Do I need to update OpenPNP?

46 views
Skip to first unread message

jbasia

unread,
Jan 14, 2026, 11:13:53 PM (2 days ago) Jan 14
to OpenPnP
Do I need to update OpenPNP?

It worked fine last time when I used it and the version is 2023-08-20_11-22-37.beb77a7

I now have the same assembly job as then, some module. I feed manually from strips. 

Issues I had before: 
Alignment of: SOT-23 and  64 pin TQFP  (not really a big issue)

Is there any point in updating? I am hesitant in fiddling with a working system.


4512e7df-e7ab-46dc-976e-4d8a6c801af0.jpg

Jan

unread,
Jan 15, 2026, 3:31:19 AM (yesterday) Jan 15
to ope...@googlegroups.com
HI Jbasia!
There is never 'need' to update. Sometimes you may feel pressured to
update, hoping that it will eliminate inconveniences. If you think, you
might be 'affected' by this, please consult the change log of the main
branch at https://github.com/openpnp/openpnp/blob/main/CHANGES.md or of
the test branch at
https://github.com/openpnp/openpnp/blob/test/CHANGES.md. (As stated many
times, the test branch is very stable and used by many for production
purposes.)
Concerning you issues:
- SOT23 is always difficult as it does not fit well in the rectangular
strategy OpenPnP defaults to. Pattern matching likely provides better
results, but OpenPnP does not yet support rotated pattern matching, just
pattern matching without rotation.
- TQFP64 is difficult to make suggests as it might be a vision hardware
issue (FoV not large enough, distorted optics, uneven lighting, bad
background) or a setup problem. It could even be a problem of a bad
nozzle tip or vacuum system that makes the part slip after vision. (You
may inspect the bottom vision debug images to find out if its a vision
or a nozzle issue.) Newer version of OpenPnP support
multi-shot-bottom-vision that capture multiple edges of larger parts.
That might help as well.

Jan

On 15.01.2026 05:13, jbasia wrote:
> Do I need to update OpenPNP?
>
> It worked fine last time when I used it and the version is
> 2023-08-20_11-22-37.beb77a7
>
> I now have the same assembly job as then, some module. I feed manually
> from strips.
>
> Issues I had before:
> Alignment of: SOT-23 and  64 pin TQFP  (not really a big issue)
>
> Is there any point in updating? I am hesitant in fiddling with a working
> system.
>
>
> 4512e7df-e7ab-46dc-976e-4d8a6c801af0.jpg
>
> --
> 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
> <mailto:openpnp+u...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/openpnp/
> f8605cfa-7b30-4742-9162-fc697737503bn%40googlegroups.com <https://
> groups.google.com/d/msgid/openpnp/f8605cfa-7b30-4742-9162-
> fc697737503bn%40googlegroups.com?utm_medium=email&utm_source=footer>.

Toby Dickenson

unread,
Jan 15, 2026, 5:26:00 AM (yesterday) Jan 15
to ope...@googlegroups.com
Hi Jbasia,

        There is never 'need' to update. Sometimes you may feel pressured to
update, hoping that it will eliminate inconveniences. If you think, you
might be 'affected' by this, please consult the change log of the main
branch at https://github.com/openpnp/openpnp/blob/main/CHANGES.md or of
the test branch at
https://github.com/openpnp/openpnp/blob/test/CHANGES.md. (As stated many
times, the test branch is very stable and used by many for production
purposes.)

2023-08 is a long time ago, and openpnp has matured significantly. My experience in that period is that openpnp has changed from a source of frustration to a reliable production tool.
 
        Concerning you issues:
- SOT23 is always difficult as it does not fit well in the rectangular
strategy OpenPnP defaults to. Pattern matching likely provides better
results, but OpenPnP does not yet support rotated pattern matching, just
pattern matching without rotation.

One of the fixes in that period is a reliable solution to sot-23 vision :-)

I have one sot-23 part where its three pads have rounded ends. Under bottom vision it looks like three round dots arranged in an equilateral triangle pattern. Openpnp is trying to fit the smallest rectangle around that shape, and sometimes it guesses the wrong orientation for the triangle.

The fix is:
1. Update your openpnp
2. Take a copy of the stock bottom vision pipeline, specialised for your sot-23 package
3. Edit the vision pipeline, and in the MinAreaRect stage change search-angle from 45° to 20°


A question for other users.... is this a common problem? would it be worth having openpn create this vision pipeline automatically?

Toby

Luke Bayes

unread,
Jan 15, 2026, 8:13:29 AM (yesterday) Jan 15
to ope...@googlegroups.com
Semi related side note:

We recently had some minor issues with an asymmetrical footprint (2-pin SMT header) that looks like a fractional SOT-23.

Created a customized vision pipeline with a reference image, limited the valid rotation angles and it worked quite well.

Also wanted to agree and express gratitude to the team that the improvements over the last year or so have really been great.

Thanks!

Luke


--
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/CAH35urdsZ-urpaQr1kX_qxqSTc%2BzmLuc342AQotF3K9oJ%3Dcb0A%40mail.gmail.com.

Jan

unread,
Jan 15, 2026, 8:22:42 AM (yesterday) Jan 15
to ope...@googlegroups.com
Hi Toby!
I too have issues with SOT23 (3pin) and I was not aware of the
limitation of allowed rotation.
Wouldn't it be a good idea to add the rotation limitation to the
default pipeline as well? I'd consider it of general use even thought
I'm not aware of other footprints that suffer from the same asymmetry issue.

Jan

On 15.01.2026 11:18, Toby Dickenson wrote:
> Hi Jbasia,
>
>         There is never 'need' to update. Sometimes you may feel
> pressured to
> update, hoping that it will eliminate inconveniences. If you think, you
> might be 'affected' by this, please consult the change log of the main
> branch at https://github.com/openpnp/openpnp/blob/main/CHANGES.md
> <https://github.com/openpnp/openpnp/blob/main/CHANGES.md> or of
> the test branch at
> https://github.com/openpnp/openpnp/blob/test/CHANGES.md <https://
> github.com/openpnp/openpnp/blob/test/CHANGES.md>. (As stated many
> --
> 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
> <mailto:openpnp+u...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/openpnp/
> CAH35urdsZ-urpaQr1kX_qxqSTc%2BzmLuc342AQotF3K9oJ%3Dcb0A%40mail.gmail.com
> <https://groups.google.com/d/msgid/openpnp/CAH35urdsZ-
> urpaQr1kX_qxqSTc%2BzmLuc342AQotF3K9oJ%3Dcb0A%40mail.gmail.com?
> utm_medium=email&utm_source=footer>.

Reply all
Reply to author
Forward
0 new messages