...

[LRBen]

More testing, new results. Looks like the outlander hearbeat mod firmware didn't have any effect on the dc-dc. What happens is that the dc-dc will work in run mode after it is first powered up. Then once you have cycled from run mode/charge mode to off mode, the next time it goes into run or charge mode the dc-dc will not work. Power cycling the dc-dc will fix this.

In 2.17, the dc-dc will work every time.

Hm... Is this something that happens overtime when VCU is kept powered?

I dont know about Zombie 2.20 but i experimented with aot93 Teensy 4.1 VCU.
When i kept Teensy powered while car was stationary i noticed CAN commands seemed to fell out of sync.
When throttle would be applied car would jump forward and lurch backwards. I tried to change to another routine to ammend this, going from FWD through NEUTRAL to REV, but it wouldnt help.
I couldnt pinpoint this behaviour to anything particular in firmware...

I resolved this when performing direct reset on the VCU or i simply removed power(fuse).
In the end i decided to only keep VCU powered when car is running or charging.
I dont know if it would help, but here it is...

I'm hesitant to think it would be anything like that. Nothing that major has changed between 2.17 and 2.20 and I've been running the Outlander inverter on zombie for 14 months now.

[tom91]

In 2.17, the dc-dc will work every time.

Please get CAN Logs of both so we can investigate what is causing this behaviour.

[arber333]

More testing, new results. Looks like the outlander hearbeat mod firmware didn't have any effect on the dc-dc. What happens is that the dc-dc will work in run mode after it is first powered up. Then once you have cycled from run mode/charge mode to off mode, the next time it goes into run or charge mode the dc-dc will not work. Power cycling the dc-dc will fix this.
...

I noticed the same with charger... if i would connect the incorrect 12V ON signal then it would strart the charger and DCDC in the wrong sequence.
Best result was when i used about 2s delay from connecting PP signal to performing precharge and connecting charger enable to 12V.
I suggest to go back to original wiring schematic and inspect what happens when you use charger enable and or DCDC enable line.

Edit: i noticed most Outlander components need heartbeat 100ms but some of them will only work with 30ms!

[tom91]

some of them will only work with 30ms!

Which components? Jamie is running charger/dcdc and heater with the current zombie release without issues.

Also logs from outlanders show the 100ms timing of the message.

[arber333]

Which components? Jamie is running charger/dcdc and heater with the current zombie release without issues.

Also logs from outlanders show the 100ms timing of the message.

I see heater will only reliably work with 30ms heartbeat. If i tried 100ms i got intermittent starting and very limited heat.
After several attempts i landed with 30ms as the surest signal for working heater.
AC compressor would sometimes dud on me as well and 30ms signal completely rectified that.
It could also be reflection on CAN lines but all my wiring was allready committed and 30ms for 0x285 just worked.

[LRBen]

I noticed the same with charger... if i would connect the incorrect 12V ON signal then it would strart the charger and DCDC in the wrong sequence.
Best result was when i used about 2s delay from connecting PP signal to performing precharge and connecting charger enable to 12V.
I suggest to go back to original wiring schematic and inspect what happens when you use charger enable and or DCDC enable line.

Edit: i noticed most Outlander components need heartbeat 100ms but some of them will only work with 30ms!

The current setup I have is the charger on perm 12v, with the dc-dc enable being on the zombie main contactor pin via a relay. This way dc-dc comes on after precharge in either run or charge mode. So far it has worked fine. The charger is on perm 12v as I use the charger to sense cp and put zombie into charge mode.

Although I do have to power cycle the charger in-between going from run to charge mode. Otherwise it will go into charge mode but will not push any amps through. It will do charge mode - off - charge mode just fine, but once it has gone into run mode it needs that power cycle. My assumption is it is something to do with the heartbeat message. It's not an ideal setup but it has worked so far. Eventually I am going to put in a Foccci board which will take care of the cp signal so the charger will be able to power cycle, so this issue will be fixed by that. It will be a much better setup than what I have currently.

The curious thing is now what has caused the dc-dc to also need a power cycle every time, not just from Run to Charge, but run or charging a second time. I had thought it was completely separate from the canbus system given it was activated by a 12v signal. I can't see anything that has changed going from 2.17 to 2.20 that would interrupt this. I will get some logs later in the week which might help shed some light on the matter.

[arber333]

The current setup I have is the charger on perm 12v, with the dc-dc enable being on the zombie main contactor pin via a relay. This way dc-dc comes on after precharge in either run or charge mode. So far it has worked fine. The charger is on perm 12v as I use the charger to sense cp and put zombie into charge mode.

Although I do have to power cycle the charger in-between going from run to charge mode. Otherwise it will go into charge mode but will not push any amps through. It will do charge mode - off - charge mode just fine, but once it has gone into run mode it needs that power cycle. My assumption is it is something to do with the heartbeat message. It's not an ideal setup but it has worked so far. Eventually I am going to put in a Foccci board which will take care of the cp signal so the charger will be able to power cycle, so this issue will be fixed by that. It will be a much better setup than what I have currently.

The curious thing is now what has caused the dc-dc to also need a power cycle every time, not just from Run to Charge, but run or charging a second time. I had thought it was completely separate from the canbus system given it was activated by a 12v signal. I can't see anything that has changed going from 2.17 to 2.20 that would interrupt this. I will get some logs later in the week which might help shed some light on the matter.

Yup, that was the same setup i had and i setup my friends Mazda RX8 that way too. We had the same problem with both charger and DCDC. all of a sudden while driving aux voltage would fail. I think it is not Zombie or VCU related. It seemed like there is internal timer after which charger/DCDC cycles and doesnt turn back on until power is reset.

For my Mazda i now turn on inverter, charger and dcdc via 2s precharge and i simply programmed my vcu to turn on DCDC after precharge. The same for the charger. I used VCU input with 2K7 pullup to sense when PP would be connected.

My friend though wanted a permanent DC connected to sense 12V via CAN bus in long intervals. I solved this by a separate relay which came on in two cases. If 12V from key is connected or if PP would be connected. Then 12V would be applied and DCDC would come on.
With charger it was a little more complicated. When CP signal is connected to charger VCU would need to apply 12V power and within 1s it would need to start 0x286 CAN telegram, persuming 0x285 heartbeat is already running.

You got me thinking now... i will write about that setup it in wiki.