Re: DMOV flag problem with soft motor

[Torsten Bögershausen via Tech-talk <tech-talk at aps.anl.gov>]

Hej Tine,

this smells a little bit like the
"done before even started" phenomenon.

The controller is commanded to move, execute is set, but it hasn't
started yet.

If this the case: We have added the
"waitNumPollsBeforeReady" feature, in the EPICS driver.

This may help:
<https://github.com/EuropeanSpallationSource/m-epics-ethercatmc/blob/master/ethercatmcApp/src/ethercatmcAxis.cpp>

(And today, we are using a different approach, with a bi-directional 
state machine inside the PLC-code. But that is a longer story)

BR
/Torsten




On 2022-09-28 16:57, Celcer Tine (PSI) via Tech-talk wrote:
> addendum: The delay is actually on the TwinCAT PLC between issuing the 
> EXECUTE command and the time of setting the DONE flag to “ACTIVE”.
> --
> Dr. Tine Celcer
> Controls Section
> Group Leader Integration and Support – Photon Science
> Group Leader a.i. Integration and Support – Large Research Facilities
>
> WBGB/013
> Phone: +41 56 310 46 29
> tine.celcer at psi.ch <mailto:tine.celcer at psi.ch>
>
> Paul Scherrer Institut
> CH-5232 Villigen PSI
>
>
> > On 28 Sep 2022, at 16:32, Celcer Tine (PSI) <Tine.Celcer at psi.ch 
> > <mailto:Tine.Celcer at psi.ch>> wrote:
> >
> > Hi all,
> >
> > We use a Beckhoff TwinCAT controller to move 3 independent axes, and 
> > it is integrated into EPICS via the s7 driver. Now I am trying to add 
> > the soft motor record support on top of that. Everything works fine 
> > except for a problem with the .DMOV field. The .DINP filed is linked 
> > to the DONE flag of the TwinCAT controller.
> >
> > When initiating the motion from “s7” EPICS PVs, the soft motor DMOV 
> > field behaves as expected, however when the move is initiated by the 
> > motor record I get the 1/0/1/0/1 pulse form the DMOV instead of the 
> > desired 1/0/1 (of course any user script monitoring the DMOV would 
> > capture that additional pulse). I notice that the first 1/0/1 pulse 
> > happens before the DONE flag changes to 0, and the remaining /0/1 part 
> > is in accordance with the processing of the DONE flag (as expected via 
> > ca link).
> >
> > I suspect that the reason is in the way the Twincat motion is 
> > implemented - one has to set the position first and then execute the 
> > “GO” command. I handle this with a sequence record that is linked to 
> > the OUT filed of the motor record. Could the fact that there is a 
> > delay between the time of writing to the motor record .VAL field and 
> > the time that the “GO” command is issued on the PLC and the DONE flag 
> > processes, cause the initial 1/0/1 pulse on the motor record .DMOV 
> > field (thinking there is no move - analog to sending it to the current 
> > position)?
> >
> > And if this is the case, is there a “timeout” I could set for that? Or 
> > any other ideas how to overcome this issue?
> >
> > ——
> > record(motor, "$(P):$(AXIS)") {
> >   field(DESC, "$(AXIS) soft motor")
> >   field(DTYP, "Soft Channel")
> >   field(OUT,  "$(P):$(AXIS)_SETANDGO PP")
> >   field(RDBL, "$(P):$(AXIS)_RBVCALC NPP")
> >   field(MRES, "1")
> >   field(RRES, "1")
> >   field(URIP, "Yes")
> >   field(PREC, "2")
> >   field(EGU,  "mm")
> >   field(TWV,  "1")
> >   field(DINP, "$(P):$(AXIS)_DONE NPP")
> >   field(RTRY, "0")
> >   field(NTM,  "NO")
> > }
> >
> > record(seq, "$(P):$(AXIS)_SETANDGO") {
> >   field(DOL1, "$(P):$(AXIS).DVAL")
> >   field(LNK1, "$(P):$(AXIS)_SETPOS PP")
> >   field(DOL2, "1")
> >   field(LNK2, "$(P):$(AXIS)_EXECUTE PP")
> > }
> >
> > ---------
> >
> > Thanks,
> >
> > Tine
> >
> >
> > --
> > Dr. Tine Celcer
> > Controls Section
> > Group Leader Integration and Support – Photon Science
> > Group Leader a.i. Integration and Support – Large Research Facilities
> >
> > WBGB/013
> > Phone: +41 56 310 46 29
> > tine.celcer at psi.ch <mailto:tine.celcer at psi.ch>
> >
> > Paul Scherrer Institut
> > CH-5232 Villigen PSI