[SOLVED] Flipsky 75100 pro V2+ QS90 (1000W) - FOC detection stalling at Flux Linkage

[simoneek]

Hey,

Running into a wall trying to get a Flipsky 75100 Pro V2 to play nice with a QS90 mid-drive (1000W PMSM, 5 pole pairs).

The Motor Setup Wizard gets a clean lock on step 1 (R/L):
- R: 182.70 mΩ
- L: 72.66 µH
- Current: 85.42 A

But step 2 completely fails. During the open-loop spin up for the flux linkage, the motor coggs heavily, spins super slowly/choppy, and the wizard dumps a 0.00 mWb result. Obviously, FWD/REV testing doesn't work at all after that since the VESC thinks there is no flux.

I tried overriding the advanced settings using the E-Bike DD Hub profile (~6kg) to give it some headroom:
- Max Power Loss: 2000W
- Openloop ERPM: 400
- Sensorless ERPM: 2500
-Poles: 10

Still nothing, it just won't spin cleanly during detection.

Anyone running this exact combo on a Pro V2? Is the 75100 shunt filtering just too noisy for heavy inrunners at low speed open-loop, or is there a specific current/duty cycle tweak I can use to force the flux linkage measurement to pass?

Attached the partial detection screenshot. Thanks.

 

[stancecoke]

Try to run the detection with much less current! 5 amps are enough.

 

[scianiac]

Try to run the detection with much less current! 5 amps are enough.

I was going to say more current, if I recall that 85A in the output is what it thinks the max motor current should be based on the power loss you gave it, it doesn't use that for detection, it uses some fraction of that. You can do the measurements in the motor tab instead which gives you more control. If it can't spin the motor up nicely it probably doesn't have enough current. Honestly I'm not even sure 5A would motor current to spin that motor very nicely.

Try lowering the ZVF to 20 or 25khz, seems to improve noise on these noisy controllers and there was something I recall about doing that for detection. I don't recall exactly but the motor may run nicer with a lower ZVF anyway.

It should be able to get a flux linkage value just fine as long as it gets the motor spinning and if I recall it's pretty accurate even for IPM inrunners. I think all of my manual flux linkage measurements ended up pretty much the same as the VESC measurements even with an even noisier controller.

After you get the flux linkage though you may want to look into installing an encoder. I mean for that size of motor with that controller you may have no issues at all because you are running that motor so far below it's limits. Honestly I'm not sure why you have such a large motor with such a small controller or maybe you're just testing it.

 

[simoneek]

Hello stancecoke ! Thank you for your reply
If I change "max power losses" to 100W, I have 20A but the motor don't want to move

 

[simoneek]

I was going to say more current, if I recall that 85A in the output is what it thinks the max motor current should be based on the power loss you gave it, it doesn't use that for detection, it uses some fraction of that. You can do the measurements in the motor tab instead which gives you more control. If it can't spin the motor up nicely it probably doesn't have enough current. Honestly I'm not even sure 5A would motor current to spin that motor very nicely.

Try lowering the ZVF to 20 or 25khz, seems to improve noise on these noisy controllers and there was something I recall about doing that for detection. I don't recall exactly but the motor may run nicer with a lower ZVF anyway.

It should be able to get a flux linkage value just fine as long as it gets the motor spinning and if I recall it's pretty accurate even for IPM inrunners. I think all of my manual flux linkage measurements ended up pretty much the same as the VESC measurements even with an even noisier controller.

After you get the flux linkage though you may want to look into installing an encoder. I mean for that size of motor with that controller you may have no issues at all because you are running that motor so far below it's limits. Honestly I'm not sure why you have such a large motor with such a small controller or maybe you're just testing it.

Hello,
Tried dropping the Switching Frequency (ZVF) to 20kHz like you suggested but nothing changed.The motor itself is definitely fine, it runs perfectly on an ASI BAC555. I'm actually switching away from the BAC555 because the ASI ecosystem is completely locked down and closed source, which is why I'm trying to get it working on this VESC setup.

I have tried with more current but the result is the same... The motor don't want to run and detect the flux.

The limitation with lower ZVF is for the eddy current in the motor, you have more losses.

 

[scianiac]

If I recall the current that the controller actually runs the motor at during detection is something like 10% of the set current. When you say you tried higher current how much more?

If you go to motor settings > FOC > General you can run the RL and lambda (flux linkage) detection at the bottom (on the PC app, I don't remember where the buttons are on the phone). There you can set the actual current parameters. Included is the ramp up which you can lower so basically it tries to spin the motor up slower. It's not a small motor and without enough current it won't be able to spin it up very rapidly.

For context my understanding is to do the flux linkage detection is spins the motor in openloop, basically it just fees the motor a sine wave that ramps up in RPM from zero to the speed at the set duty cycle. So to do that it needs enough current for the motor to "follow" the sine wave and needs to ramp up slow enough for the same reason. The controller is not looking to see if the motor isn't following the sine wave hence the failure. Once the motor is spinning it then does the flux linkage detection by reading the back EMF.

 

[simoneek]

I have tested with 107A

This is what I have when I test lamba parameter, the RL test works. The motor doesn't move.

 

[simoneek]

I have also tested at 10A, nothing changed.

 

[simoneek]

I have another QS90 motor on hand. Do you think it's worth trying a swap to see if it makes a difference?

Also, could swapping the motor phases help, or does the VESC not care about phase order during detection?

 

[thepronghorn]

I think trying a different motor could be valuable to make sure it's not a motor problem. I would also try spinning the motor with the "foc_openloop" commands in the VESC tool terminal.

 

[stancecoke]

Have you switched off the phase current filtering? The Flipsky 75100 hardware doesn't support this feature.

 

[simoneek]

Hi!

I disabled phase current filtering, but nothing changed.
I tested the command: foc_openloop 4 500.It works without the Hall sensor (with the connector unplugged), but it does not work with the Hall sensor connected!
I also tried manual calibration; it works without the Hall sensor, but the issue returns as soon as I plug it in.

I tested another motor as well, but only without the Hall sensor because I don't have the correct connector

 

[simoneek]

Breaking news !! ahah
I cut the temperature sensor and it works perfectly ! With the hall sensors !

Do you have any idea why ? Could I drive without this sensor ?

 

[scianiac]

That is bizarre, they should have nothing to do with each other. Although maybe check that the temp sensor is reading correctly and that the temp sensor limit is set sensibly. Usually if the temp sensor is reading way off it just yells at you but it could be reading too high and maybe that's limiting the power so measurements are incorrect, or maybe some noise is coming back into the temp sensor wire causing it to read erratically.

You can run it just fine without a temp sensor, probably not an issue either as you aren't exactly running the motor very hard.

 

[simoneek]

yes that is bizarre.. I change the topic to solved.

Thank you for your help !!