[DRIVING] Renault Clio Electrique -97

[arber333]

To clarify, the BMS issue is probably due to the isolation issue. I get a lot of large spikes on the +12V system when I tried to run the motor.
If I don't run the motor the BMS communication works fine.
So my plan is to solve the isolation first and then see if the BMS communication is ok.

As I have not seen any reports from others having issues with isolation on the Toyota inverters I highly suspect I made some mistake in my wiring or I have a damaged inverter. Luckily I have a spare Yaris inverter on the bench which can be used for reference measurements if needed.

Plan forward is as follows:
I will first disconnect all other cables from the inverter when measuring, (heater core supply, charger AC input at MG1 and motor at MG2).
I will then test to load down the voltage with 1M resistor. I might also test with lower resistance using a filament lightbulb or similar.

I recommend as Johannes said, take a high resistor and cross the voltage you are measuring. Might be you are measuring floating voltage.
Try disconnecting the cable in center of battery and measure voltage and resistance from contactor to chassis GND.
Also observe if you have any electric coolant heaters connected to your HV. Those can easily be the cause of isolation loss. When a strand of wire inside heater would be touching the tube wall HV would get to chassis through water.

BMS comm wires must be routed independently from HV and 12V wires. Any fluctuation on power lines can cause comms blanketing and HV PWM is a huge fluctuation. I recommend you put BMS comms lines on the side of the car where no other power lines are routed. Remember EM signal strength rises/falls by the square of the distance (proximity).

[bexander]

I've done some measurements.
Seems I have close to 2Mohm from either HV bat plus or HV bat minus to gnd inside the inverter, which to my understanding is normal.
This is good since I was very concerned, isolation error can be very dangerous.
If anyone is interested I measured this, after a tip from a friend, by using a power supply of 36V and injecting that between either HV bat plus or HV bat minus and gnd while measuring the current with a DMM set to uA. I was not able to measure this directly with the DMM set to Mohm.

I still have severe disturbance on the 12V system when I push motor current through the inverter. I can see spike of +-10V on the 12V with a frequency of 8,8kHz, i.e. the switching frequency x 2. I have tried to find how this gets out on the 12V but with no luck so far.
Tested to set the deadtime value (126) to twice of what I normally use (default 63). No change.
Tested to disconnect the positive output of the DC/DC to the car, hence disabling the DC/DC from starting. No change.
Tested to supply the inverter from a power supply instead of the 12V system. No change.

My guess is that the disturbance is so powerful that it should be traveling through cables but at this point I'm thinking it might be traveling by air?
Will need to check wiring and measurement setup to see if I somehow can effect the magnitude or get rid of the disturbance.

[Isaac96]

I had similar interference issues for a time, though I never saw it actually cause issues; it was the same 8.8KHz and showed up whenever the motor got power. I solved it by grounding everything (with large straps) and putting braided shielding around the motor wires. Though there was still some noise visible on the scope...
I drove the car 200 miles like that until the motor shaft snapped.

[arber333]

Tested to set the deadtime value (126) to twice of what I normally use (default 63). No change.
Tested to disconnect the positive output of the DC/DC to the car, hence disabling the DC/DC from starting. No change.
Tested to supply the inverter from a power supply instead of the 12V system. No change.

My guess is that the disturbance is so powerful that it should be traveling through cables but at this point I'm thinking it might be traveling by air?
Will need to check wiring and measurement setup to see if I somehow can effect the magnitude or get rid of the disturbance.

Yes well i use deadtime 180 which would be 3us... You can try this also. Toyota engineers might cut normal PWM pulses short to protect power stage and that might cause spikes in your 12V wires.

What is the length of phase cables?
Though i found out early in the process that phase cables are not as much a problem. DC cables on the other hand may span through the whole vehicle and strange harmonics can travel on them back and forth...
To some extent we can mitigate this by intertwining DC cables HV+ and HV- and there is some effect there.
Even more effect is using HV cable shield with good grounding along the car.

