Coaxial Lighting

[Jason von Nieda]

Has anyone tried something like https://www.aliexpress.com/item/COL-40-Machine-Vision-Lamp-Coaxial-Light-Source-38-42mm-Industrial-LED-Lighting-Detection-Source-White/32919890175.html?

I've been looking at a lot of examples and it seems like it might be pretty great for top vision, especially, but also bottom vision.

I might pick one up and try it out - but I'm also interested in, if it works well, replicating it much cheaper. It's just a black box, a two way mirror and an array of red LEDs.

Jason

[Michael Anton]

I can't figure out where they intend to mount the camera with that unit.

[Jason von Nieda]

The camera goes beneath (or above). Like this: https://www.keyence.com/Images/ca-d_ws_sr_ca_d_lineup08_ca-ds_1_1341117.jpg

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 post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/0f461775-97c4-4c5a-989a-5d7b3e010cf4%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Michael Anton]

Here is a smaller one that is much cheaper: https://www.aliexpress.com/item/Machine-Vision-LED-Lighting-Coaxial-Light-Source-25-25mm-White-Light-Adjustable-Brightness-Identification-Test-Industrial/32915436395.html.  I understand now how these work, but the pictures on the first link didn't show the back side of the device, so I was confused.

On Friday, February 15, 2019 at 11:07:59 AM UTC-7, Jason von Nieda wrote:

[Marek T.]

Hi Jason,

So have you (or anybody else) tried it already?

I made small "investigation" to buy this or build. I know now the the goal is this 45deg angled plate which is so-called two-way mirror (venezian mirror, fenice mirror).

It is reflecting some 80% of light at one side and passing some 80% of light coming from second side.

There are two options I have found and I'm not sure is it better if the LED light is reflected and camera looking through the plate (this two-way mirror):
https://www.wodop.com/wp-content/uploads/2016/03/26-COP.pdf

or camera reflected but LED light passing through the plate:
https://www.wodop.com/wp-content/uploads/2016/03/25-COA.pdf.

I'm scared with the picture distortions anyway. No matter if camera vie is reflected or coming through - the plate is a point of distrortions introducing.

Some thoughts?

[Mike Menci]

Marek 

Good example is here ; https://www.diyphotography.net/use-mirrors-get-perfect-axial-lighting-macro-subjects/

Mike

[Mike Menci]

and here http://www.visionlux.net/latest-news/ 

details : https://www.dropbox.com/s/f7h0ug8u78x2tc6/VisionLux%20Catalogue%202019.pdf?d

[Marek T.]

Thx Mike but it doesn't answer for distortion question... However the lens correction can be still applied so maybe no reason to worry about it too much.

[bert shivaan]

hi Marek, I received the acrylic 2 way mirrors. here is a picture showing the shift from passing though it. To get this pic I held the mirror so the edge lined up with the pins. A true glass mirror will likely work better, but so far i am not impressed. I get better results by simply adjusting the light to shine on the part it seems like.

[Mike Menci]

Have a look at this old post ; https://groups.google.com/forum/m/?utm_medium=email&utm_source=footer#!search/Up$20camera$20light$20mike$20menci/openpnp/BrOtYDYFFnI
You might find a simple workin solution which I use there - any question don’t hesitate to ask.
Mike

[bert shivaan]

hi mike, that link does not seem to work for me, just takes me to google gropups :(

anybody else have this issue?

-Bert

[Marek T.]

Hi Bert,

Link doesn't work for me too.
Your picture looks really shi** but I believe it must be better. Maybe other mirror material is really needed. I have ordered some from China and will get it within one week or so so, at some bigger shipment occasion. I'll come back when after some tests.

PS. Mike's link doesn't work for me too.

[Trampas Stern]

The Manncorp uses a prism for up looking camera on head. This allows them to move prism under part and do the part alignment while head is moving (aka on-the-fly alignment).  They have two different light sources one looking up and one looking down. If the turn on the down looking light the camera now sees the PCB, when the up looking light is on the camera sees the part/nozzle.  This allows some neat tricks, for example it is easy to do the nozzle offset from camera.  

Note the Manncorp has two cameras for the part rotation (one per nozzle), and one down looking main camera.  Then they still have an up looking camera on base for doing alignment of large parts (BGAs, big TQFP, etc).   However to align main down looking camera they turn on the head cameras looking down, and you pick a point on a PCB. Then they ask you to find the same point with second nozzle camera and the main down looking camera.  

