45 Degree angle mirror for upward facing camera

[thereza]

Wondering if there is a reason we can't use a 45 degree mirror with the upward facing camera? I am building my unit on 3/4" MDF and figure I should be able to route out a space for it and add the camera. Doesn't have to be too accurate as you should be able to calibrate out errors.

Reza

[JuKu]

We want to calibrate the needle when its tip is at PCB surface level. In principle, a mirror should work just like straight view. On the other hand, you need to carve in to the table anyway, and I wanted to avoid the potential complexity of a mirror, so I put the camera under the table.

[thereza]

I'm using a 3/4 inch MDF board as the base of my unit and it would be easy to carve out a section in that for the camera rather than going through that and into the table below it.

[PeterST]

it would be easy to carve out a section in that for the camera rather than going through that and into the table below it.

I didn't even think about it that way because we just got ourselves a nice IKEA table for it (80 euros including legs). But it would be true that it would make the unit better portable that way. So it might be a good idea after all ? (it wont bother me though)

Other option perhaps :
We use the same table tops as for our work bench, with the difference that the work bench is not supported by legs but by (cheap IKEA again) two drawer cabinets (and that several times so you have a loooong work bench). Now the table top just rests on those cabinets and can be taken off when desired. The hole for the up looking camera would be in between the cabinets.

Peter

IMG_1614a.JPG (136.08 KiB) Viewed 8913 times

[JuKu]

...it would make the unit better portable that way.

A loose tabletop with really short legs or even some 2x4's under it could make the machine movable.

[thereza]

I'm going to try using a mirror from another USB camera that I have

Capture.JPG (24 KiB) Viewed 8903 times

[Covert]

How did you go with the mirror ? I have some single surface mirrors here and I 3D printed. I'm thinking to do the same.

[JuKu]

Do note, that the needle calibration will be done at PCB surface level. In other words, the machine will lower the needle to do the measurement. If your goal is to avoid making a hole in your table, you need to mount the PCB higher.

And writing this, I realize that this poses a problem with camera scaling. I need to think of this: The software needs a z-position dependent camera pixel size feature in the measurement routines. Possibly, target height (looking at components) should somehow taken in to account as well.

[Covert]

Do note, that the needle calibration will be done at PCB surface level. In other words, the machine will lower the needle to do the measurement. If your goal is to avoid making a hole in your table, you need to mount the PCB higher.

And writing this, I realize that this poses a problem with camera scaling. I need to think of this: The software needs a z-position dependent camera pixel size feature in the measurement routines. Possibly, target height (looking at components) should somehow taken in to account as well.

Thanks for your thoughts on this JuKu. I was going to edit the source to raise the needle calibration once I get there, for the mean time I will just raise the PCB level. Any issue you can see with raising the calibration Z level ? (I have yet to even look at the source code).

[mrandt]

Hi Covert,

I have played a lot with alternative camera solutions - see photos and descriptions here:

http://liteplacer.com/phpBB/viewtopic.php?f=3&t=46

I just want to stress Juha's point here - needle wobble calibration HAS to be done at the same Z-level as PCB surface - otherwise it is pointless. If you measure it at an arbritrary height, the wobble may be different from what you have when the needle is lowered to surface level (mechanical reasons).

Also, for the measurements to work, you have to calibrate the pixel to mm ratio (box size for uplooking cam) at the same Z-level. The needle tip (and any offset from the center) appear smaller the farther they are away from the cam and larger the closer they move in.

Bottomline: If uplooking cam is on top of the table, you probably have to raise the PCB level as well, but room is limited. So I am inclined to think that mounting the camera inside or underneath the table is the easiest and safest solution.

@Juha, regarding variable component height I had a longer discussion with thereza. As I am using high-res cameras, the perspective error on both cameras is much more significant If I elevate the PCB by just 2mm, the pixel-mm ratio changes by almost 10%. The same of course applies to larger components, e.g. electrolytic caps or thick ICs as they will be nearer to the camera. Reza had a simplistic idea, how to solve this - basically move components incrementally until they are centered and take mechanical offset (steps) and not measured (cam) to compensate pickup error:
http://liteplacer.com/phpBB/viewtopic.p ... rt=20#p422

Regards
Malte