Experimental Physics and
Industrial Control System

Till Straumann <[email protected]> · Thu, 15 Apr 1999 14:38:12 +0200

Hi.

We are using motorRecord (a locally modified version 3.4) with external
readback
(via link) enabled.

The OMS 58 controller overflows at  +/- 33.554.432 steps, which seems to
be
a lot. However, we are actually using almost the full range.

If external readback is enabled, the motorRecord is using relative
motion commands
causing step count register drift and occasional overflows. When this
happens, things
go haywire.

My fix (only active if URIP and RDBL present) calculates the true
position based upon
the readback value when the motor stops and writes that value to the
motor hardware.

Cheers, Till.


---------------------------------------------------------------
This patch fixes:

 - if an external readback link is used, the position should be written
back
   to the controller hardware to avoid drift which may overflow the step
register.
   by T. Straumann (PTB)

Index: motorRecord.c
===================================================================
RCS file: /opt/csr/CVS/till/epicsApps/records/motorRecord.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -r1.3 -r1.4

--- motorRecord.c   1999/01/28 15:08:38 1.3
+++ motorRecord.c   1999/04/01 16:33:38 1.4
@@ -937,6 +941,26 @@
                if (pmr->mip & MIP_DELAY_ACK || (ticks <= 0)) {
                    pmr->mip &= ~MIP_DELAY; MARK(M_MIP); /* done
delaying */
                    (void)maybeRetry(pmr);
+                   /* T. Straumann 4/1/1999:
+                    *   If we using an external readback, we should
+                    *   write the position back to the motor hardware.
+                    *   E.g. the OMS VME58 controller overflows at
+                    *   +/- 2^25 steps. With an external readback,
+                    *   steps may be lost which may cause drift of
+                    *   the position value maintained by the
controller.
+                    *   Eventually this may cause overflow.
+                    *
+                    *   We happen to use almost the whole range of
steps;
+                    *   therefore, after running for some time an
overflow
+                    *   occasionally occured.
+                    */
+                   if ( pmr->urip ) {
+                       double  newpos = pmr->dval / pmr->res;
+                       /* dump the position to the controller */
+                       INIT_MSG();
+                       WRITE_MSG(LOAD_POS, &newpos);
+                       SEND_MSG();
+                   }
                } else {
                    int status;
                    pmr->mip |= MIP_DELAY_REQ; MARK(M_MIP);



Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System