I have found that having light from the side creates some shadowing effects that can help vision algorithm performance.   The Manncorp allows you to set the brightness of the light based on the part.  I had thought about doing this with RGBW light sources. That is per part determine which color light works best, then store the light setting with each part such that when I place the part the PnP machine will use the correct lighting for the part.   Along the same lines is to store the camera settings, brightness, contrast, etc. per part. 

[Mike Menci]

Here the link - https://groups.google.com/forum/m/?utm_medium=email&utm_source=footer#!search/Up$20camera$20light$20mike$20menci/openpnp/BrOtYDYFFnI
Open the zip file for images

[Mike Menci]

New link
https://groups.google.com/forum/m/?utm_medium=email&utm_source=footer#!search/Up$20camera$20light$20mike$20menci/openpnp/BrOtYDYFFnI
See the zip enclosure and this is as well indirect light to nozzle tip /mirror up

[Mike Menci]

Finally I think I have the right link;

Here the link : https://groups.google.com/d/topic/openpnp/BrOtYDYFFnI/discussion

and the enclosure files are here now as well; 
Acrylic Ring as well as Dwg enclosed.

Mike

[Mike Menci]

To stay close to the subject - hard drive discs (2pcs = up and down (1)) are used to reflect light - illumination to centre and work as two mirrors.. 

 Mike

[James Carter]

In building a coaxial lighting rig with a PtGrey Flea 3 camera.

For this I have a 50/50 transmission mirror (glass) from a optics company.

Only 25% of light from the source hits the camera, so I am planning on overdriving the LEDS for the frame time of the camera.

I have the assembly machined already but am yet to design the LED board for this rig. Happy to report back this results.

For your reference I think the mirror (50x70mm approx) was around £60.

From what I've read coaxial lighting is the best for looking at subjects such as ours. Items with a high reflection index.

[bert shivaan]

Mike do you have any build details for your HDD mirror camera? I have WAY WAY WAY too many HDD's not to try this out.

[Mike Menci]

If you look at the picture posted - on the left you will find parts- listed used items. I use Led strip 12V DC which can be cut to length needed - I post the link Here
https://www.google.com/search?q=ebay+12VDC+led+tape&safe=active&rlz=1CDGOYI_enID614ID615&hl=en-US&prmd=isvn&source=lnms&tbm=isch&sa=X&ved=2ahUKEwju14igkf_iAhVO3qQKHV5BB6wQ_AUoAXoECA0QAQ&biw=375&bih=539#imgrc=RyzS8kHKlElBXM
Let me know how it works for you!
You can buy cheap Led dimmer from eBay - I needed to decrease a bit the illumination !
Mike

[Florian Chende]

Does this looks ok to build a coaxial illuminator?

https://www.aliexpress.com/item/32964314615.html?storeId=4046006&spm=2114.12010612.8148356.2.2c5eee2bMNE9Bc

I'm not very good at optics, it is better to ask.

I was unable to find bigger than 30x30mm. Does this mean that I will not be able to illuminate components bigger than 30x30mm?

Thank you.

[Florian Chende]

Or this one: 

https://www.aliexpress.com/item/33013750958.html?spm=a2g0o.cart.0.0.44bc3c00KkDsBd

[Marek T.]

For coaxial lightning you need two-way mirror (venezian mirror) and difusing plate.

[Florian Chende]

That looks like a two way mirror, in the description is "50R/50T", I assume that is 50% reflection 50% transmission. But I am not sure.

[Scott Wilson]

Yeah- thats definitely 50/50, and worth trying. I wonder if 80/20 would work better? I had been thinking of using acrylic from ebay, which is closer to 80/20.

This stuff is cheap enough I'm going to order both and see how it works, but it'll be quite a while before I have a chance to try.

[Marek T.]

I don't know too much about parameters of these two plates yet, so not sure whwther 50/50 is ok, but I think (guess) it should be rather non-symetrical.

My mirror and diffusor lays in DHL store waiting for clearance with some other goods.

