Microstepping results vary markedly with the motor characteristics.
I am getting excellent results by PWMing between two static stepper
states and slowly varying the proportion of time that one or other of
the codes is applied. I am driving a small (similar in size to those
in OLD small hard disk drives) (NEMA 17 frame = 1.7") with a ULN2803
open collector driver and small series resistors (10 ohms ???) in
each motor lead. This gives a crude approximation to current drive.

Results are amazingly good - the motor moves smoothly between static
positions at low speed - some resonance is visible depending on
load - I am driving a pointer and I have to ensure that it isn't
resonant. The sound is interesting - you can hear the analog signals
that the PWM is simulating overlaid with intriguing beats and
harmonics.

The motor I am using comes from RS (RS440-436) (about $US40!) but I
found that a similar motor from Farnell is much less satisfactory.
There are many motors surplus from US suppliers but the
characteristics of these will be uncertain. Start with a 1.8 degree
unit (the smaller the step the better) - you can get less than this
but they are uncommon. Some manufacturers specifically spec their
units for Microstepping applications. Have a look at their data
sheets for the dark arcanery involved (field strength and shape with
distance from rest position. Permanent magnet versus hybrid versus
...).




Russell McMahon

From: Lawrence Lile <lilel@toastmaster.com>
>I've been workoing on a stepper motor project.  The challenge is to
take it
>a little ... ah.. STEP  into the microstep domain by doing two-bit
current
>control.  (Using a TEA3718 gives three steps of current and one OFF,
which
>is a little like microstepping current levels.)   I can get about
1/6th step
>out of this thing, so far so good.
>
>But I notice that I can't step SLOWLY.  If I have too long a wait
between
>steps, the motor just sits there and sings.  (Stepper motors can
make a lot
>of noise.)   Any ideas why ya can't go slow with these things?