Hey, I try to run my 4 pole IPMSM with PWSVM and use a typical controller with PI-controller with a current loop and speed loop. When I run my system with DC-supply 100 V with a steady-state speed of 700 rpm, I see that both Iq and Id go towards 0 A, since I don't have any load except some friction. The Vd and Vq computed out of the current controllers have constant value as well. But when I run my system with DC-supply 150 V with the same speed, the output of the current controllers is no longer stable. It seems like the position measurement is gradually getting out of phase. Until the point, I get an overcurrent fault in the inverter. Have anyone had a similar problem?. Is it possible for the system to get unstable with a higher DC-supply?
The output of the speed controller (Iq_ref), does not change much compared to Vd and Vq, but the signal gets more noise.
Maybe the problem comes from the SVPWM algorithm, test it alone with a resistive load, check the radius of the reference.
Are you using simPowerSystem or simulink to model the machine?
The control code is made in Simulink and I'm running it on "dSpace MicrolabBox". I suspect it has to be something with the algorithm. I'm been testing with closed-loop hysteresis control and it can run with a supply on 190 V up 1200 rpm.
New update. Have tried to run the code on hysteresis. Finally was able to tune the speed controller right and can now run with DC-supply 260 V and run at 1500 rpm. The problem with SVM still persists.
Sveinung Attestog this confirmes that the problem comes from the SVM algorithm, have you tested it with resistive load? also check the sector detection program, sometimes the detected sector is shifted from the actual real sector, so try also to make a manual shift on sector detection.
Well, was able to run the system now with DC-supply 280 V and run at 2000 rpm with SVM control. Can't figure out why the position in the PI controller gets out of phase. So I just added an angle compensator, where I decide to set Vd_ref = 0 V and integrate the error (Vd_ref-Vd) with an integral gain -1. Not sure if this is the correct way to control IPMSM, but it works with no-load.
Did a small check, in a simulation model in Simulink and see that keeping Vd=0 V seems to be OK for no-load condition, but not when it's loaded. So if probably will need to adjust the Vd_ref according to load.
be sur to use the correct Vdc inside the SVM algorithm
Why are you controlling Vd_ref? why are you not controlling dq currents ??
can you please show us a general schematic of your control scheme ?
I am controlling Id and Iq in a typical FOC scheme with PI controllers. But I noticed when I operate at no load Vd, stays at 0, thus to adjust the input angle for the Clarke-Park transformations.
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running