Here you have few links. In some of them you can find the exact sizes and drawings, maybe it can be useful for you.

https://www.keyence.com/products/vision/vision-sys/ca-d/lineups/lineup-08.jsp
https://www.stemmer-imaging.com/en-gb/knowledge-base/standard-coaxial-illumination/
https://www.advancedillumination.com/wp-content/uploads/2018/10/DL225-Series.pdf

http://www.chinaacrylicsheet.com/two-way-mirror-sheet/two-way-mirror-sheet.html
http://www.chinaacrylicsheet.com/acrylic_diffuser/

[Marek T.]

seems to me like Scott said...

[Florian Chende]

Ok, let's try it. I ordered also some pieces for testing.

For illumination, I am thinking at an array of white leds in 0402, 30x30mm pack, as tight as possible, and a diffuser above.

[Mike Menci]

I have this -http://moritex.com/model/1-2-1-7-2-1-08.html
See the pdf and dwg. I Use dimmer with it

[John Plocher]

> For illumination, I am thinking at an array of white leds

Some articles I’ve seen suggest red monochromatic is better than white for both contrasts and reflections.  You may want to play with both mirrors and colors...

   John 

[Brynn Rogers]

I would use RGB LEDs, then you can get whatever color you want, and also 'white'

Brynn 

[Mike Menci]

Here photos 

Light is in 3 stages 12V DC

Mike 

[Florian Chende]

Ok, ws2812 it is. I just need one month of patience for the mirror to arrive.

[Sandra Carroll]

my main hobby is astronomy and right now I’m playing with spectrography.   So thought I would share  what I’ve been learning.   Take it FWIW

 

So just an FYI,  RGB LED’s don’t’ give a real white, actually no LED gives a real white,  I’ve been playing with spectrum of LED compared to true white which the reference is our sun so daylight.

LED’s tend to be very cool or very warm,  ie towards the blue or red with dips between the R,G and B colors especially B to G depending on the LED.

 

The attached is of a typical Cool White LED,   you can see its strong in Blue, dips before going Green then Red rolls off

 

This is not to say to you to not to do this but to be aware unless you can measure it, don’t’ be sure your white is really white.

camera sensors also tend to be more sensitive to some colors that others.   Both of these can have effects that you may not get results you’re thinking you’ll get

--
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 post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/916c33ae-547d-462f-a4cb-9e1fffdbebe7%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Mike Menci]

I think Colour Red, Blue .... is the history - White is now more and more used ... 

there are plenty used working machine vision units out there & good for Open PnP
https://www.ebay.com/sch/i.html?_from=R40&_nkw=Machine+Vision+Coaxial+Light&_sacat=0&_sop=15

Mike

On Friday, 5 July 2019 19:32:01 UTC+2, Sandra Carroll wrote:

my main hobby is astronomy and right now I’m playing with spectrography.   So thought I would share  what I’ve been learning.   Take it FWIW

 

So just an FYI,  RGB LED’s don’t’ give a real white, actually no LED gives a real white,  I’ve been playing with spectrum of LED compared to true white which the reference is our sun so daylight.

LED’s tend to be very cool or very warm,  ie towards the blue or red with dips between the R,G and B colors especially B to G depending on the LED.

 

The attached is of a typical Cool White LED,   you can see its strong in Blue, dips before going Green then Red rolls off

 

This is not to say to you to not to do this but to be aware unless you can measure it, don’t’ be sure your white is really white.

camera sensors also tend to be more sensitive to some colors that others.   Both of these can have effects that you may not get results you’re thinking you’ll get

 

 

From: ope...@googlegroups.com <ope...@googlegroups.com> On Behalf Of Brynn Rogers
Sent: Friday, July 5, 2019 12:47 PM
To: OpenPnP <ope...@googlegroups.com>
Subject: Re: [OpenPnP] Re: Coaxial Lighting

 

On Friday, July 5, 2019 at 11:19:37 AM UTC-5, John Plocher wrote:

> For illumination, I am thinking at an array of white leds

 

Some articles I’ve seen suggest red monochromatic is better than white for both contrasts and reflections.  You may want to play with both mirrors and colors...

 

   John 

 

 

I would use RGB LEDs, then you can get whatever color you want, and also 'white'

 

Brynn 

--
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 ope...@googlegroups.com.

[Jason von Nieda]

Don’t use WS2812. PWM frequency is too low and causes strobing in the camera image. Use constant current or just solid color.

--

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 post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/3de520da-533a-4f58-92d0-c624dcb8b6d3%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Sent from my BeOS enabled toaster

[Florian Chende]

Good point, did't think at that. Thank you.

