steppermotor record problems

[Len Lawrence <L.Lawrence@roe.ac.uk>]

At the Royal Observatory Edinburgh we have a problem in driving
several motors at once via the EPICS "steppermotor" record and the OMS 
8-axis stepper motor drive card. We are being advised by our colleagues 
at the Royal Greenwich Observatory to dump the record in favour of the 
"motor" record which is in use at APS. Have others experienced similar
problems?

The motors we are driving control (currently) four mechanisms within the
Michelle imager/spectrometer being built here for use on UKIRT and
Gemini. The datuming of the mechanisms involves "stop on microswitch"
operations (steppermotor STHM to 2 or 3) and also "binary search" motions 
in which a mechanism edges towards its datum microswitch whose state is 
sensed after every move. Control of the sequencing involved in these
searches is through SNL code triggered by EPICS records, specifically
Gemini CAD records.

We find that mechanisms datum reliably individually or in pairs, but
that attempts to datum four mechanisms at the same time invariably fail
unless successive move demands are issued from the SNL at a very slow 
rate (< 2Hz). Three motors at once usually works, but sometimes doesn't.
When a mechanism fails (not always the same mechanism or the same state)
it appears that either the expected move does not start, or else the current
motion does not complete.

After exhaustively walking through the SNL code, and peppering it with
diagnostics, we are reasonably confident that this is not a case of the
variables within one state set trampling over those of another.

Len

==============================================================================
Len Lawrence                       Tel.   : +44 (0)131 668 8255 (direct)
Royal Observatory                           +44 (0)131 668 8100 (switchboard)
Blackford Hill                     lcl@roe.ac.uk
Edinburgh EH9 3HJ                  aka lcl@tarazed.demon.co.uk
UK                                 WWW    : http://www.roe.ac.uk/lclwww/