[bexander]

After a lot of measuring today it seems the disturbance is leaking out from HV battery through main contactors and bms slave communication.
The BMS slave communication is today isolated by capacitors. I will replace them with transformers for better noise supression.
Regarding main contactors I did test a RC-filter (1ohm/2200uF) on the 12V supply to GND with very little effect. I moved the GND connection closer and that did have some effect so I think I'm getting close.
I will try and figure out what kind of filtering and GND connection I can implement to block the interferance from getting in to the BMS which needs to be on the same supply as the contactors.

EDIT:
Both motor cables and HV battery cables are shielded with connections to GND at both ends.
Motor cable is approx 1m.
Battery cable is approx 2,5m, running under the car, sepparated from anything else up until inside the battery box.

[arber333]

After a lot of measuring today it seems the disturbance is leaking out from HV battery through main contactors and bms slave communication.
The BMS slave communication is today isolated by capacitors. I will replace them with transformers for better noise supression.
Regarding main contactors I did test a RC-filter (1ohm/2200uF) on the 12V supply to GND with very little effect. I moved the GND connection closer and that did have some effect so I think I'm getting close.
I will try and figure out what kind of filtering and GND connection I can implement to block the interferance from getting in to the BMS which needs to be on the same supply as the contactors.

EDIT:
Both motor cables and HV battery cables are shielded with connections to GND at both ends.
Motor cable is approx 1m.
Battery cable is approx 2,5m, running under the car, sepparated from anything else up until inside the battery box.

Hm... On my early BMS i observed signal spikes on the optocoupler inputs. Positive side of BMS was isolated by optos while on the negative side i had direct connection to the PIC12F chips. This wrecked havock with BMS comms for some time untill i decided to scope one optos (master/slave). Good lord! There was grass growing in the UART signal as soon as PWM started... Then i tried to make RC filter by adding various caps on the opto input resistor. At one point i got really good curve and comms went live... This would be the simplest method.
But my comms were at 9800bd which is really low so complete voltage roundup lasted just under a second.

[bexander]

With 100nF ceramic capacitors in parallel with the contactor coils the spikes on the supply is reduced to a level I can live with.
I have ordered parts to implement transformer isolation on the BMS communication. I hope this will solve the issue.

Will focus on getting the charging up and running while waiting for the parts.

Also took the car out for a very short drive in the close neighborhood. Felt good to finally be able to drive it again.

[bexander]

I have now implemented the isolation transformers to the BMS communication. Voila! Now it works while pushing current through the motor.
Will try to tidy it up for semi permanent installation...

Plan is to disassemble the inverter to make adjustments to the current measurement for the charger (MG1). The signal measured is a bit to small so will try to increase signal level by removing voltage divsion and use a op-amp circuit instead. Should increase signal level by 5 times.

[bexander]

I finally got a somewhat working BMS, both HW and SW.
After modifying the current measurement and the dc-bus1 voltage measurement in the inverter I can now charge the battery. Remains to install Type2 connector and test up to full power (22kW). As of now only tested with 4kW, which is enough for every day charging.

Getting close to drivable car...

[johu]

Very nice! So you are using Prius buck/boost for charging?

[bexander]

Yes, I'm using Prius Gen3 buck/boost supplied with 3-phase for charging with celeron55:s SW with minor tweaks to both HW and SW. Not much tested yet, but enough to conclude it works.
With my obsession for weight reduction, it is really amazing that the Prius inverter provides inverter, DC/DC and charger in a very compact package with a weight of only 13kg!!

[johu]

I split the stuttering discussion off: viewtopic.php?f=7&t=1546

[bexander]

I finally got a chance to test drive my car up to resonable speeds. Unfortunatly the results are disapointing but expected after tests done by tom3141.
I'm getting good acceleration up to around 40-45km/h and then field weakening is required and reduces the fun to about 11kW or 30A battery current with the settings used. The car still accelerates uphill so I can get upp to my desired speed of 80km/h. I would prefer to have 20kW available up to 80km/h for some more margin. Running a 96S battery pack at 360V. To make this motor work good I guess you would need a lot higher voltage...

Screenshot_2021-04-17-06-22-48.png

Prius_MGR_FW5_00FOC_2021_03_31.json

Maybe time to admit defeat regarding MGR and try to find another motor solution.. Kia HSG might be an option, even though on the weak side...