[Sandra Carroll]

I’m nor sure what you are trying to say.

My reply was about white LEDs are not white (Fact),  they either Warm or Cool (towards Red(gives the yellowish look) or Blue) 

 

Sandra

To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/558a7e11-bd19-4791-b107-a252cc863c73%40googlegroups.com.

[Brynn Rogers]

I think he was responding to Jason's post - that the PWM freq of the WS2812 is too low and causes flickering.

The WS2812 is a serial controlled driver chip integrated into an RGB LED.    You can dial in the RGB with 8 bits worth of PWM (on each color)  from one serial pin- maybe if you choose 255 it will not flicker, but I think your better off with a constant current driver than such a part.

[Florian Chende]

Sorry if I created confusion. I was thanking Jason about pointing out WS2812 problem, that led chip is creating various luminosity levels by pwm modulation and I forgot that this will interfere with the electronic shutter of the camera.

[John Plocher]

My point on monochrome leds was partly the pen issue and the fact that all rugby leds have the 3x components are in different orientations, and can cause slight orientation or focus issues (I.e., the red Light reflection and the blue reflection may be in different places/planes because the emitters are fractions of a mm off center...)

Maybe I’m over thinking this...

  John 

On Fri, Jul 5, 2019 at 12:09 PM Florian Chende <chende...@gmail.com> wrote:

Sorry if I created confusion. I was thanking Jason about pointing out WS2812 problem, that led chip is creating various luminosity levels by pwm modulation and I forgot that this will interfere with the electronic shutter of the camera.

--
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 post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/ed3acf93-7cf3-4b53-85c6-2f416cc6922f%40googlegroups.com.

[Mike Menci]

Have a look and explanations here - https://www.opto-e.com/basics/light-in-machine-vision

On Friday, 5 July 2019 21:30:50 UTC+2, John Plocher wrote:

My point on monochrome leds was partly the pen issue and the fact that all rugby leds have the 3x components are in different orientations, and can cause slight orientation or focus issues (I.e., the red Light reflection and the blue reflection may be in different places/planes because the emitters are fractions of a mm off center...)

Maybe I’m over thinking this...

  John 

On Fri, Jul 5, 2019 at 12:09 PM Florian Chende <chende...@gmail.com> wrote:

Sorry if I created confusion. I was thanking Jason about pointing out WS2812 problem, that led chip is creating various luminosity levels by pwm modulation and I forgot that this will interfere with the electronic shutter of the camera.

--
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 ope...@googlegroups.com.

[Jason von Nieda]

That's a really nice reference Mike, thanks for posting it.

Jason

To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/b81a4e18-f5fd-4f7d-8aca-63bffc76b90b%40googlegroups.com.

[Florian Chende]

Ok, let's see if I got this right. I attached a rough sketch.

I reversed the led and the camera, it is easier for me to include it into pnp machine this way.

-led array, from the direction B, through diffuser (for light uniformity), will transmit 50% to direction D (useful) and 50% to direction C (wasted)

-reflected PCB light coming from direction D, goes 50% to direction A (camera) and 50% to direction B (wasted)

-camera sees 50% from direction D and 50% from direction C. In that case, direction C should have a really black object, ideally nothing should be reflected from there, 50% of light from array hits it.

Am I right?

[Florian Chende]

Hi,

For the above setup I purchased some optical components and I made a rough test fixture. It works very well, illumination is uniform, as expected. It is a bit difficult to adjust all the angles without a special designed fixture with adjustments (which will come in the future), but now was just for proof of concept.

However, I have a small problem and I need an advice. On the schematic I posted, direction C is 50% visible to the camera and it receives 50% light from leds, as initially suspected. I was not able to find a material black enough to absorb all (or most) of the light received. Reflected light worsen the contrast of the image. I tried with black plastic cases, they reflect a lot of light, I also tried with photographic film listed black. It is opaque, it blocks light, but it has a shiny surface and reflects what is coming.

Can anyone recommend me something affordable (paint, material, whatever) which can absorb light?

Thanks.

On Friday, February 15, 2019 at 8:07:59 PM UTC+2, Jason von Nieda wrote:

Has anyone tried something like https://www.aliexpress.com/item/COL-40-Machine-Vision-Lamp-Coaxial-Light-Source-38-42mm-Industrial-LED-Lighting-Detection-Source-White/32919890175.html?

