How to re-try on part size detection error

[JW]

In the event that a part is not present on the nozzle, by the time an up looking vision operation is performed (because it fell off, failed to pick etc, or even was picked end or side on), is itnpossible to simply discard and try again?

For simple passives, their cost is so low that is literally costs more in the time it takes to stop what you're doing and go to tend the machine.

Presumably with the extensive retry options available, this is a capability and I'm just missing where to configure it.

[Toby Dickenson]

Hi JW,

You are a few days too soon. I hope to be merging PR 1898 later this week.

https://github.com/openpnp/openpnp/issues/1891
https://github.com/openpnp/openpnp/pull/1898

Toby

> --
> You received this message because you are subscribed to the Google Groups "OpenPnP" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/openpnp/383814b2-a02d-4940-9418-b686eea8e70dn%40googlegroups.com.

[JW]

Hi Toby, fantastic - I've had a read through the issue discussion and added some comments in there directly. Great work, the feeder system has had some major opportunities for improvement for a while.

[Toby Dickenson]

Hi JW,

> Hi Toby, fantastic - I've had a read through the issue discussion and added some comments in there

I will reply here, because I think what you describe is orthogonal to that PR.

Since forever, openpnp has had two "error reporting policies":

"Alert" is what I use during job setup. Any job error causes the job
to pause, and an error message is shown. It is the operator's
responsibility to correct the error. When resumed, the job repeats the
step which caused the error. So you will get the same error again if
nothing has been done to correct it. Openpnp conveniently auto-selects
rows in the feeder and parts table, to assist in your manual
correction of the problem.

On github JW wrote:
> on a multi-nozzle machine, if a vision op fails and the part is discarded, the job cannot proceed without discarding all of the other parts on the other nozzles, as attempting to resume the job throws "No part loaded on nozzle".

So I guess this scenario means you are using "Alert", and you are
getting an error message box during vision. You then manually discard
the part, which changes the error but does not correct it. Resuming
the job is redoing the failed vision step, but now it fails because
that nozzle is empty. The next step in this workflow, after *manually*
discarding the part, is to *manually* feed and pick its replacement.
Resuming the job will redo the vision step and this time it should be
successful.

NB since the user is forced to correct all errors, then at the end of
the job (if you get that far!) the panel will necessarily have 100%
placements completed.

The alternative "error reporting policy" is "Defer". This causes the
error to be logged only. On a multi-nozzle machine the non-errored
placements are placed, then the nozzles are cleaned by discarding the
errored placements. This is all automatic. At the end of the job it
output all the error messages into the log for a second time, and
reports that the job is finished but less than 100% complete. This is
good for keeping the machine working without waiting for operator
input. Since forever, these two error reporting policies have been a
per-placement setting. In version 2.3 we added the "bell" button to
toggle. My habit is to toggle onto "alert" (bell) when I am next to
the machine, and "defer" (no bell) when I walk away.


So let me re-answer your initial question:

>> is it possible to simply discard and try again?

Yes!
Since you are using the "alert" error reporting policy (bell) then you
need to manually discard the part, and manually re-pick a replacement.
Having the same done automatically in "defer" mode is PR 1891.