Dead zone in stepper power curve

I was wondering if anybody had encountered this problem. I'm
developing my own stepper controller system and experience the
following...

when I run a smooth speed ramp, there are certain speeds at which
the motor doesn't like to run. Not fast speeds, but slow ones. I'm
running assembly on a microcontroller, and the pulses are very
accurately timed.

Depending on how I set the winding power (it's a chopper drive), I
can tune out these dead spots. Usually by decreasing the power.

Do you think this is a resonnance problem? Or is it more likely to
be a problem with my code?

I've scoped the waveforms and they are uniform and glitch free
through the problem speed zone.

Any thoughts?

Thanks,
Ted