I've been looking at a lot of examples and it seems like it might be pretty great for top vision, especially, but also bottom vision.

I might pick one up and try it out - but I'm also interested in, if it works well, replicating it much cheaper. It's just a black box, a two way mirror and an array of red LEDs.

Jason

[Mike Menci]

The arrangement shall be as per sketch ; Coaxial illumination geometry (standard and collimated). Follow above link

[Andreas Bergmann]

On Friday, 2 August 2019 10:11:36 UTC+2, Florian Chende wrote:

Hi,

Can anyone recommend me something affordable (paint, material, whatever) which can absorb light?

Thanks.

Xou might try Ventablack https://en.wikipedia.org/wiki/Vantablack

If this doesnt help .... ;)

[Mike Menci]

https://www.opto-e.com/basics/light-in-machine-vision

[Florian Chende]

Looks good but vantablack it's not available to average joe.

Instead I found some paint called 'black 2.0' which looks good for my purpose. I ordered one and now I'm waiting to arrive.

[Jason von Nieda]

Black 2.0 is the way to go. Vantablack is all wrapped up in a bunch of licensing BS. Black 2.0 is basically an "open source" protest against Vantablack :)

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/f3999933-861f-4f50-85b3-89d0720c4864%40googlegroups.com.

[Scott Wilson]

I think an 80/20 mirror would be better- you want most of the LED lighting reflected or you're going to light up your subject and the 'backstop' equally.

-Scott

[Florian Chende]

Unfortunately most mirrors (or beam splitters) easily available on aliexpress are 50/50, so I used those.

Anyway, except the problem with that surface which must be black, I'm very happy with the result. The illuminator I used is an array of 0402 leds, around 200 pcs crammed in a 25x25mm space, built by me.

Among other things observed during the tests are that I may not really need the diffuser above leds and the protective window (at least the one I use) is very reflective, so it must be mounted at an angle, not perpendicular to viewing direction.

I will post some pictures after I will build a proper casing, but that may take some time.

[Brynn Rogers]

Isn't Black Felt the poor man's VANTA?      isn't something with texture (like felt,black fur, or a forest of nanotubes[=vanta]) better than a flat surface painted black?

Brynn

On Friday, August 2, 2019 at 11:14:35 AM UTC-5, Jason von Nieda wrote:

Black 2.0 is the way to go. Vantablack is all wrapped up in a bunch of licensing BS. Black 2.0 is basically an "open source" protest against Vantablack :)

Jason

On Fri, Aug 2, 2019 at 10:56 AM Florian Chende <chende...@gmail.com> wrote:

Looks good but vantablack it's not available to average joe.

Instead I found some paint called 'black 2.0' which looks good for my purpose. I ordered one and now I'm waiting to arrive.

On Friday, August 2, 2019 at 11:25:07 AM UTC+3, Andreas Bergmann wrote:

Xou might try Ventablack https://en.wikipedia.org/wiki/Vantablack

If this doesnt help .... ;)

--
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 ope...@googlegroups.com.

[Florian Chende]

I prefer paint. It's not just that surface which must be black, I observed other parts (like mirror edge, supports, etc) inside the assembly which shines and which may cast unwanted reflections into the lens. Having paint which is easily applied anywhere looks better for me.

[Brynn Rogers]

In my infinite distractions today, I went and looked up this:   https://arxiv.org/ftp/arxiv/papers/1407/1407.8265.pdf

   compares some of the different blacks commonly used.

[Scott Wilson]

I ordered this: https://www.aliexpress.com/item/32967047805.html?spm=a2g0s.9042311.0.0.67f64c4dwmv4Ug

I have it on hand, but I don't have my 3d printer up and running yet, so I haven't been able to make a jig to test.

-Scott

[Mike Menci]

Is it a square or round ? I see comment that the seller did not indicate that they are selling this as round glass - not as square on the picture?
Mike

[Scott Wilson]

I ordered the 25.4mm version and its square. They were pretty fast (for aliexpress) too- standard shipping took 20 days to be delivered to the US.

-Scott

[Peter Chaisty]

Hi Scott

Thanks for sharing.

Can't wait to see how you get on

Peter

[Scott Wilson]

Finally was able to try some things- and I'm not hopeful. First off that mirror I linked from Aliexpress is definitely the wrong thing. It transmits 80% and reflects 20%, so entirely backwards. To keep things moving I got a piece of acrylic that is closer to 80R/20T and tried that. The PCB is illuminated, and after painting the 'far side' of the mirror I don't get any unwanted reflections.

