Hello,

I am trying to build my own ESC (Brushless DC controller) and I managed to get the motor running starting from a ramp to a steady PWM duty Cycle and Commutation period and these are the BEMF waveforms that I see. My question is, where is the "zero crossing event" and why is it different (somehow inverted) from the regular BEMF waveforms displayed on various application notes? Kindly refer the waveform images below.

http://static.rcgroups.net/forums/attachments/3/6/9/4/5/8/a4175768-10-NewFile0.jpg

http://static.rcgroups.net/forums/attachments/3/6/9/4/5/8/a4175769-43-NewFile1.jpg

Here-in, it is almost impossible to detect '0' crossing. Also, in my scheme I am PWMing the upper MOSFETS, with the lower ones being used/made high during relevant commutations.

 

With unipolar PWM it is very difficult to filter out the zero crossing, depending on if your pwm'ing the high side or low side, you need to sample the bemf on either the pwm off time or during the pwm on time.

I suggest reading up on sensorless BLDC to get a bit more knowlege and insight, the appnotes from freescale, microchip, st and atmel is a good place to start.

However, draw an imaginary line in the middle of the waveform, whenever the unpowered phase crosses that line, theres your zero crossing.

Myself i am also working on a sensorless bldc drive, but im using bipolar pwm which makes detecting the back emf way easier, but also plan to include a unipolar pwm mode for those hard to drive low inductance motors.

 

Hello ,

What do u mean by Bi polar PWM. Does it mean providing PWM on both pins. Are u giving them same PWM or complementary PWM. If you see my waveforms , it wld seem that the '0' crossing occurs almost at the begining. ARe you getting better results with your bi polar approach.

Thank you.

 

This is bipolar PWM:
http://i.imgur.com/0NxBJFd.png
And the circuit to make it:
http://i.imgur.com/u1q4uWY.png
And the resulting virtual hall signal:
http://i.imgur.com/aorwetF.png

However note that bipolar pwm is not usable with low inductance motors as it will generate big losses and alot of heat in the motor windings.

 

Also I will need dead time interval implementation, right.

 

Only with the bipolar PWM, with unipolar pwm your just interrupting a standard six step commutation sequence.

Like i said, do some gooling on sensorless bldc, my project is a result of years of research, and i still have long ways to go.