Re: Cheap Motor controll

"Arne Chr. Jorgensen" wrote:

>
> First PMW versus analog:
>
> A PWM bridge is what you would like for a large motor, - it is far
> more efficient. An analog would have to take up the same wattage as
> the motor itself - and then have a good heatsink.
>
> Let say you just wanted a small motor, - say a "dremel" tool. If
> you feed the +-10V to a power opamp, - or a transistor buffered
> opamp, then you wire this up as a closed loop voltage fashion. A dc
> motor will have an rpm according to the voltage you supply it. The
> rpm may drop a bit because of load. A bit more electronics, and you
> can compensate for this rpm drop. So a opamp and a power
> transistor, you have a simple servo controller for a few dollars. (
> Yes, it is a servo ! The one used in a tape recorder or a tune
> table - but the often use a velocity feedback. Small tape recorders
> often use a motor with a tachometer in the same housing )
>
> Okay, - this is an analog type of motor control, - for a just a few
> dollars. You can also use a PWM bridge, - adding a few dollars.
> Now it is a little harder to have a voltage feedback. Add a few
> dollars - and you can have that too, without any tacho generator.
> But the best performance would be with a dc-tacho generator for
> feedback.
>
> All I have just said, is for a small motor - like a "dremel" tool.
> But add larger transistor output stage, and it can turn big motors.
> This would add more dollars to it - until you reach designs in Jon
> Elson's category.
>
> But back to the very simple design - you could use it to control a
> "dremel" tool, or small motor. If you add an encoder feedback, -
> then you will close the loop for position. The EMC will handle
> that, so this very simple design could be used as a simple
> interface to EMC. It will stop in position, and crank up as others,
> - and if the load don't vary much, it could run on several axis.
>
> The dc feedback tacho - is used to keep a constant velocity. Say
> 0-10V is equal to 0-100% on your rpm scale. Output 8 volt, and the
> motor should output 80% rpm. The EMC will home it in to the correct
> position, so way bother if the RPM drops some % ? Why use a dc
> feedback ?

Because you want the machine to be at the right position at all times, not
just the start and end of the move. I built a similar voltage amplifier circuit
to quickly drive my machine while I worked on the PWM servo amp.
The performance was AWFUL! The following error was at least .002"
per IPM of the move. That means that the machine had an error of 1/8th
of an inch at 63 IPM! You can't make any parts with that! The following
error was about .020" even with the machine motionless!

Clearly, a simple voltage amplifier won't cut it!

> Let say you would like a 45 degree motion on your X & Y. If you
> output 50% rpm to both motors, then it will move in a straight
> line, but say if your X axis drops to just 30% of the commanded rpm
> - what then ? You will not be moving in a straight line, - the Y
> motor will reach it's position before the X motor catches up - but
> it will zero out to it's commanded position.
>
> A lot of commercial motor-amps, have compensation for velocity drops
> and for a lot of uses, - you don't need any dc - tacho. A large XY
> table or different loads, - may not work so good.
>
> But for a little test - we could make a "stupid simple" motor
> control for less than $10, and it could work okay too. If we added
> a simple counter, then the RPM on this could use the stepper
> interface too. This could make the interface you would need for
> controlling the RPM on a router or something too, without any STG
> board. This little setup could interface to a 3phase motor
> controller, - from the simple par-port.
>
> A "real good" motor-amp, can be made for less than what Jon Elson
> has designed. I don't think he would mind to adjust his design a
> bit either. If you wanted something cheaper, he could drop the bar
> led display and maybe use smaller output transistors if you don't
> have the power need. This would save you some dollars. If he use
> the same boards, you could add the bar graph LED later on - if you
> think you would need it.

Sure, just leave those parts out! But, then, when you go to tune it up,
you'll really want the info display, or else you'll have a DVM hooked up
all the time.

> So, there is endless of possible solutions, - the main thing is to
> get answers as to what your needs is. What motor size, what use,
> and so on.
>
> I have not studied the EMC code, - but Fred may have put in some
> velocity control loop. In other words, it might work well with a
> less than perfect motor-amp. That is, you may not need any high
> quality dc-tachometer, and Jon Elson states that he has use a small
> dc motor instead.

It COULD be done, but it has not. The STG card's ADC could be
used to read the armature current, and the encoder counter could be
used to compute the velocity (although it needs to be filtered or
de-glitched, as we now know the LS7166 encoder counter chips
occasionally read glitched data into the CPU).

Jon