Back EMF sensing?

Thread Starter

lagfish

Joined Jun 12, 2011
6
Hi,
I am trying to measure the velocity of a voice coil actuator by using back EMF.The coil is controlled using a PWM scheme. The conventional method of sensing is to stop the current going into the coil by switching off a transistor, and wait for the energy to dissipate through a flyback diode or some sort of snubber circuit, then measure the back EMF. However, for my application, I can't wait for the energy to dissipate because it takes too long. My question is, can I just put a resistor in series with the flyback diode then measure its voltage drop after a fixed time after switching off the transistor to infer the velocity? The voltage drop across the resistor should be a function of the initial current through the coil, which is fixed, and the back EMF of the actuator, correct? Essentially my reasoning is that if the actuator is moving fast through a magnetic field, the energy stored inside the coil will be dissipated faster.

Does this make sense to anyone? Thanks
 

Papabravo

Joined Feb 24, 2006
21,094
My recollection of trying to do this for the arm of a hard disk drive was that the signal is way too noisy to be of much use without heavy filtering. As we know heavy filtering destroys the information that we need to do the job. That was then (1980), and this is now so things may have changed.
 
Last edited:

Thread Starter

lagfish

Joined Jun 12, 2011
6
Another idea: what if instead of using PWM I used a transistor in linear mode. This way we can eliminate the Ldi/dt portion of the voltage and cancel out the IR portion from using a current sense resistor, then just left with the back EMF voltage? If we're using PWM and just dissipating the energy through the coil itself via freewheeling diode, the efficiency should be similar right?
 

Thread Starter

lagfish

Joined Jun 12, 2011
6
It's not an ideal inductor so it has a voltage drop. This is countered by the back EMF voltage due to the movement of the coil through a magnetic field. Is this not right?
 

John P

Joined Oct 14, 2008
2,025
What if you had a measurement phase of operation, where instead of trying to kill the current completely, you used some sort of electronic system to keep the current constant over the measurement interval? Then you'd have 2 components of voltage; one would be the IR drop, which you can calculate because you can measure the current, and you should know what the resistance of the coil is. The other component is what you need to figure out, and it's going to be equal to the velocity times some constant, which you should also know, or at least be able to measure. So there would be the velocity. Does that make sense? (But you might need to deal with the change in resistance if the coil heats up significantly.)

This ought to work better than doing the same thing with a motor, as you don't need to deal with changes in the parameters as the motor rotates, and of course brush noise. With a voice coil, it should all be linear.
 

Papabravo

Joined Feb 24, 2006
21,094
It's not an ideal inductor so it has a voltage drop. This is countered by the back EMF voltage due to the movement of the coil through a magnetic field. Is this not right?
The resistance of a typical inductor might be measured in milliohms so it takes a large current to produce an appreciable voltage drop. Large voltages are the result of large values of di/dt. This is part of the problem.
Go ahead and try this if you want, but I'm betting you will not be happy with the results. We ended up using embedded digital information from the surface of the disk between the sectors as our feedback mechanism. That actually worked quite well.
 

Thread Starter

lagfish

Joined Jun 12, 2011
6
John:
That was kind of my idea with using a transistor in linear mode. I don't think the current even needs to be constant as long as we know what it is. The problem is the resistance changes with an increase in temperature. The voice coil is expected to operate -20 to +70C, so will see more than 15% change in resistance.

Papa:
The voice coil I'm using has a DC R of 6Ω so it may be alright.
 

crutschow

Joined Mar 14, 2008
34,201
Perhaps I'm missing something, but can't you just cut off the voltage to the voice coil and then measure the open circuit voltage? The open circuit voltage should be a measure of the voice coils movement. If the initial inductive spike from the speaker inductance is a problem, a zener diode will rapidly dissipate that energy and leave the small voltage from the voice coil movement.

Interestingly, I have a Rythmik Audio subwoofer speaker that has a two windings - the normal voice coil winding where the power is applied, and a secondary winding with much finer wire. The secondary winding voltage is a indicator of the voice coil movement and that is used in a servo negative feedback loop to reduce speaker distortion from non-linear speaker movement and also extend the low end frequency response. Seems to work quite well.
 

John P

Joined Oct 14, 2008
2,025
If you do what you say you can't do, and kill the current, then the resistance of the coil is no longer a factor. If you have to keep the current flowing, I don't see how you can avoid dealing with the resistance. You could put a temperature transducer near it if it's ambient conditions that set the temperature rather than self-heating of the coil.

I'm not sure about measuring a changing current. If the current is constant, then I'd expect one term to drop out of the sum of voltages that make up the total.

i.e. Vtotal = IR + Ldi/dt + Kv
where small v is velocity

It seems as if you'd need to know how fast the current is changing. It should be possible, but it would be simpler not to do it.
 

Thread Starter

lagfish

Joined Jun 12, 2011
6
Crutschow: What would be the purpose of using a zener diode vs a schottky diode? The back-EMF on the voice coil is in reverse-bias direction of the diode isn't it?

John: you're right, I don't know how I missed that.

I am looking into using a thyristor as opposed to a regular resistor to reduce L/R time. Anyone can provide any insight? Thanks
 

crutschow

Joined Mar 14, 2008
34,201
Crutschow: What would be the purpose of using a zener diode vs a schottky diode? The back-EMF on the voice coil is in reverse-bias direction of the diode isn't it?

.............................

I am looking into using a thyristor as opposed to a regular resistor to reduce L/R time. Anyone can provide any insight? Thanks
You could use a diode to absorb the inductive transient as long as the signal you want to see is less than the forward diode drop. Why you want a Schottky?

Don't see how you can use a thyristor since thyristors do not turn off until the current drops to zero.
 
Top