Enova Combi Board [SOLVED]

[tom3141]

Hi everyone

First post here, but you may have seen me on the DIY Electric Car Forum with my Suzuki Cappuccino build. I've got an Enova inverter that's controlled using Damien's combi board.

I've finally got my Cappuccino wired to the point where I can try it out albeit powered using a lab PSU. The bad news is that I still have an intermittent problem with the inverter gate drivers/IGBTs. I had this before when testing on a bench but I decided to get it all wired in properly before investigating further in case it was a wiring problem with my test setup.

This is the process I went through today:

- I turned the ignition on, and then to the momentary start position after a couple of seconds. The power supply hit the current limit (20A). (no throttle input and neither forward or reverse selected)

- I knew the problem before had been related to the gate drivers/IGBTs so I unplugged the lead between the driver board and the first IGBT. The whole system then turns on fine and forward or reverse can be selected.

- I then plugged this lead back in. Everything turned on this time but the PSU hit the current limit as soon as I selected either forward or reverse. (no throttle input). If I then flick back to 'neutral' it stays at the current limit but if I turn it off and back on again it will be fine until I then select forward or reverse again.

- Unplugging the other driver-IGBT leads doesn't help at all so it seems related to the first IGBT lead that I unplugged.

- Eventually, after a few minutes of turning it on and off and fiddling with these leads I got it to the point where I could select forward or reverse without it hitting the current limit but upon applying some throttle the PSU would cut in and out and the motor would turn rather jerkily.

- It then just started working perfectly, spun the wheels up fairly quickly without hitting the PSU current limit

- After taking some videos and enjoying seeing the wheels spinning for the first time it then stopped working again. This time there was no PSU current limit but the motor would only jerk back and forth however much throttle I applied (as if one phase was not working?).

- It then went back to hitting the current limit every time forward or reverse was selected

I had some suggestions that I am yet to try from when I had this problem a while ago on the bench but if anyone has any more ideas or suggestions from this description they would be gratefully received!

Cheers,

Tom

[nailgg]

Hi Tom,

So you could spin the motor successfully at least for a while, that means your IGBTs are good. I am not sure about the schematic of the logic board you have but as far as I know Combi boards incorporate ACPL-337J gate driver ICs, which has the desat/uvlo protection. Those ICs might be giving some kind of error and when it happens one of the six IGBTs doesn't switch anymore.

* Is your NAND protection chip plugged in? When a desat/uvlo occurs and one of the gate drivers refuses to work, that will inhibit the PWM so your motor doesn't move jerky.
* Have you measured all of the DC/DC converters' output voltages? Not sure what kind of DC/DC you are using but you must see something around +24V between +Vout / -Vout (assuming that you have +15V/-9V DC/DC converters). I have seen when a gate driver is faulty it goes short and DC/DC output voltage goes down to +3V. So make sure that you have them all good.
* Have you scoped all the gate driver outputs? Just remove the high voltage power supply, start the inverter operation, select a direction and use a scope to view between gate-emitter on all six IGBTs. You should see 50% duty cycle on all of them, swinging from +15V to -9V.

I think the one of the low/high side gate driver ICs in the first phase is faulty and needs to be replaced.

[tom3141]

Thanks for your suggestions. I've done some more investigating this morning, results below.

First, I checked all the DC/DC output voltages (with no high voltage connected and the inverter in stop mode). +Vout measures 15V and -Vout measures -5V on every one as I believe they should (they are Murata MGJ2D051505SC).

I then checked the gate driver outputs (between gate and emitter) with the inverter in stop mode and they all measure -5V which again I believe is correct. I will need to borrow a scope from work to check the gate driver outputs in operation.

They are ACPL-337J ICs but I'm a bit out of my depth with the desat and NAND protection stuff. I got the board from Damien populated and tested so I haven't built it myself as a lot of you guys have. The electronics and software side of things is where I have the least knowledge unfortunately.

