Post by thread:[EE] Back EMF and brushed DC motors.

I'm currently doing a project (persistance of vision clock) and I'm experimenting with a couple componets and ideas that arent really needed for the clock, but could be useful in other applications. I will be using a brushed DC motor to turn the clock, and I was wondering about feedback from back EMF. Ive seen back emf being used by stopping the pwm output to the motor and measuring how fast the motor is going by measuring the voltage generated. But I was thinking more along the lines of something that would work without removing power from the motor. I was thinking something like putting an inductor in series with the motor, and watching for spikes/dips in the motors voltage, as the brush crosses the commutator. I dont know if the motor would momentarily go open circuit as it crosses the joint in the commutator, or if it would halve in impedance if 2 coils got connected in parallel for a short moment. I guess that would really depend on the motors construction. Anyone ever experiment with this or seen anything about it? all my google searches lead to either the method of turning off the motor output, waiting for the current to stabilize, and then measuring the voltage across the motor, either that or stuff about brushless DC motors. Either that or a way to make a stepper motor turn @ 1200rpm fairly quietly, I was thinking about a VFD (variable frequency drive) but making my own VFD with an output that is somewhat close to a sinewave to keep the motor quiet might be a bit much for me now. Thanks, Jonthan Hallameyer

Hi Johnathan, If I understand your application properly, I would think that you need to know absolute position in the rotational cycle, not just roational speed. If that is truly what you need to know, then I don't see how you are going to get it from motor back EMF or noise from the brushes. I'd recommend using either an optical or magnetic sensor to give you a single pulse as the motor passes some rotational reference point. By counting the number of these pulses per second, you can get speed. Then, you can make an estimator which resets itself to 0 deg when it sees the pulse and then counts forward from there based on time and the latest best estimate of rotational speed. As for back EMF sensing, I have never tried it. For an application such as this, I think measuring speed would be very doable because torque will be a monotonic function of speed. This means that the current in the motor will increase as speed increases, so that the voltage across the motor terminals (which is the back EMF plus I*R), will increase monotonically with speed. You could calibrate this and then use a lookup table or equation fitted to the data and get a pretty good estimate of speed. Assuming that your PWM is operated in such a way that there is an "on" period and an "off" period (where the motor is freewheeling), and that the motor inductance is high enough to keep the current fairly constant during the "off" period, then you could just take your measurements when you know that the PWM is in the "off" part of the cycle. Another possibility is that you might find that there is a very consistent relationship between applied PWM duty cycle and rotational speed and you might not need a measurement. The method of counting spikes of noise when the commutation happens is probably doable but would, as you say, depend on the motor construction. I am under the impression that most motors never totally break the circuit but rather short several windings together at commutation step points. Sean On 3/31/07, Jonathan Hallameyer <spam_OUTjmhtauTakeThisOuTgmail.com> wrote: {Quote hidden}> I'm currently doing a project (persistance of vision clock) and I'm > experimenting with a couple componets and ideas that arent really needed for > the clock, but could be useful in other applications. I will be using a > brushed DC motor to turn the clock, and I was wondering about feedback from > back EMF. Ive seen back emf being used by stopping the pwm output to the > motor and measuring how fast the motor is going by measuring the voltage > generated. But I was thinking more along the lines of something that would > work without removing power from the motor. > I was thinking something like putting an inductor in series with the motor, > and watching for spikes/dips in the motors voltage, as the brush crosses the > commutator. I dont know if the motor would momentarily go open circuit as it > crosses the joint in the commutator, or if it would halve in impedance if 2 > coils got connected in parallel for a short moment. I guess that would > really depend on the motors construction. Anyone ever experiment with this > or seen anything about it? all my google searches lead to either the method > of turning off the motor output, waiting for the current to stabilize, and > then measuring the voltage across the motor, either that or stuff about > brushless DC motors. > > Either that or a way to make a stepper motor turn @ 1200rpm fairly quietly, > I was thinking about a VFD (variable frequency drive) but making my own VFD > with an output that is somewhat close to a sinewave to keep the motor quiet > might be a bit much for me now. > Thanks, > Jonthan Hallameyer > -

Sean, I was planning on using a phototransistor and LED to index the LED array every time it went around on a rotation, and the method of pausing the PWM would work pretty well when the rotor is on the back side of the clock, I just wanted to see what I could do with brush noise for later robotics use. I have a plan which I could get the clock working with, just gotta overcomplicate it a bit and see if I can use that somewhere else. -Jon On 3/31/07, Sean Breheny <.....shb7KILLspam@spam@cornell.edu> wrote: {Quote hidden}> > Hi Johnathan, > > If I understand your application properly, I would think that you need > to know absolute position in the rotational cycle, not just roational > speed. If that is truly what you need to know, then I don't see how > you are going to get it from motor back EMF or noise from the brushes. > > I'd recommend using either an optical or magnetic sensor to give you a > single pulse as the motor passes some rotational reference point. By > counting the number of these pulses per second, you can get speed. > Then, you can make an estimator which resets itself to 0 deg when it > sees the pulse and then counts forward from there based on time and > the latest best estimate of rotational speed. > > As for back EMF sensing, I have never tried it. For an application > such as this, I think measuring speed would be very doable because > torque will be a monotonic function of speed. This means that the > current in the motor will increase as speed increases, so that the > voltage across the motor terminals (which is the back EMF plus I*R), > will increase monotonically with speed. You could calibrate this and > then use a lookup table or equation fitted to the data and get a > pretty good estimate of speed. Assuming that your PWM is operated in > such a way that there is an "on" period and an "off" period (where the > motor is freewheeling), and that the motor inductance is high enough > to keep the current fairly constant during the "off" period, then you > could just take your measurements when you know that the PWM is in the > "off" part of the cycle. > > Another possibility is that you might find that there is a very > consistent relationship between applied PWM duty cycle and rotational > speed and you might not need a measurement. > > The method of counting spikes of noise when the commutation happens is > probably doable but would, as you say, depend on the motor > construction. I am under the impression that most motors never totally > break the circuit but rather short several windings together at > commutation step points. > > Sean > > > On 3/31/07, Jonathan Hallameyer <jmhtauKILLspamgmail.com> wrote: > > I'm currently doing a project (persistance of vision clock) and I'm > > experimenting with a couple componets and ideas that arent really needed > for > > the clock, but could be useful in other applications. I will be using a > > brushed DC motor to turn the clock, and I was wondering about feedback > from > > back EMF. Ive seen back emf being used by stopping the pwm output to the > > motor and measuring how fast the motor is going by measuring the voltage > > generated. But I was thinking more along the lines of something that > would > > work without removing power from the motor. > > I was thinking something like putting an inductor in series with the > motor, > > and watching for spikes/dips in the motors voltage, as the brush crosses > the > > commutator. I dont know if the motor would momentarily go open circuit > as it > > crosses the joint in the commutator, or if it would halve in impedance > if 2 > > coils got connected in parallel for a short moment. I guess that would > > really depend on the motors construction. Anyone ever experiment with > this > > or seen anything about it? all my google searches lead to either the > method > > of turning off the motor output, waiting for the current to stabilize, > and > > then measuring the voltage across the motor, either that or stuff about > > brushless DC motors. > > > > Either that or a way to make a stepper motor turn @ 1200rpm fairly > quietly, > > I was thinking about a VFD (variable frequency drive) but making my own > VFD > > with an output that is somewhat close to a sinewave to keep the motor > quiet > > might be a bit much for me now. > > Thanks, > > Jonthan Hallameyer > > --