The problem I have is that the reflected glare from the PCB is still a major problem- its certainly not any better than just using a ring light. I'm guessing that I don't have the alignment dialed in _perfectly_, but I spent quite a bit of time making fine adjustments and it never got much better.

My final thought is: if this was easy then those coax lighting modules probably wouldn't be $150.

[Florian Chende]

Hi,

Sorry to hear bad news. My setup worked fine, but it is with 50/50 mirror. I still have to build a proper housing for it.

Can you post some pictures of your setup?

Flo.

[Scott Wilson]

I have a 50/50 mirror on order, but it'll be a month or two before it gets here. I don't feel like that'll make a huge difference, but since yours is working its definitely worth a try.

I'm not where I can take pictures, but I attached the STLs. The mirror is 25mm, so the whole assembly is quite small. I wonder if scale matters?

-Scott

[Scott Wilson]

I uploaded to thingiverse so theres an online viewer:

https://www.thingiverse.com/thing:3841789

[Florian Chende]

Looks fine to me...

My test setup was much more rudimentary than yours. I used some copper thick wires to support various components.

What was the light source you used? Afaik, light source has to be uniform over the entire area of interest, if it is spot type, it's not good.

Mine was an array of 12x24 leds, 0402. Initially I thought I need a diffuser above them but it worked just fine without one.

[Scott Wilson]

My lighting source isn't too different, its an array of 1206 LEDs. I suppose it may be too bright. I don't have an easy way of dropping the brightness, but I may experiment with that in the coming weeks. Probably made worse by the 80% reflective mirror.

[Florian Chende]

Yes, that 80/20 mirror may be the primary problem. I am curious to see your results with 50/50.

I don't think the absolute brightness matters as long as you have full control over your camera parameters.

Mine are also very bright, they pierce through retina at full brightness and also heats everything too much for continuous use, but with careful adjustment of camera parameters, the results are good.

[Scott Wilson]

You inspired me to try again and the results were much better- with the camera adjusted the 'glare' was dealt with and the result is pretty good. I'll be tweaking a bit more but I think this a good way to go.

[Jason von Nieda]

That looks quite good Scott! Very nice contrast on the fiducial, and that is hard to get with a ring light.

Jason

On Mon, Sep 2, 2019 at 5:31 PM Scott Wilson <scott.t...@gmail.com> wrote:

You inspired me to try again and the results were much better- with the camera adjusted the 'glare' was dealt with and the result is pretty good. I'll be tweaking a bit more but I think this a good way to go.

--

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/7a4e665b-4ba6-4aaf-91a3-2addae2528d9%40googlegroups.com.

[Juan-Antonio Søren E. Pedersen]

https://www.dino-lite.com/products_detail.php?index_m1_id=4&index_m2_id=24&index_id=128

Just found this little coaxial fella. Don’t know the price though..

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2BQw0jwRq6OCUrk5mkOiN_-j%2B7%3DX3hyCRfF6s7chJ3MAumHoAw%40mail.gmail.com.

[bert shivaan]

looks like $109

https://digital-microscopes.com/product/n3c-a-edge-series-coaxial-illuminator-cap/

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2B7Q0cP6b%2BNdxC5Ya9LQ8ZmW88okzSs--eGto7Bed9heX-BLLg%40mail.gmail.com.

[Juan-Antonio Søren E. Pedersen]

Nice design though ...

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2BKNHNypwf7P_zCBJnnn0s4%3DML-abX7gu5ib7FhzjhX2BsvSRQ%40mail.gmail.com.

[Florian Chende]

Hi Scott,

Looks good, but, from what I can see in the picture, your illuminator center is not in the camera center and the size of the illuminated area is smaller than camera viewing area. If you can do something about it, to match the viewing area with illuminated area, I think it would look better. In the above situation, autoexposure will try to correctly adjust the entire image, therefore border area are underexposed and inner is overexposed. The xtal, solder joints and fiducial are too shiny.

[ma...@makr.zone]

Not just the camera viewing area but the viewing area at double the viewing distance (the cone being reflected back).

