Here's the Science:
I created a spreadsheet to determine how the PIC timing needed set up. It gave me the absolute timing specs that I needed to get the performance I needed.

I used the max clock of my PIC: 20Mhz, coupled with the fastest clock division I could use (division of 1). The pic timers already use 1/4 of the clockrate as the timer clock so that gives us a max resolution of 5Mhz.

I also used the following specs to target my calculations: range of .5revolutions/sec to 30revolutions/sec



click here to download the .xls spreadsheet


The calculations show that I need 128 increments between changing the output to step the motors in full-step mode vs. 32 increments in 1/4 step mode. 1/8th step mode would be half of that, or 16 increments. Seeing that 16 clock increments leaves virtually no time for any calculation, debug output, or capture I opted to configure the motor controller for full-step mode and use 128 steps between toggles.

Here Comes the Proof:
I used my GDS-820C oscilloscope to check the output of my PIC stepper driver.

First I calibrated my code until I could specify a 2mS period waveform (1mS between toggles). I set up triggering on the rising edge. Here is the output waveform:




At 0V, the circuit generates the fastest waveform for the stepper motor (spreadsheet: clock incs/sec = 129) here is the waveform:




At Vdd, the circuit generates the fastest waveform for the stepper motor (spreadsheet: clock incs/sec = 7716) here is the waveform:



As you can see this deviates from the theory a little bit (theory: 19.44kHz, reality: 15.34kHz; theory: 324Hz, reality: 324Hz).

I think that is because the Analog-to-Digital conversion and math in the middle of the loop is slowing the program down a little bit. Though these values still give a range of .5rev/sec to ~25rev/sec which is pretty damn good as far as I'm concerned at the moment.

Meanwhile
Testing the circuit with the stepper motor, I can only get it up to 2.8kHz before the motor stalls. Also I have to baby the thing (ramping speed up gradually) so it doesn't stall. 2.8kHz is approximately 5rev/sec. Almost there...

Code (7/5/2006):
This compiles for the 18F2520 chip
PIC Program: main.c
Header File: stepper_driver.h
HEX File: main.hex