Re: These servo motors worth a darn?

Jon Anderson wrote:

> I'm not concerned with physical accuracy here. Surely having the encoder
> directly attached to the screw is optimimum for accurate measurement of
> what the screw is doing. What I am not sure of is if having the motor
> and the encoder separated by a timing belt might introduce any sort of
> servo error. If there is even the slightest wrapup or lag between the
> belt and screw (most noticable when accelerating a heavy load from a
> stop) will the controller see no motion for the given output and start
> ramping in more current than really required. This is a big difference
> between stepper and servo systems. The encoder tracks motor velocity as
> well as position.

I have used such a system on my mill, and had no particular problem with
it. No matter how things are coupled, there is some springiness in everything,
leadscrews, belts, shaft couplings, motor shafts, etc. All these things have
their own natural frequency, and when you assemble the whole system, it
adds even more natural frequencies, and the system performance becomes
quite complicated. But, as long as the more significant vibrational modes
are above the bandwidth of the servo system, they don't cause much trouble.
If you put the motor several feet from the shaft, with a very long belt coupling
them, then you could see a problem. I have the feedback DC tach on the
leadscrew, too, so it is on the other side of the belt from the motor, as well as
the encoder. The velocity feedback bandwidth is about 1300 Hz on my system,
while the positioning loop is limited to less than 500 Hz by the 1 KHz sampling
rate of the EMC servo code. This bandwidth is probably excessive, too, but
I wanted to see how far I could push it.

Jon