I had another go at powering it all up just in case my fiddling had changed anything. It's still the same, hitting the current limit of the PSU when I first switch on and then gradually getting better. I've taken a video (see link below) of the power supply when I had the inverter powered up and forward selected but no throttle input. The motor is not turning (or trying to turn) here, I've literally just selected forward and it starts doing this. The noise is from the power supply not the motor or inverter. It doesn't always do this, sometimes it's fine and draws no current in this state, other times it hits the current limit and stays there, other times it does this:

https://1drv.ms/v/s!Am-VyIvBzOs6g80WmA69luXigBCCIQ

I also took the video below when it was working slightly better. I got the motor turning but it was very jerky (causing the vibrations you can see) and the current and voltage were all over the place. When I've had it working properly the voltage will barely drop and it will just ramp up smoothly to 10 or 15A.

https://1drv.ms/v/s!Am-VyIvBzOs6g80XOlvrk-DduRjStA

I may be completely wrong here but I'm wondering if the problem could be related to the DC bus capacitors? Either that they are faulty (I don't know how long they have been sat unused, they came with the Enova) or due to the low voltage I'm using. It may have nothing to do with it but it just crossed my mind this morning as the first time that I managed to get it to turn on without immediately hitting the current limit was when I had left it in the precharge state for a couple of minutes (not a few seconds). Can capacitors such as these become dodgy/fail intermittently? Would that produce my symptoms?

[nailgg]

So if you bought it already populated and tested, NAND protection IC is probably there.

What is the value of "deadtime" parameter in the inverter software? What is the value of the gate resistors? If the gate resistor value is too large for your IGBTs, and if the deadtime is too small, shoot-through might occur when PWM is initialized (when fwd/rev is selected even with no throttle input).

I'm not sure with the capacitors Do they keep drawing current after 10 seconds when they're in the precharge state?

[tom3141]

I can't exactly remember the deadtime parameter off the top of my head but I know I played around with increasing it when I had the same problem during bench testing. I think it was set at '63' and I tried increasing it with no positive results.

Again not sure on the gate resistor value, I'll have to check when I'm back in the garage. I think Damien initially had exactly the same set up on his drift BMW project so I would think it's more likely to be a fault with something than incorrect components.

I can't actually see any current draw on the PSU during the precharge state. The PSU display is pretty slow to respond so I don't know if it's just too quick or maybe that suggests there is a problem?

[nailgg]

The initial charge should already be completed in a few seconds so it's normal that you can't see the current. Actually I was asking if they keep drawing current after the precharge is completed, but looks like they don't.

I don't think that your issue is with the capacitors, that wouldn't make the motor spin jerky. If a capacitor was damaged, what would happen is probably the bus voltage would overshoot to a high voltage because the capacitor just couldn't smooth out the voltage. In this case, either your IGBTs would pop or you might get a desat error. Or, if a capacitor is damaged it might not be able to handle the ripple current too, but the Enova inverter has many capacitors in parallel so even if one was damaged the rest would handle these stuff, at least for a no-load operation, in my opinion.

ACPL-337J supports peak 4A charging current, so since you have +15V/-5V DC/DC converters, value of the gate resistor shouldn't be lower than 5 Ohms ((15-(-5))/4). Otherwise, the output stage in the gate driver IC would pop. Not sure what IGBTs you have but some IGBTs already have internal gate resistors too, so in that case a lower gate resistance could be used on your gate driver.

What I'd do in your case is, I'd unplug the Olimex from the main board and remove the high voltage supply, connect all IGBTs to the gate drivers, power the main board with +12V so the gate drive circuitry is powered up, then insert a 3.3V square wave (8.8kHz) to the mainboard where Olimex PWM output pins are connected (TIM1_CH1, TIM1_CH2, TIM1_CH3, TIM1_CH1N, TIM1_CH2N, TIM1_CH3N, you can check Olimex STM32-H103 datasheet for the pinout) to simulate the PWM operation. When you apply a 3.3V square wave to the non-isolated side of one of your gate drivers, the current drawn from +12V supply would increase about 100mA because of the IGBT switching operation. Do this for all six of them and record the current readings. If one of the six gate drive circuitry is broken, you should see a different current value. If they are all equal and current readings look normal, you can focus on the power stage for further inspection.

[tom3141]

Just done some more investigation with a scope.

