
> Change the program code so that the selection of the “manual”, “houming” mode takes effect without restarting the program.
Sorry I don't understand. I need a longer explanation how it
should behave in what case.
However, I feel that the real problem is the pipeline. You need
to make it more robust, so that failures almost never happen. I'm
confident this can be achieved.
I consider calibration as part of homing a very good practice and
failure should also fail the homing cycle. That's sort of a
pre-flight check for the machine and vision in particular, same as
visual homing.
If you "postpone" the vision problem it will instead fail in the
middle of a job where it is much more inconvenient, or could even
lead to spoiled parts/boards.
_Mark
--
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 on the web visit https://groups.google.com/d/msgid/openpnp/b0ebf4e8-7c6b-479b-b37e-4603f0b06eb0n%40googlegroups.com.
After 6 you could home again. No need to restart the program.
Maybe I'm still missing something?
_Mark
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/85e3b288-0157-4209-88df-aa89b342a4a5n%40googlegroups.com.
Hi Tony
We could create a new calibration setting called
"FirstNozzleUse". It could postpone calibration to the first use
of the nozzle/tip that matters, e.g.
before the first pick or before the first nozzle tip unload.
But I'd like to keep the MachineHome the way it is.
I wrote this logic and my intention was to make the calibration a part of the homing cycle. I have 0.2mm run-out on that nozzle tip holder*, I guess you can see that running the machine that way is a no-go. Before the nozzle is calibrated, the machine is simply not considered homed.
The same applies to visual homing. Only the homing fiducial is
stable on my machine, the Liteplacer has the end-switches mounted
with only one screw plus they are in harms way when a crash
happens, so you can never rely on end-switches on a Liteplacer. So
if visual homing fails, I equally consider the machine unhomed.
Like I wrote earlier, this extended homing cycle is sort of a
pre-flight check for the machine. If it does not work due to
pipeline problems, it is a good thing the machine would not
continue. You then need to solve the pipeline problem before
anything else.
Many things get tested through a successful vision cycle. From basic motion, to light, to camera settling etc. This is also essential for testing during development. It has saved my machine from crashes many times already.
_Mark
*) Yes, I have a replacement nozzle tip holder that I should
mount, but that's beside the point.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/2842aa0d-d7e4-425e-b6a7-6f3a297cf9d2n%40googlegroups.com.
> But why block the ability to fix manually?
First of all, you can't calibrate when the machine is not at
least mechanically and/or visually homed. Otherwise, the Machine
does not yet know how far and which way to move to the bottom
camera.
So to support what you propose, we would have to store multiple states "Mechanically homed", "Visually homed", "Homed and Nozzle tip calibrated", etc. and for each homing related calibration we'd need to add one new state. And then you'd have to start assigning "minimal homing states" to the various user functions. That's becoming very complex for something that is IMHO really not a true problem.
Like I said: Set to Manual. Home. Fix the pipeline. Set to
MachineHome. Problem gone for good.
_Mark
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/0468d39f-409f-4111-9ff1-86c385355dacn%40googlegroups.com.
Hi, Mark. I would like to interject a “use case”. I never switch nozzles, I always use a 3cm diameter bellows. When I calibrate the nozzle I set the pipeline to see the brass fitting in the middle and calibrate, and all is OK. But if I want to calibrate the camera, then the nozzle is moved so far away that the camera can’t see the fitting, and the calibration fails. To calibrate the camera I have to switch to a very small nozzle so I can have a pipeline that can calibrate both the nozzle and the camera. It would be good (for me) if there were separate pipelines for calibrating the nozzle step and calibrating the camera step.
By the way your comments about the camera “learning” the shape were very interesting. I could see that would be very useful.
Best, Jim
--
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 on the web visit https://groups.google.com/d/msgid/openpnp/b0ebf4e8-7c6b-479b-b37e-4603f0b06eb0n%40googlegroups.com.
--
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 on the web visit https://groups.google.com/d/msgid/openpnp/85e3b288-0157-4209-88df-aa89b342a4a5n%40googlegroups.com.
--
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 on the web visit https://groups.google.com/d/msgid/openpnp/2842aa0d-d7e4-425e-b6a7-6f3a297cf9d2n%40googlegroups.com.
--
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 on the web visit https://groups.google.com/d/msgid/openpnp/816ccb98-07d7-9588-7aed-ed95e25d3415%40makr.zone.
Hi James
I don't understand why having a separate pipeline will help, if
the brass fitting is not seen by the camera. Can you explain?
Also, calibrating your camera should more or less be a "one-time" thing. It should not be needed routinely, so changing the nozzle tip a few times seems acceptable to me. Or is there something special about your machine that requires frequent recalibration?
@Tony, I agree that the camera calibration belongs into the
camera, but why a new pipeline? Are you using the nozzle tip to
calibrate? If yes, isn't it OK then to reuse the pipeline that this
nozzle tip defines?
_Mark
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/47fc22a3-efd0-4019-b977-1ac9e6808a0cn%40googlegroups.com.
> but I know the top camera can't use a nozzle tip for
calibration so it will need its own pipeline.
Just use a fiducial (part), like visual homing. This way we have
all the functionality in place and if that functionality evolves
(which is likely, see the discussion in the group right now), so
will the calibration. The fewer "reinvented wheel" pipelines we
have, the better. ;-)
_Mark
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/a8dccb35-a2e9-46c0-864a-d713972b9731n%40googlegroups.com.
The brass fixture is about 1.5cm deep inside the bellows. So when the nozzle is not exactly over the camera the brass fitting is obscured by the bellows rubber of the nozzle. No, there is nothing too special about my machine. I have been changing the optics (like changing the camera-part distance or focal length of the lens and when I do, I have to recalibrate. I agree, it is not a big deal. I actually threw this question out there because I thought I might be missing an already existing feature of Openpnp.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/9e7cbb8c-f790-2a76-6ded-492e60e5b3ee%40makr.zone.