Re: Cheaper servos

Phil Plumbo wrote:

> From: psp@... (Phil Plumbo)
>
> >Sorry, but I think you're nuts!
>
> Well, Jon, you're not the only one...no hard feelings. I wanted to generate
> some discussion. I'm happy to play the role of the guy that asks the stupid
> questions to that end, if you all won't mind.

Glad you didn't take offense.

> >Servo motors are designed to have minimal
> >'torque ripple' and 'velocity ripple', meaning that with a constant DC voltage
> >applied, they provide constant torque if the speed is held constant, and
> >turn at constant speed when free to turn. Cordless drill motors, etc.
> >are not built this way, and won't give smooth motion.
>
> I see, so we rely on the motor characteristics as well as positional
> feedback? My recollection of the gist of the Byte article was that with
> positional feedback, even cheap motors can give accurate and repeatable
> positioning. For a robot. My question is really, "if OK for a robot, why not
> for a home shop machine tool?"

Well, what does the robot do? If it only needs to pick up a 3" wheel and
put it on a conveyor belt, it doesn't need much accuracy or smooth motion
at all. If it is putting IC's on a circuit board with .01mm accuracy required,
that is another matter. I'm doing fairly precise machining, and don't ever
want to preclude some really hairy work, like making a steam turbine or
jet engine blades. So, I want very good accuracy and precise motion.
Now, the 'lumpy' servo motors would work fine for positioning, but
for smooth motion, the CNC control would have to compensate too much.

> I'd like to know more about how a DC motor is constructed to have these
> desirable ripple characteristics.

One of the big tricks, for slotted armatures, is to 'twist' the armature laminations
as they are pressed onto the shaft, so that the slots are not in a straight line,
but have a twist to them of about 1/n. So that at the radial position where
one slot starts at one end, the next slot ends at the other end. This way,
the pole between the slots does not break out of the field flux all at once,
and the wire in the slot doesn't, either. Now, for the finest servo motors,
they eschew slots entirely, and the wires are wound onto a smooth
mandrel. This is the way many DC tachometers are made. But the
whole aim here is to have the same number of current carrying wires
in the field flux at every possible position of the motor.

> >My servo system
> >moves smoothly down below .01"/minute, and up to 105"/minute.
> >With a little higher DC voltage, it could go faster.
>
> Smooth is good, and the range is impressive... I'm still not clear on why,
> with good positional feedback, we can't use cheaper motors.

Well, maybe you can. But, at some speed, the torque or speed ripple could
come out at the smae frequency as some resonance in the system, and then
it would excite oscillation. Now, maybe you can know that these frequencies
will never match up, but on a retrofit, it is hard to know.

> Do you have any suggestions as to which motors might possibly be acceptable?

You just have to test them.

Jon