With the inverter in operation and a direction selected I have a 50% duty cycle on 5 of the 6 gate driver outputs, swinging from +15V to -5V. The other gate driver output remains at a constant -5V. It's the EM_H and GATE_H of phase 3 if that makes any difference.

I then scoped the non inverting voltage control input (pin 2, VIn+) on all of the A337J chips. They were all showing a 3.3V PWM signal apart from the one that has the problem. I've attached a photo of the bad one compared to the other 5 A337Js.

This suggests to me that the output from the Olimex could be dodgy. Any thoughts on the most likely cause?

[nailgg]

Yes! That’s probably the faulty STM32 output. I had a very similar issue with this uC before. All 5 pins had a 0-3.3V square wave but one of them (it was CH3N, in my case) had something like 2.5V-3.3V, effectively keeping the gate at “on” state all the time. Luckily, the Olimex board costs only several $

[tom3141]

Thanks for your help.

I'll get a new Olimex ordered, hopefully that will fix it.

The only thing that concerns me is that if this is a fairly common problem, what's to stop this happening again when my battery is connected, blowing up an IGBT? Bit of a worry!

[nailgg]

It probably happens due to the ESD, no need to worry unless you touch your board too much. But this is one simple example to why automotive industry has too much safety standards. A professional inverter made by an automaker would probably test the uC and gate driver outputs every time the inverter is powered up, in a very small amount of time. It's also possible to use some additional logic circuitry to avoid shoot-through due to a damaged uC. To achieve these, however, you'll need additional hardware and software. With the simple hardware we've got here, we can only rely on the desaturation protection.

[johu]

Thats an interesting point. I wonder what faulty gate drivers outputs read when used as inputs?
Tom, if you still have the damaged board I could hack up a firmware that configures the 6 PWM pins as inputs and reads their state. Would be a cheap check

[tom3141]

Interesting, hadn’t thought of that. I haven’t touched the faulty Olimex yet, it’s still plugged into the PCB. A new Olimex arrived yesterday and I was planning to swap them over on Saturday. I’m happy to try some hacked up firmware to investigate this before I swap them over.

[johu]

Ok, please use the attached firmware and let us know the value of "pwmio"

[tom3141]

I'm having some problems. Previously, to set up parameters etc, I used this web interface - https://github.com/dimecho/Huebner-Inverter. The PC that I used last time has since died so I've had to set it up on a new laptop.

Previously, I would just connect to the combi board via USB, run the Huebner Inverter.bat file and it would work fine. Today, I'm getting the error shown in the attached photo. Any ideas?

Also, I've realised that I'm not actually sure how to flash the brand new Olimex. Do I need a special programming tool or can I just do it over USB?

Sorry for all the questions, I'm pretty new to this kind of stuff!

[nailgg]

@tom3141, in order to flash a bare STM32, you need some adapter like either ST-Link v2, or ARM-USB-TINY-H. There are cheap Chinese ST-Link clones available in the market, you can just use the cheapest one you can find in the local shops around you.

For the serial communication, you should be able to see the COM port in the device manager. Just type devmgmt.msc to the Start->Run menu (Win+R) and find your device under Ports (COM & LPT). Without flashing the software to the STM32, you can open the COM port but you won't be able to send/receive data, of course.

[tom3141]

Thanks, I suspect one of my work colleagues might have one so I’ll ask around.

I don’t think I was very clear regarding the serial comms. The photo I sent was when I was trying to connect to the original board that is flashed and working intermittently. I want to connect back to this before I swap to the new Olimex so that I can download the current parameters and try out the test firmware that Johannes sent. I’ve connected to this previously with no problems.

[nailgg]

That web interface was developed by @Dima and he can assist you better, but I suggest you download PuTTY from here: https://www.putty.org/ and connect to the COM port you see in the devmgmt.msc with 115200 baud rate.

Or you can try finding the file Web\js\serial.json, manually edit the "COM2" port to the one you see in devmgmt.msc, and try opening the web interface again.

[tom3141]

I've managed to connect to it using PuTTY and retrieve the parameters I wanted, however the web interface is still showing the same error. The com port settings are set the same.