EDIT: To get from fstat(Hz) to motor speed(rpm) multiply by 15 and to get from motor speed(rpm) to vehicle speed(km/h) divide by 65,8,

[bexander]

Time to once agin test with boosted voltage to find out if I can get 20kW up to 80km/h. This time boost from 360V up to 500V or if I dare 550V or even 600V...
My aim is high efficiency and the boosting will loose some efficiency but longterm plan is to use boosting only during acceleration and heavy load as it seem to work ok with field weakening to maintain 80km/h on flat road. Will require more advanced SW, specially as I want to use the buck/boost as a charger as well but if I can make the MGR work as a motor for this car it will be worth the trouble.

[thEVan]

I don’t have anything technical to add. But I wanted to send you some encouragement because I thing this build is amazing. I really like the focus on aerodynamics and weight to minimize consumption. Keep it up and keep us updated!

[bexander]

I have decided to give up on the MGR at the moment.
I want my car to be drivable again (been standing still for almost a year now) so I have decided to replace the MGR with a Leaf motor. This will cost about 40kg extra but is a lot more tested and proven with the openinverter so will most likely be plug-n-play. It still have the potential of increasing efficiency over the stock motor in this car and is lighter (stock motor 78kg vs Leaf EM57 53kg).
My plan forward is to mate the Leaf motor to the stock gearbox which have a 1:8,02 which is very close to the Leaf gearbox so should be ok. This will require an adapter plate, shaft coupler and one motor mount to be fabricated.
Will keep the Prius inverter as it includes DC/DC and "charger" and is lighter the Leaf inverter.

[mikmur]

too bad MRG didn't live up to your dreams.
I learn from your experiments.
I will probably refuse too.

[arber333]

I have decided to give up on the MGR at the moment.

That is unfortunate. What about outlander phev rear motor with original gearbox? Or Leaf with original gearbox.
I think Leaf motor will be too cubersome on your original gearbox. Dont forget it turns up to 10krpm. Your bearings might protest...

On the other hand Leaf motor is a beast. Last week i drove home from work 150km/h as the road permitted and motor was only 45deg when i checked at home. I am sure power limit is at battery side or inverter. So 350A x 360V because of batteries and main fuse.

[bexander]

To be clear, the MGR is still a motor with great potential but at the moment the MGR in combination with the openinverter is not working properly.
I have considered buying another inverter but soon discarded that idea when I realized the cost of this was past €3000.

There are indeed other motors out there but I was given the Leaf motor in return for helping a friend so I thought, why not test it in the Clio. I will limit the motors torque and power to make the original gearbox survive.

[arber333]

To be clear, the MGR is still a motor with great potential but at the moment the MGR in combination with the openinverter is not working properly..

I wonder how MGR would fare with Lebowski core. It would be interesting since it doesnt care about pole count. Also you can directly command FW current up to 75% of phase current. And at that rpm it would run in sensorless mode.
Maybe we could have two motor pairs, MGR(lebowski) vs MGR(openinverter) run to better set the openinverter parameters?

[bexander]

I still have the MGR in my car and are considering another go at it.

I have come across SW for a evaluation kit for motor drive HW from a large power electronics semiconductor manufacturer. Will try to interpret the equations used and see if they can be implemented in the openinverter SW. So far I have found two different variants, one for salient motors and on for regular motors. I belive most OEM EV motors to be of salient type as interior permanent magnet motor usually are. For regular motors regulation for 0 Id is normally used but for salient it is different.

[johu]

Good to hear.
Did you ever try this software? viewtopic.php?p=29757#p29757

Lets you specify id/iq per throttle separately. So you could try 2A/% id and 1A/% iq or so.

[bexander]

No, I must have missed it.
However I don't think that SW is a permanent solution, more a testing thing? I already know there is potential in the MGR so I will focus more towards finding a permanent solution with good control far out in the FW domain...

[bexander]

Today the Clio passed the yearly car inspection! So now it is possible for me to do some proper road testing. Preliminary data indicates increased efficiency, more tests will follow.
Working on the charger SW at the moment as I'm not satisfied with how the current sampling/measurements work. When this is done I will go into more detail of how to get the field weakening working in the inverter SW for the MGR.

[mikmur]

Congratulations. I'm happy for you