You are here

Motor detection accuracy in 3rd party hardware (old FW vs new FW)

4 posts / 0 new
Last post
shaman
Offline
Last seen: 4 days 15 hours ago
VESC Free
Joined: 2018-12-09 15:59
Posts: 58
Motor detection accuracy in 3rd party hardware (old FW vs new FW)

I've been working on an open source low-cost VESC 6 based ESC that uses TO-220 FETs, low-side sensing, and a few other hardware differences. I've noticed that the older motor detection process in the older FW (FW3.62) yields more consistent motor parameters than say the newest firmware (FW4.1). The newer FW can have a very difference values (especially inductance) and cause non-ideal motor operation in comparison. I was wondering if anyone knows all that factors that could affect the measurement in this case. I suspect things like the through-hole FETs or low-side sensing possibly altering the measurement but then why would the old FW be fine? Has anyone noticed this sort of thing with 3rd party hardware?

vadicus
Offline
Last seen: 11 hours 53 min ago
VESC Free
Joined: 2018-08-17 07:26
Posts: 228

I've noticed this too. The older firmware (pre 4.0) that measures R and L with a few ticking and then loud low pitch growling sounds provides consecutive measurements with a deviation of maybe 10-15mH. 

The new method that does the same ticking and then a a quick chirp/hi pitch sound may produce the inductance measurements within 50mH deviation or even more. This may vary also with the voltage. Lower voltage, e.g. 33v DC link, tends to produce higher inductance values, 100-150mH on my motor, and 86v may push the inductance down to 89mH. At the same time, the resistance measurements stay consistent at any voltage.

I almost always have to half the detected inductance value to be able to run pretty much any motor (all big heavy hubs and inrunners).

 I am not sure if this has to do with my hardware or the software.

Sometimes, I also have to reduce the flux linkage to remove the stutter. By the way, how does flux value affect the battery current measurements? I see that if I tweak it, the DC current changes.

 

   

 

district9prawn
Offline
Last seen: 1 hour 34 min ago
VESC Free
Joined: 2018-04-26 12:18
Posts: 106

It's weird on my a200s boards the new hfi inductance measurement gives more consistent measurements but I have some hw6 based boards that have large variations with the new method.

Vadicus by stutter I'm assuming you mean the rough motor and abs over current faults at low speed high current? I've also found lowering inductance to help but never tried going so low as half. I will try that. I haven't played with flux linkage as its essentially kv and I figured it should be accurate.

Isn't the battery current just calculated from phase currents and modulation?

vadicus
Offline
Last seen: 11 hours 53 min ago
VESC Free
Joined: 2018-08-17 07:26
Posts: 228

Yes, a motor would start rough even without load right after passing the detection. My motor is detected at around 230uH and would not work smooth unless I drop the inductance to 120uH or so. It seems to be a consistent pattern, I have to do it on all three motors that I tested. Two of them are hubs and one is a large/heavy inrunner. Maybe that's a thing for larger motors and smaller RC motors are detected correctly. I am yet to test an RC motor.