_Mark

[Morpheus]

Hi 

I have sketched a, what could be a solution for coaxial lighting for the mgrl ligt/cam mount. 

https://a360.co/2PMYlkF

The mirror/glass is a standard microscope sample plate 26mm wide, cut to length. I have experimented with the aperture of the 8mm lens for the ELP 1MP cam. Its about 5.4mm leaving the lens. 10mm from the lens i guess 8mm aperture will do. 

Do any of you know the exact FOV for those lenses. 

I like the idea of the black felt backing. Will try this out when i get around to it.

The mgrl design have led´s pointing up into the dome, to defuse the light down. I will mount some led´s on the back side of the mount defusing the light through the mount, using transparent PET. 

 

[Maple_Dude]

looks neat

[Sandra Carroll]

Instead of black felting check out Black 2.0/3.0 available here

https://www.culturehustleusa.com/products/black-v1-0-beta-the-world-s-mattest-flattest-blackest-art-material?_pos=1&_sid=c9045e1d9&_ss=r

 

This is a extremely flat black,  99+% non-reflective.   Cost for the 2.0 is 15.99

This will kill reflections better than felt.

 

In astronomy we used to use the black felt but it does still cause some reflection,   we’ve been moving to this

The other thing we found was the felt wore out over time and began flaking off, 

Professionals use vantablack which is only commercially available with license .  the Black 2.0/3.0 are equivelent

 

If you really want to kill reflection, give it a try

 

Sandra

--

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/9859d258-639f-4969-aa50-0cd97b8b6646%40googlegroups.com.

[Scott Wilson]

I can second that- I used black 2.0 on the inside of my assembly and it completely killed the reflected image.

-Scott

[Florian Chende]

Same here with black2.0. However, I had to put an angled surface behind the mirror, a perpendicular one still gave some reflection. 

In the mean time I managed to build a 3d printed housing.

[Juan-Antonio Søren E. Pedersen]

Good to know.

I will order the black 2.0 and do a test with this design: There is just about room for two 45° angles in the back. The box needs to be printed in to or three parts and then glued together.  

Den fre. 6. sep. 2019 kl. 15.33 skrev Florian Chende <chende...@gmail.com>:

Same here with black2.0. However, I had to put an angled surface behind the mirror, a perpendicular one still gave some reflection. 

In the mean time I managed to build a 3d printed housing.

--

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/58bfda74-db6a-4da0-8548-c220daf316d8%40googlegroups.com.

[Scott Wilson]

I'll be posting my cad files, but they're more or less the same as before. I just made some changes to accommodate a larger mirror, and some mounting tabs so that I can actually use it. I'm still using an 80/20 acrylic mirror. I need to do some tweaking still, and probably install a diffuser. Attached are screen shots of clear tape, black tape, and a fiducial. The clear tape recognition is amazing imo. The fiducial is better than I originally had, but needs some work.

-Scott

[Jason von Nieda]

Looks very good! Adding a diffuser will help with the fiducial a bit, I bet, but I think it might also reduce the contrast on the clear tape, since the specular reflections seem to be what is providing the contrast. Definitely interested to see how it goes.

For what it's worth, one of the commercial machines I work with has multiple lighting options on each camera, and when you set up vision for an object you can specify how it should be lit. I wonder if this is a situation where coax + ring might work better for the clear tape, while coax alone works better for the fid.

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/b10f7b49-5c68-4855-a4c0-e014eae38996%40googlegroups.com.

[Shai]

What about using something like this? It's super cheap and just about the right size. Or am I missing something here? https://www.aliexpress.com/item/32954439897.html?spm=a2g0o.productlist.0.0.23195103yvCbua&algo_pvid=d9c17215-6337-464a-8545-ff312db559e2&algo_expid=d9c17215-6337-464a-8545-ff312db559e2-1&btsid=19119a2e-c48f-4a3e-97b2-0d3c613a0cb1&ws_ab_test=searchweb0_0,searchweb201602_6,searchweb201603_52

[Florian Chende]

I used something like that for tests. It was from other vendor from aliexpress but basically same parameters. It is not ok for final design, as 30x30mm is too tiny and won't cover the entire field of view of the camera (at least my camera, it has a cctv lens on it). But it is ok for proof of concept. In a post above there is a picture with a test fixture of mine, that one has a 30x35mm 50r/50t mirror. That one is also not enough to cover the entire fov. For the final design I will use an 80x80mm.

[Александр Зендриков]

Hi. Did you manage to complete the project?

