unexpected part rotation applied to every part

[CodeDataTech]

For some unknown reason, openpnp seems to be adding an extra rotation angle (-26.563) to every part it places. Every part on the PCB and in the job is on a multiple of 90 degrees relative to the X and Y axes. Yet, it is placing every single part on the PCB at this exact same angle offset from the correct orientation.

Where could it be getting this angle from?

I am pretty sure I have rerun every pertinent calibration procedure for anything I have readjusted to try to get it aligned as closely as possible with the X and Y axes.

I have tried different combinations of options with regard to nozzle rotation with no change in outcome.

Has anyone ever seen anything like this happen before?

What am I missing...??? Any ideas?

Please see attached picture from bottom camera looking up at part.

[Charlie Jenkins]

What is the board rotation from your fiducial check?  That would apply the same rotation to all if not correct but 26 deg is huge.

--
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/3d78d7b0-bc6d-47cd-b2f1-ae74c407a72bn%40googlegroups.com.

[CodeDataTech]

Panel fiducials rotation = 0.060; Board fiducials rotation  (4 PCBs on panel) = 0.055, 0.055, 0.053, 0.056

By the way - here is a picture of the resulting placements of one test...

[CodeDataTech]

After sending it through the oven...

Many of them were pulled into place by the reflow - which is good; however, it is not really good enough to rely on it - I need to get the correct placements consistently...

[Jonathan Oxer]

The vision pipeline isn't working properly.

You can see in your bottom camera pic that the bounding box has been drawn so that it includes an artifact that's offset from the part itself, most likely a reflection from the side of the nozzle.

Try controlling the lighting by covering the machine or reducing the room lights, load a part on the nozzle, and run the "test orientation" feature to see how it's processing the bottom vision.

Jon

To view this discussion visit https://groups.google.com/d/msgid/openpnp/2cb5f209-395e-4267-ac12-2e5cd070af14n%40googlegroups.com.

[CodeDataTech]

Thank you! I will look into that...