Re: motorRecord with phytron support: record send goto 0 command before reference is done

[Kevin Peterson via Tech-talk <tech-talk at aps.anl.gov>]

Bernhard,

With ACCL=2.0 the desired acceleration would be 410 steps/s^2, but the 
driver hard-codes the minimum acceleration to 4000 steps/s^2:

https://github.com/epics-motor/motorPhytron/blob/master/phytronApp/src/phytronAxisMotor.h#L21

The minimum acceleration is enforced here:

https://github.com/epics-motor/motorPhytron/blob/9acb974257dc61512017f18f9b823be0495e948d/phytronApp/src/phytronAxisMotor.cpp#L705-L706

I thought the problem might be due to an acceleration time that is 
significantly longer than the poll period, which might result in the 
controller still decelerating when EPICS sees the limit is hit, but that 
doesn't explain the behavior you're seeing.

Does the motor record that fails to home properly have a nonzero 
backlash distance (BDST) or a non-zero retry count (RTRY)?

There are only four places in the motor record code where moves are 
commanded.  One is where normal moves and retries are commanded.  The 
other three are related to backlash correction.

Kevin


On 3/29/22 09:06, Bernhard.Kuner wrote:
> Hallo Kevin,
>
> thank You for the help to understand what happens in the motorRecord,
> nasty  beast this record!
>
> The homing type is here on the positive end-switch "Mn.m+". Poll period
> 0.5s but ACCL=2.0
>
> There was a bug in the phyMotion software if you use the positive
> end-switch, but there is an update from phytron and now it works if you
> give this command by the phyLogic software.
>
> Ok, You say the record recognises a motor action to be done by
> pC_->motorStatusDone_  This is set in the poll routine by the command
> "Mn.m==H". In the log you can see that there is no "M1.1==H" with the
> answer "E" for done after hitting the end-switch and before the new move
> comand occurs. So we suppose a problem in the record.
>
> By the way, we have a support for the phytron IO modules. There I
> introduced support for stringout -in Record to communicate with the
> motor controller. If you send a home command this way, it works as 
> expected.
>
> Bernhard