The advantage of PWM is reduced power dissipation in the controller and the relatively easy digital generation of the signals.

The tradeoff are 1.) High electrical noise 2.) high stress on many of the switching components 3.) Lower ultimate resolution.

Zhang, unless you have some real budget for torque motors, look at using two drives, a cheap slewing motor system and the refined tracking system.

Assuming you are using an ALT-AZ scope, decide on the maximum size of the zenith blindspot to set your maximum drive dynamic range during tracking.

I would use linear drives for the servo drives themselves rather than switchers, or yuo can get really clever. Have a switched controller followed by a linear controller for the really tiny adjustments. It works very well.
Couple your slewing and tracking drives with a differential gearbox to the azimuth drives, and split your problem into driving two systems.

Or, if you only have a small scope (<50cm ) use microstepped driven stepper motors instead. Check out Mel Bartel's compu-dob project.

If you have any more questions email me at steve@astronomycentre.org.uk