fiducial pipeline

geo0rpo

unread,

Feb 3, 2021, 4:14:29 PM2/3/21

to OpenPnP

Hello people!

Is there an easy way to recognize 1mm HAL fiducials?

I tried many ways but no luck, unless I scratch the surface of the pad.

Jason von Nieda

unread,

Feb 3, 2021, 4:19:19 PM2/3/21

to ope...@googlegroups.com

The best way is to use coaxial lighting so that the fids show up nice and bright. Failing that, add a threshold stage to the pipeline, turn on "invert", and then adjust the threshold until the fid looks like a white dot surrounded by black. See image below.

Screen Shot 2021-02-03 at 3.17.15 PM.png

Jason

--
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/b924c755-f285-429b-9651-988450ab6896n%40googlegroups.com.

geo0rpo

unread,

Feb 3, 2021, 5:16:53 PM2/3/21

to OpenPnP

OK, good approach but did not work for me.

I do not understand how MatchTemplate works. To my eyes it should match...but no.

johanne...@formann.de

unread,

Feb 3, 2021, 5:19:45 PM2/3/21

to OpenPnP

Three options:

- Use ENIG (my solution, since only few prototyes)

- Switch to coaxial lighting

- try to move the picture a bit "off center" and correct that with a camera offset , should help in a few cases

geo0rpo

unread,

Feb 4, 2021, 2:48:48 AM2/4/21

to OpenPnP

I am not 100% sure about the meaning of coaxial lighting. It needs a prism of some type. Right?

Mike Menci

unread,

Feb 4, 2021, 3:04:54 AM2/4/21

to OpenPnP

Ask Google - answer: https://www.google.com/search?q=coaxial+lighting&safe=active&client=firefox-b-d&sxsrf=ALeKk03pMLen2mKCZtezRt0pez_0nsEjbw:1612425108869&tbm=isch&source=iu&ictx=1&fir=UxZUu7Sjnb0vsM%252CtZfy_0LoFv8SfM%252C_&vet=1&usg=AI4_-kR54WZquE3rVOfpe8PMtis2h4EQ2w&sa=X&ved=2ahUKEwii4NzO38_uAhWNuIsKHZ1WAqsQ9QF6BAgeEAE#imgrc=oz-BNSMTiYSJnM
Mike

3d.m...@gmail.com

unread,

Feb 4, 2021, 4:29:20 AM2/4/21

to OpenPnP

My commercial machine has an option to measure the fiducial offset. I find for HASL that I usually have to use a 4mm offset, or it is spotty at best. With ENIG, on center works fine. This is with the equivalent of a ring light for lighting.

I really want to try to adapt the machine to use coaxial lighting for just this reason.

geo0rpo

unread,

Feb 4, 2021, 4:35:49 AM2/4/21

to OpenPnP

Yes I know how to search google :))) I was asking if there is a specific part already used in openpnp.

geo0rpo

unread,

Feb 4, 2021, 4:04:17 PM2/4/21

to OpenPnP

Lesson learned. I think coaxial lighting is the only way to go.

Thanks for the enlightening tip.

Jason von Nieda

unread,

Feb 4, 2021, 4:11:59 PM2/4/21

to ope...@googlegroups.com

Wow, great demonstration! Nice work!

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/708562cd-4b49-4711-893f-f205b6e777e2n%40googlegroups.com.

geo0rpo

unread,

Feb 4, 2021, 4:17:16 PM2/4/21

to OpenPnP

I am impressed. So simple yet so perfect.

Now I have to design and print a coaxial lighting box :)

Thank you guys for all the support and knowledge.

Anatoly

unread,

Feb 4, 2021, 8:27:56 PM2/4/21

to OpenPnP

Coaxial Lighting

https://groups.google.com/g/openpnp/c/OtT6iLNAufU/m/vHWb9nGjBwAJ

пятница, 5 февраля 2021 г. в 04:17:16 UTC+7, geo...@gmail.com:

geo0rpo

unread,

Feb 5, 2021, 11:31:03 AM2/5/21

to OpenPnP

First test with success.

geo0rpo

unread,

Feb 5, 2021, 11:32:06 AM2/5/21

to OpenPnP

ma...@makr.zone

unread,

Feb 5, 2021, 11:58:19 AM2/5/21

to ope...@googlegroups.com

Great!

Just plain glass?
And no thermal problem?

_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/7e96f985-54a7-49ee-8f4e-aff875949557n%40googlegroups.com.

geo0rpo

unread,

Feb 5, 2021, 11:59:34 AM2/5/21

to OpenPnP

Just plain 3mm glass.

I do now know what you mean by thermal problem... Should I get scared?

ma...@makr.zone

unread,

Feb 5, 2021, 12:11:41 PM2/5/21

to ope...@googlegroups.com

Just the LEDs producing warmth, captured in that box. Probably a question of the combined LED power and how well the glass reflects.