Is there a command so that I can update the firmware via PuTTY? At least then I can put the test firmware on to check the value of 'pwmio'

[nailgg]

To update the firmware without an adapter, you can use Johannes' Python based updater here: https://github.com/jsphuebner/tumanako- ... bootloader

You need to install Python if it isn't installed on your computer. I haven't used this updater before by the way, I'm not even sure whether it's usable in Windows as the bootloader based update instructions was provided in Linux:

Code: Select all

python updater.py -f stm32_sine.bin -d /dev/ttyUSB0

You can give it a try by replacing /dev/ttyUSB0 with COM2 or whatever port your device is named.

[tom3141]

I got an ST link v2 this week and managed to flash the new Olimex. I can still only communicate with the old Olimex via PuTTY. The web interface and even the ST link would not connect to it so I’ve been unable to try the firmware to test the pwm inputs/outputs.

The new Olimex appears to have fixed the problem although I need to borrow a scope to confirm. The motor is not spinning up nicely though, I think it might just be parameters as I still can’t get the web interface working. I’ve been setting parameters via PuTTY. I’ll have to investigate the web interface problem further.

[nailgg]

Some STM32s sometimes require connecting under reset, have you tried that? You can set it from Target->Settings in the ST-Link software.

So the motor is not spinning nicely, how are your parameters set? What's the battery voltage? Are you running in the manual mode or software mode? You have a fresh Olimex and already confirmed before that the gate drivers are nice, so it's might be because of incorrect boost, fweak and fslipmin/fslipmax values. To make sure with the fslip, simply try setting fslipmin/max values at the same value, around 1.5-2. Set boost to a value such that you're providing at least 15-20 volts at the startup. You can try setting boost to a value around 2000 and set udcnom to 360 for example. By doing this, the software will scale up your boost value automatically with a factor of 4x when you're providing 90 volts battery, or 6x for 60 volts battery etc and you won't have to change it all the time whenever your battery voltage changes. Then set fweak to some value near to 50 or whatever your motor's rated frequency is. Udcnom will also scale the fweak automatically and you will have a lower fweak when you're providing a lower battery voltage.

[tom3141]

I managed to get the web interface working on another laptop. I think the problem might have been a firewall on my work laptop that I couldn't turn off.

I had a few problems with the precharge circuit that turned out to be due to my 12V aux battery being flat. Everything would turn on but then the main contactor wouldn't close when it should. Fixed this now by using a fully charged 12V battery!

I'm still having troubles getting the motor to spin up properly and after some investigation with a scope this morning I think I still have the same problem that I had with the old Olimex. I checked all the PWM input pins on the A337J and they all show a 3.3V square wave apart from the one that I had the issue with last time. I've not had the motor spinning properly since I swapped the Olimex over so I don't think I have actually fixed the problem. I'm not having the shorting issue on start up anymore but the motor just rotates very slowly and jerkily - I guess this would make sense as 1 of the phases isn't working.

Any ideas on what else could be causing this? Is there anything else on the PCB that would affect this PWM signal? Or could it be a firmware issue? Just a dodgy solder joint? I might try to measure the output pin on the Olimex (rather the pin on the A337J) but I need to figure out which pin this is.

[Peter]

Hi tom3141. I am using the Combi V8 board too. Have you checked you are getting good noise free(ish) signals back from your encoder ? Also I had problems trying to run my Enova motor and spent weeks adjusting parameters to no avail. Contacted Johannes who set up a test with a Combi board (thanks once again Johannes) I was running version 2 with the Olimex, check if its version 1 or 2 when you read parameters, Johannes suggested I remove R21 which is to do with RPM output, once removed it reverts to version 1 (I snipped one leg at such a point where I could re-solder the leg if needed later without removing the pcb) the motor ran great immediately !! Worth a try as my motor would jump about crazily like it was missing a phase and tripping out on overcurrent and encoder signal, check to see what errors you have. Might be worth a try. Fingers crossed. Peter

[tom3141]

Interesting. Not sure if this could affect the PWM output to the gate driver but worth some investigation.

Are you talking about the ‘hardware version’ in the parameters? Mine is set to v2 if so.

I presume R21 is a resistor on the Combi board?

[Peter]

Hi tom3141. Yes its the hardware version, R21 is on the Combi board. Simple enough to test I guess. Peter