Re: [CAD_CAM_EDM_DRO] axis clamps/brakes was Re: Measuring backlash more accurately?

ballendo wrote:

>Juha,
>
>My reply on this was thinking steppers, even though you mentioned
>servos... I believe the rutex drives close the loop within the drive,
>like geckos? In this case, I think you can treat the system like
>steppers...
>
>Jon, Won't the "violent jump" be limited to the "hunting" error/size
>in these type servo drives? I would think that clamping an axis
>driven by these would be nearly the same as clamping a stepper axis?
>
>

It depends on the amount of backlash between the motor and encoder as to
how many
encoder counts can build up with the axis clamped. If the encoder is an
integral
part of the motor, it may be no problem. If the encoder is on the
leadscrew, and
the motor drives it via a belt reduction, it could cause all sorts of
problems, such
as overheated drives and motors. I know that in my system, which uses
the above
mechanical setup, that if the motors are locked, the servo amps will
build up to
high current in about one second and then cause an overcurrent trip.
This is because
the open-loop gain is very nearly infinite, and it requires the closing
of the loop
through the mechanical coupling to keep the system stable.

I also know from texts that any discuntinuity or abrupt non-linearity in
the transfer
function of any component in the system can lead to instability. I
actually ran
across this in a precision rotor speed control in my photoplotter. I
didn't have
a negative power supply for the servo amp, so it could accelerate the
motor but
not brake it. (The motor's back-EMF was less than 3 V, and Darlington
transistors
were used, so that back-EMF was not enough to sink current to ground.)
It caused bursts of instability that causes the whole system to go haywire.
When I changed the connection of the - terminal from ground to -5 V for
the servo amp, the system became VERY stable, because the discontinuous
transfer function was fixed.

Jon