Hello researchers,
I am implementing sensor less control of PMSM from estimated back emf technique. I am getting an angle in the range of o to 2*pi which is perfect, however when I try to convert it into speed (rad/sec) by (differentiating or using PLL) I get lower dip in speed waveform at frequent interval exactly when theta is changing from 2*pi to zero. How can I get rid of this problem?
If the increment has a magnitude larger than pi, add or subtract 2*pi.
Sagar,
As Olof has rightly stated, this is a numerical issue as the angle is proceeding as
0, 0+d, 0+2d,....
"after" the cycle completion point, and as
...2*pi-2d, 2*pi-d, 2*pi.....
"before" the same point.
So AT the cycle completion point, the speed computation is giving you:
(2*pi + d - 0)/T = (2*pi + d)/T
while you should compute as
(0+d - 0)/T = d/T
Just "adjust" the "additional 2*pi" from the calculation, and the signal will be corrected!
-Sanjay
Thank you for your reply
I have attached the screen shot of the issue. I have tried your suggestion but I am still getting almost same output.
Something is wrong with your implementation of the speed algo.
Difficult to say based on the waveforms. Maybe you need to check the thing for bugs.
-Sanjay
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering