decay mode for inductive load current control

[brashear]

Hi,
I am trying to understand and figure out what is the best way to control the motion of a voice coil motor using current control. I am using a "hysteresis mode" feedback seen here:

http://www.cs.uiowa.edu/~jones/step/currentf/10.gif

First of all, why is hysteresis even needed? Is it because high frequency switching will increase losses?

For my main question: when the sensed current is larger than some reference value, the comparator output will be off. During this off period, the H-bridge FETs can be arranged to provide 1) slow decay, in which the inductor is essentially shorted and discharges through the resistance of itself and the FETs or 2) fast decay, in which the inductor is essentially run in reverse. Am I correct so far? The part I don't understand is how exactly does slow decay mode also create braking for the motor? If I had to use slow decay mode, would I not have to account for this extra braking effect as a sort of damping if I had to model the system for controlling it?
In fast decay mode, it's kind of like locked anti-phase drive right? How come in this mode there is no braking effect?
I have read a lot of articles online but can't seem to be able to answer these questions.

 

[davidbeach]

Is this for school?

 

[brashear]

if it was for school I'd ask my prof

 

[sreid]

In the mode where you short the coil the current is sustained longer but no Braking function happens. Can you use a Pulse Width Modulation control [PWM]?

 

[sreid]

Is your application high performance, high precision?

 

[brashear]

From what I've read, when you short the coils (slow decay), the current is sustained longer but there IS braking. From what I understand, there will be an induced current in the coils moving through the magnetic field of the permanent magnet, which will generate its own magnetic field in opposition to the field of PM. The induced current and thus the coupled magnetic field will be quite high because the resistance of the shorted coil is small.

In fast decay, I think the electromotive force induced by the moving field is being passed onto the source. Wouldn't this mean that the rate of decay of current would depend on the velocity of the motor?

I don't know what you mean by high performance, but precision is definitely important.

 

[sreid]

You are correct . if you short a coil that's moving through a magnetic field, a current will be induced that causes "Braking." However, if there is some kind of controller that's trying to keep the current at some value, it will apply whatever average voltage is required to the coil to keep the current at the correct value.

I'm a bit of an expert on voice coil;s and servo control of voice coils. By High Performance I mean High Bandwidth [starting with a high bandwidth current loop]. Often linear amplifiers are used.

 

[brashear]

Can you elaborate where linear amplifiers are used and how this would be beneficial? I am using a simple "chop drive", where a comparator compares the motor current via a current sense resistor and some reference voltage that the microcontroller sets. The MOSFET then turns on and off the current to the motor via fast decay mode, but it is operating in saturation and not linear. Thanks

 

[sreid]

The biggest objection to hystersis control is that the switching frquency varies and is often in the audible range.

The hysteresis ripple causes additional voice coil heating.

From your schematic it appears that you have to select positive or negative current. This may not be a problem for you but if you have to switch the current direction, a delay would be needed so a top and bottom transistor are not on at the same time. This results in a current dead band around zero current. This is also true for some PWM schemes.

Any switching amplifier creates electrical noise.

Linear amplifiers have none of these problems. The problem they have is that they get hot.

 

[brashear]

It's not critical whether the switching frequency is audible.

Coil heating is an important factor, and something I had been thinking about. Something I thought about though, is to install a variable output DC-DC converter, and limit the voltage during period when the force is constant.

I think what you are talking about with the top and bottom transistors being on at the same time is shoot through. Most H-bridge ICs have a delay where all FETs are off. However, this delay is very short (~200ns), and the FETs themselves have diodes to prevent being destroyed during this off time.

When you say linear amplifiers get hot, you mean the amplifiers themselves and not the voice coil, correct? I am guessing the way this control scheme works is the desired current is sent to the amplifier as a voltage. Since the amplifier is operating in the linear region, the drain current will be proportional to this voltage. Can these amplifiers handle ~3 amps? How about the cost when compared to hystersis feedback?

 

[brashear]

Well, I guess a variable output DC-DC converter would consist of a linear amplifier and there's no point in having both control schemes at the same time.