Someone wrote:
> Each servo you want to control must have a pulse of between 1 and 2 ms with a
 duty
> cycle of 18 to 20 ms. Your duty cycle timer must take account of the duration
 of
> the control pulse. I.E. after the counter for the pulse has expired you must
 also add
> the rest of the 2ms wait before entering the duty cycle timer.

In my experience the repetition frequency is of little importance when
testing servos. I have used, for some years, a program that pulses a pin
on the Centronics port of a PC for such tests. This repeats at the
"timer-tick" interrupt rate, which is 18.6 ticks/sec, or a 54 mSec
period. This is a perfectly adequate arrangement. You can speed up the
timer-tick if you want to.

The idea is to write a small TSR, to which the timer-tick interrupt is
redirected 18.6 times per second. The TSR generates the 1-2 msec pulse.
It gets its pulse-width data from a Basic or other program which runs in
the normal way; this program takes its input from the keyboard, and
passes it to the TSR.

In fact, instead of connecting a servo to the Centronics port, why not
connect the transmitter output stage to it, and use the PC as the
transmitter? Then you can program it in Basic. It's a bit bulky to carry
around of course, especially as you also need a Honda motor-generator to
power it all with.

John Blackburn.