Erna, thank you very much. The following really helped me grok it better: "So the motor is designed to keep the torque constant over 60°. DC has no phase angle. There is just one closed loop running one current. If you switch the loop from phase to phase, that doesn't make DC an AC."
That implies there was a concerted effort to design the structural layout of the windings and magnets, to keep it steady through that 60° segment, for the purpose of making the motor phase management electronically as simple as possible.
I presume the trapezoidal labelling just comes from the current waveform to energise and de-energise the windings of each segment. ie: The voltage switchover can be a sharp box shape.
Let me say it this way: It is the state of the art. And if we remember: Chip made the Propeller different, as the state of the art is just old wine in new bottles.
If anybody is interested in finding the rotor position at standstill he may digest the attached document ;-)
I presume the trapezoidal labelling just comes from the current waveform to energise and de-energise the windings of each segment. ie: The voltage switchover can be a sharp box shape.
It is even worse. There is no trapezoidal current waveform. The current is constant, it is just commuted from phase to phase, and as there is inductance, the current rises fast, but limited due to the BEMF and sinks fast, as now the BEMF helps. Commutation takes 3-4 PWM Cycles max. What is "trapezoidal" is the voltage at a phase: you apply positive voltage for 120°, then the terminal is open for 60°, you measure "the BEMF", with acutally is the BEMF times the angle, or just the angle, and then you apply a negative voltage for 120°, then the terminal again is free and the voltage will ramp up.
And: when I say: you apply a voltage I do not necessary mean a "clean" analog voltage, it can be, and mostly is, PWM, either symmetric or asymmetric. So the voltage is always "box shaped". The floating pin is a sensor output, not a "voltage".
This is how actual current looks like. The current is not perfectly flat, as the BEMF is not perfectly constant, that is, if the BEMF goes a little down, more of the applied voltage comes across the inductance and so current rises. But in the same moment, this additional current still creates less torque. This is due to the fact, that for a given motor the motor constant both determines how much BEMF is created per rotation/sec and how much torque is created per Amp. It is one constant, if you find two in the documents, this is due to the fact, that A is Coulomb/second, while rotation is counted per minute.
@ErNa said:
... What is "trapezoidal" is the voltage at a phase: you apply positive voltage for 120°, then the terminal is open for 60°, you measure "the BEMF", with acutally is the BEMF times the angle, or just the angle, and then you apply a negative voltage for 120°, then the terminal again is free and the voltage will ramp up.
Got it, thanks. And reading that linked TI PDF also shows that.
Pity the TI PDF doesn't compare using an AC motor. Instead they seem to go into using a DC motor in more complex ways. EDIT: I ask this because industrial servo motors went AC decades ago. Basically as soon as the DSP became a thing all the big names quickly dumped hall+encoder and went to resolvers.
Comments
Hi
This is a good explanation- if you can get your head around it...
https://www.ti.com/lit/ml/slyp711/slyp711.pdf
dave
Erna, thank you very much. The following really helped me grok it better: "So the motor is designed to keep the torque constant over 60°. DC has no phase angle. There is just one closed loop running one current. If you switch the loop from phase to phase, that doesn't make DC an AC."
That implies there was a concerted effort to design the structural layout of the windings and magnets, to keep it steady through that 60° segment, for the purpose of making the motor phase management electronically as simple as possible.
I presume the trapezoidal labelling just comes from the current waveform to energise and de-energise the windings of each segment. ie: The voltage switchover can be a sharp box shape.
Let me say it this way: It is the state of the art. And if we remember: Chip made the Propeller different, as the state of the art is just old wine in new bottles.
If anybody is interested in finding the rotor position at standstill he may digest the attached document ;-)
Tra> @evanh said:
It is even worse. There is no trapezoidal current waveform. The current is constant, it is just commuted from phase to phase, and as there is inductance, the current rises fast, but limited due to the BEMF and sinks fast, as now the BEMF helps. Commutation takes 3-4 PWM Cycles max. What is "trapezoidal" is the voltage at a phase: you apply positive voltage for 120°, then the terminal is open for 60°, you measure "the BEMF", with acutally is the BEMF times the angle, or just the angle, and then you apply a negative voltage for 120°, then the terminal again is free and the voltage will ramp up.
And: when I say: you apply a voltage I do not necessary mean a "clean" analog voltage, it can be, and mostly is, PWM, either symmetric or asymmetric. So the voltage is always "box shaped". The floating pin is a sensor output, not a "voltage".
This is how actual current looks like. The current is not perfectly flat, as the BEMF is not perfectly constant, that is, if the BEMF goes a little down, more of the applied voltage comes across the inductance and so current rises. But in the same moment, this additional current still creates less torque. This is due to the fact, that for a given motor the motor constant both determines how much BEMF is created per rotation/sec and how much torque is created per Amp. It is one constant, if you find two in the documents, this is due to the fact, that A is Coulomb/second, while rotation is counted per minute.
Got it, thanks. And reading that linked TI PDF also shows that.
Pity the TI PDF doesn't compare using an AC motor. Instead they seem to go into using a DC motor in more complex ways. EDIT: I ask this because industrial servo motors went AC decades ago. Basically as soon as the DSP became a thing all the big names quickly dumped hall+encoder and went to resolvers.