Board Z parallax errors

[Toby Dickenson]

Hi all,

A common recommendation on the lumenpnp discord is to set the board Z
0.3mm lower than reality. The claimed advantage is to thoroughly push
parts down into the paste.

I wonder if doing this has other disadvantages. Board Z is used for 3d
vision, so giving an incorrect board Z will cause an error on fiducial
X/Y positions in the common case where the fid is not exactly in the
center of the camera field of view. The parallax error due to 0.3mm Z
error will be small of course. But are there other errors or problems
that would be caused by having an intentionally incorrect board Z? I
wonder whether there is value in changing openpnp to have a "placement
compression" field that these users could set to 0.3mm to achieve
their paste compression, while using an accurate board Z.

Toby

[bert shivaan]

Prolly I am just being dense again, but I fail to understand the advantage. If board is set to correct height, the part is pushed down to the PCB in theory. The paste is above that level and not accounted for AFAIK? So it will always be pressed into paste because of this IMHO.

--
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/CAH35urdW%2B3MtOOq4BEemD2P2kQ65tQjyBNquytCxT0bHSp-Gdg%40mail.gmail.com.

[Toby Dickenson]

Yes, I agree. It is not something I have done, or would ever advocate,
but I was surprised that I could not see any reason why it was a dumb
idea.

Nonetheless it is a common suggestion from other users on the lumenpnp
discord. I have seen two reported advantages:
1. "When the vacuum turns on the surrounding air pressure pushes the
nozzle inward against the spring. If you set your z-locations with the
vacuum off, the nozzle tip will end up a little ways above those
locations with the vacuum on." lumenpnp uses CP40 nozzles, and I have
never seen this effect on my CP40 nozzles.
2. It is a margin of safety. If you have any errors in measuring part
height or board height, or any "bed leveling" imperfection, the
consequences for releasing the part too high are worse than for
releasing too low.

> To view this discussion visit https://groups.google.com/d/msgid/openpnp/CA%2BKNHNxUJV89NkuASiYSr1J7-tNayPujq4%2BX%2BB2%3D57zFf3qamw%40mail.gmail.com.

[Luke Bayes]

We have observed the problem being described with CP40 nozzles pulling up under vacuum.

As an openpnp user, I would likely use this setting if one existed.

Thanks!

Luke

To view this discussion visit https://groups.google.com/d/msgid/openpnp/CAH35urc5hHbxrAz5A7aPWDRbeRYMLoQNR1%2B1N3F6Y6E0JCbVkA%40mail.gmail.com.

[bert shivaan]

I like the idea of having an "extra deep setting" on a per nozzle basis maybe. Or per part.

To view this discussion visit https://groups.google.com/d/msgid/openpnp/CALEDPUvUJBC6GoQqvx4BiLgorgVPnzmoTVWKsnEPR2bcGEOcGw%40mail.gmail.com.

[Luke Bayes]

Another data point:

We solve a similar problem (and other related problems) on our SMT550 using the software "part placement offset" that can be configured for each placement (reference designator), but they make it easy to apply the offset to all (or a selected subset of) parts.

I haven't checked to see if that's already available in openpnp.

Thanks!

Luke

To view this discussion visit https://groups.google.com/d/msgid/openpnp/CA%2BKNHNzJg95e7b-BLv0xbjdVKjB1Su5B9MJy5Aw0f%2ByVsQvU9A%40mail.gmail.com.

[Toby Dickenson]

A preliminary implementation of this feature is here:

https://github.com/openpnp/openpnp/pull/1939

> We solve a similar problem (and other related problems) on our SMT550 using the software "part placement offset" that can be configured for each placement

Bizarrely I think openpnp already has that feature! In the Board tab,
the placement location only has a gui editor for X, Y, and rotation.
But you can edit the Z field for each placement in the xml and I
strongly suspect it would be effective.

Toby

[Luke Bayes]

ooooh - 

Nice work on the update and also, thanks for the tip!

lb.

--
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/CAH35urdjiri1bNiWUYk2QA9B9d-ODA3pLnL2aGx8%2B6m8QJGFQQ%40mail.gmail.com.