I was just asking, because one needs good strong light, so you can dial down the exposure of the camera, because of this effect:

https://makr.zone/camera-fps-cpu-load-and-lighting-exposure/519/

_Mark

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/063db3a2-856a-40a3-b151-abdd1e3072e4n%40googlegroups.com.

Harjit Singh

unread,

Feb 5, 2021, 10:57:56 PM2/5/21

to ope...@googlegroups.com

What are you doing for the LED diffuser?

--
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/d4f269ae-e3d5-4c3d-80ea-60f70d63e1c2n%40googlegroups.com.

Clemens Koller

unread,

Feb 6, 2021, 6:35:11 AM2/6/21

to ope...@googlegroups.com

On 04/02/2021 22.04, geo0rpo wrote:
> Lesson learned. I think coaxial lighting is the only way to go.
>
> Thanks for the enlightening tip.

Yes. You could additionally work with polarizers in case you have to deal with reflections - which is the case of HAL oder ENIG surfaces. We used this stuff in the past:
https://www.ccs-grp.com/products/series/317
https://www.ccs-grp.com/products/series/205

Ideally, you can vary with a ring light + coaxial light + diffuser / polarizers (and wavelength).
But be careful, some of these products ridiculously overpriced.

Clemens

On 04/02/2021 22.04, geo0rpo wrote:
> Lesson learned. I think coaxial lighting is the only way to go.
>
> Thanks for the enlightening tip.
>
>

> Graphic1.png
>
> On Thursday, February 4, 2021 at 11:35:49 AM UTC+2 geo0rpo wrote:
>
>
>
> Yes I know how to search google :))) I was asking if there is a specific part already used in openpnp.
>
> On Thursday, February 4, 2021 at 11:29:20 AM UTC+2 3d.m...@gmail.com wrote:
>
> My commercial machine has an option to measure the fiducial offset. I find for HASL that I usually have to use a 4mm offset, or it is spotty at best. With ENIG, on center works fine. This is with the equivalent of a ring light for lighting.
>
> I really want to try to adapt the machine to use coaxial lighting for just this reason.
>
> On Wednesday, February 3, 2021 at 3:19:45 p.m. UTC-7 johanne...@formann.de wrote:
>
> Three options:
> - Use ENIG (my solution, since only few prototyes)
> - Switch to coaxial lighting
> - try to move the picture a bit "off center" and correct that with a camera offset , should help in a few cases
>
> geo...@gmail.com schrieb am Mittwoch, 3. Februar 2021 um 23:16:53 UTC+1:
>
> OK, good approach but did not work for me.
>

> 1.jpg

>
>
> 2.jpg
>
> I do not understand how MatchTemplate works. To my eyes it should match...but no.
>
>
> On Wednesday, February 3, 2021 at 11:19:19 PM UTC+2 Jason von Nieda wrote:
>
> The best way is to use coaxial lighting so that the fids show up nice and bright. Failing that, add a threshold stage to the pipeline, turn on "invert", and then adjust the threshold until the fid looks like a white dot surrounded by black. See image below.
>

> Screen Shot 2021-02-03 at 3.17.15 PM.png
>
>
> Jason
>
>
>
> On Wed, Feb 3, 2021 at 3:14 PM geo0rpo <geo...@gmail.com> wrote:
>
> Hello people!
>
> Is there an easy way to recognize 1mm HAL fiducials?
> I tried many ways but no luck, unless I scratch the surface of the pad.
>
>
>

> fid.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.

> To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/b924c755-f285-429b-9651-988450ab6896n%40googlegroups.com <https://groups.google.com/d/msgid/openpnp/b924c755-f285-429b-9651-988450ab6896n%40googlegroups.com?utm_medium=email&utm_source=footer>.

>
> --
> 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 on the web visit https://groups.google.com/d/msgid/openpnp/708562cd-4b49-4711-893f-f205b6e777e2n%40googlegroups.com <https://groups.google.com/d/msgid/openpnp/708562cd-4b49-4711-893f-f205b6e777e2n%40googlegroups.com?utm_medium=email&utm_source=footer>.

geo0rpo

unread,

Feb 8, 2021, 5:41:10 AM2/8/21

to OpenPnP

For the diffuser I print with "natural color " PLA at 100% infill.

I also designed coaxial light for the bottom camera. Working perfect.

I could share the STL if anyone is interested.

Thorsten Wilmer

unread,

Feb 8, 2021, 6:27:51 AM2/8/21

to ope...@googlegroups.com

Hi

Nice build 👍. Do you have also pictures taken with the camera?

Has someone else or you tried a coaxial light without a diffuser?

There are some 10W led cob modules available and the surface area is similar to what is available when a 30*30mm glass is at 45°.

I'm sure one would have to dim such a 10W led considerably to avoid over exposure and avoid melting the plastic.

Kind Regards

Thorsten

--
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.