lewood
Newbie level 3
Hi all,
Couple of questions I'm hoping someone can give me a hand with. I'm implementing a motor controller, basically three cascaded control loops to control position, velocity and current of an electromechanical actuator. Motor is 3 phase brushless driven by 4 quad operation pwm.
First question, I am sensing current using a sense resistor at the bottom of the bridge. Sensing in this manner loses the sign of the phase current as the sense resistor sees both positive and negative current for a given phase current depending on whether the FETS are on or off. To get around this sporadic current sign I rectify the current waveform and specify the drive direction from the sign of the current demand. Seen this in a few app notes but it does introduce slight non-linearity into the control. Is there a better way of sensing current than this (without moving to using more than one current loop and controlling phase current for each phase seperately)?
Second question, at light loads the current waveform becomes discontinuous due to conduction through the diodes and effectively reduces the gain of the power stage and the resulting torque loop around it. I haven't necessarily seen many problems relating to performance but was wondering if there any techniques to get around this? Is is just accepted that the gain will reduce under light loads? What about locked antiphase drive? I have looked at this and it looks as though this would get around the problem as the diodes are no longer used (other than in dead time between switching) and the linearity of the control should be maintained even about the zero crossing.
Any help would be appreciated.
Couple of questions I'm hoping someone can give me a hand with. I'm implementing a motor controller, basically three cascaded control loops to control position, velocity and current of an electromechanical actuator. Motor is 3 phase brushless driven by 4 quad operation pwm.
First question, I am sensing current using a sense resistor at the bottom of the bridge. Sensing in this manner loses the sign of the phase current as the sense resistor sees both positive and negative current for a given phase current depending on whether the FETS are on or off. To get around this sporadic current sign I rectify the current waveform and specify the drive direction from the sign of the current demand. Seen this in a few app notes but it does introduce slight non-linearity into the control. Is there a better way of sensing current than this (without moving to using more than one current loop and controlling phase current for each phase seperately)?
Second question, at light loads the current waveform becomes discontinuous due to conduction through the diodes and effectively reduces the gain of the power stage and the resulting torque loop around it. I haven't necessarily seen many problems relating to performance but was wondering if there any techniques to get around this? Is is just accepted that the gain will reduce under light loads? What about locked antiphase drive? I have looked at this and it looks as though this would get around the problem as the diodes are no longer used (other than in dead time between switching) and the linearity of the control should be maintained even about the zero crossing.
Any help would be appreciated.