вторник, 10 сентября 2019 г., 12:04:52 UTC+3 пользователь Florian Chende написал:

[Florian Chende]

Hi,

It is still work in progress, this covid situation severely delayed my project. Currently I am working at x-y mechanical parts, but I am stalled for 1 week, the mill I use is broken, waiting for spare parts to arrive. Only after those are finished, I will work at lightning.

[Александр Зендриков]

I understand you. I think what sizes to order a glass 80*80 or 50*50.

If you buy 80*80, a very large case will turn out.

I found this glass:

https://aliexpress.ru/item/4000336706378.html?spm=a2g0o.productlist.0.0.190737c4y8tdcn&algo_pvid=ca22daa2-85a9-49e6-a51d-aa9a187d9232&algo_expid=ca22daa2-85a9-49e6-a51d-aa9a187d9232-1&btsid=0b8b034c15895207736314386e3e7a&ws_ab_test=searchweb0_0,searchweb201602_,searchweb201603_

пятница, 15 мая 2020 г., 10:20:41 UTC+3 пользователь Florian Chende написал:

[Marek T.]

Area of enlighted field is twice smaller than the size of the mirror. I've got mirror 30x30 and I see clear bright square 15x15 on the board, out of it there begins the shadow as deeper as further from the center. This is enough for the fiducials and placed parts review.

I'm not very sure why anybody may need to see so large FOV for the top camera, like mentioned 80 (in fact 40).

[Александр Зендриков]

As I understand it. The best option is 50 x 50?

пятница, 15 мая 2020 г., 10:48:24 UTC+3 пользователь Marek T. написал:

[Jarosław Karwik]

How big assmebly head do you have - how many nozzles etc. ? I would stay with 30x30 for the top vision and 80x80 for bottom vision.

I bought on ebay used 30x30 ready made setup and will have a lot of problems to find reasonable place for it on my head - even if I have monster size machine ( Phiolips CSM)

[Александр Зендриков]

The width of the Z axis is 80mm, one nozzle. I will use 30*30 glass well. If it's not enough, I'll redo it. My machine is small. Do you want to use the same glass for the lower camera? What will this design look like? Do you have a ready-made solution?

пятница, 15 мая 2020 г., 10:56:39 UTC+3 пользователь Jarosław Karwik написал:

[Marek T.]

I have the 3D case where is needed glass 32x45. Note the mirror is 45deg, so if you want have FOV xy=30x30, you need have glass like 30x45.

For the bottom vision I plan to do get FOV 40x40 so need the glass like 80x120, designing it just now but must improve my 3D printer first.

I can send you PM my project. It's custom for my machine and ELP, made in FreeHand.

[Александр Зендриков]

I would be grateful if you would share your project. You can save the project in .step format

пятница, 15 мая 2020 г., 11:56:17 UTC+3 пользователь Marek T. написал:

[Marek T.]

I'm not 3D designer. There is a lot of different sub-parts in the project, many needed and not needed.

So better to you install Freecad and take what you want converting choosen components to step.

I will mail it later to you.

[Александр Зендриков]

Thanks. I am waiting. Boil the file here? Or by email?

пятница, 15 мая 2020 г., 12:37:08 UTC+3 пользователь Marek T. написал:

[Jarosław Karwik]

If possible, I would like to have a look as well.

[Marek T.]

oki

[Marek T.]

I will see the size, maybe some ftp if large.

[Александр Зендриков]

How do you think this camera will work?

https://world.taobao.com/item/540821850895.htm?spm=a312a.7700714.0.0.OmBgem#detail

пятница, 15 мая 2020 г., 13:03:13 UTC+3 пользователь Marek T. написал:

[Florian Chende]

For most applications, smaller mirror is ok. I need larger because of its typical use in my case. I do not use bottom camera for small components, up to SO-8. Those will use laser alignment. Bottom camera is for larger parts only and I have modules up to 40x40mm. Another issue is the mirror angle. The mirror is square but the viewing area is not because of the mirror being placed at 45 degree. So an 80x80 mirror means an  80x56 viewing area. Yes, on the unmodified dimension, from 56 to 80 is a waste but I was not able to find suitable rectangular mirrors.

[Александр Зендриков]

To save space, can you cut off the extra part of the glass?

пятница, 15 мая 2020 г., 13:23:22 UTC+3 пользователь Florian Chende написал: