Re: lost steps

Jon,

Good evening and Happy New Year to you! The direction input may seem
strange but it really is not.

The step pulses are processed (delayed) by the midband stabilization
circuit. A little bit of theory is need to explain why.

Step motors are first-order mechanically resonant, (mass-spring)
systems. This simply means there is an inheirent 90 degrees phase lag
between between velocity and torque.

Think of a coil spring suspended from the ceiling with a weight
attached to the end. Leave it alone and the weight has a rest
position. Now, pull down on the weight and then let go. The weight
will oscillate up and down.

When the weight passes its former rest position it will have maximum
velocity and the spring will have zero force. When the weight reaches
the top and bottom extent of its travel, it's velocity will be zero
and the spring will have its maximum force. That is what I mean by a
90 degree phase lag between velocity and torque (force).

At low speeds a drive presents itself to the motor as a current
source. As you know, current mode amplifiers have a near zero phase
lag and all is well with the motor.

As speed increases, the current source runs out of "head room" and
reverts to a voltage mode amplifier. Voltage mode amplifiers have an
implicit 90 degree phase lag of their own.

As a system, the motor and amplifier phase lags total 180 degrees at
this point. That is a setup for positive feedback and unconstrained
oscillation if the system gain is above unity (it is).

The G201 drive introduces a phase lead component of about 60 to 80
degrees to the system loop to stabilize it.

This done by sensing the rate of motor load change (the first
derivative of torque) and then phase modulating the step pulses with
this sensed component. Phase modulation means you slightly advance or
retard the timing of each received step pulse so as to damp out
oscillation (loop phase lag of 120 degrees or less).

The way it works out is from the time a step pulse is sent until it
is "clocked in" by the drive can be a variable period of time between
5uS to 20uS. As long as the direction is "true" at the active edge of
the step pulse until the next active edge, the step will be taken in
the correct direction.

Mariss

> --- In CAD_CAM_EDM_DRO@yahoogroups.com, Jon Elson <elson@p...>

wrote:

> > Gecko drivers have a strange requirement for the direction

signal.

> It
> > must stay in the right
> > state AFTER the step pulse is issued, until the step actually

goes

> out
> > to the motor. With
> > microstepping and step multiplier drives, that can be a long

time,

> > depending on the conditions
> > at that moment.
> > Most other step drivers require the direction signal to be stable
> some
> > microseconds before
> > and hold a microsecond or so after the leading edge of the step
> signal,
> > and then it can
> > change. This causes trouble with some programs. I had to modify
> my
> > stepper controller
> > board to accomodate these drives. It has a flip flop that

latches

> the
> > direction signal when
> > the step pulse occurs and holds it in that state until the next
> step pulse.
> >
> > Jon