Build Automated Stack and Stitch Table with OpenBuilds

[mjkzz]

I had this thing built for me by a vendor for under 300USD long ago, after settling in new facility, I finally have time to try it. It is built from OpenBuilds parts, no custom work.

It is super stable, very precise, no wobble, no backlash, LOW PROFILE. The only drawback is that it can NOT be mounted vertically as the pinion drive does not do well vertically. That said, it can handle a lot of weight horizontally, I put a 1.2kg camera on it with heavy lens, it does not budge at all.

Since a lot people here have their own electronics with Arduino, etc, this is probably one way to build an automated SnS system, or even a semi-automated one -- at least move the table electronically.



Here is a video showing it is in action:

Youtube: https://www.youtube.com/watch?v=4Iyg5uFBC_c
Flickr : https://www.flickr.com/photos/60552763@ ... ed-public/

[ray_parkhurst]

Looks nice! And it answers my question about driving the SnS XY directly with timing belt. What is the effective step size without microstepping?

I still need to build my XY system for sensor-pan stitching. I have the basic design complete, but I'm vacillating again on timing belt vs precision rail for moving the camera. It's a lot of work to build and get it wrong, so would appreciate your feedback on step sizes and such. I don't really see much disadvantage to belt drive, do you?

[mjkzz]

Ray,

For XY table, I find belt driven system works fine horizontally. If you want to pan the camera, along with Z axis, you can remove the top aluminum extrusion and mount your Z axis rail right on top of the wheel plate for Y rail (in my coordinates) in my setup. You probably can make the X axis longer so you can remove it from the subject setup, then move it back in place.

Step size wise, without microstepping, with 0.9 degree motor, my XY table has resolution of 100um, with microstepping at 4th, it is 25um, at higher microstepping mode, it is even smaller. I think this is enough for a lot of situations at 4th. For coins, with full sensor at 5X and 30% overlap, step size is about 5mm or 5000um for X and Y, so even at 100um, this is pretty good.

[mjkzz]

Ray,

Though, I am reluctant to disassemble it, this is what I meant . . . mounting Z-axis on the table, cut the top aluminum short (at red line) or remove it completely.

Sure, you might be limited by how far your lens extends out from the Z rail.

[ray_parkhurst]

Ray,

For XY table, I find belt driven system works fine horizontally. If you want to pan the camera, along with Z axis, you can remove the top aluminum extrusion and mount your Z axis rail right on top of the wheel plate for Y rail (in my coordinates) in my setup. You probably can make the X axis longer so you can remove it from the subject setup, then move it back in place.

Step size wise, without microstepping, with 0.9 degree motor, my XY table has resolution of 100um, with microstepping at 4th, it is 25um, at higher microstepping mode, it is even smaller. I think this is enough for a lot of situations at 4th. For coins, with full sensor at 5X and 30% overlap, step size is about 5mm or 5000um for X and Y, so even at 100um, this is pretty good.

Yes, the step size can be large, but needs to be accurate for stitching to work without distortion. I guess you're using a 20-tooth gear on the motor? That would give 100um full-step resolution. Typical step accuracy for these motors is 0.05 deg, or +/- 6um if the system full step is 100um. But I believe that this is still 0.05 deg no matter how many steps you take, so for a 5mm movements (50 steps), this should still be +/-6um, or 0.12%. That would be perfect!

[mjkzz]

I am not sure about distortion. Distortion comes in when things are not squared up (not parallel), wobble can be one cause of this.

For X and Y, step size and accuracy is not that important (not to say it is NOT important :-)), it is the stability (wobble), backlash that are key. I find belt driven setup is perfect for this (backlash), provided you have the right belt tension (else backlash might be there) and the OpenBuilds system has excellent stability, very little or NO (so I claim under 10x) wobble.

[ray_parkhurst]

I am not sure about distortion. Distortion comes in when things are not squared up (not parallel), wobble can be one cause of this.

For X and Y, step size and accuracy is not that important (not to say it is NOT important :-)), it is the stability (wobble), backlash that are key. I find belt driven setup is perfect for this (backlash), provided you have the right belt tension (else backlash might be there) and the OpenBuilds system has excellent stability, very little or NO (so I claim under 10x) wobble.

I suppose wobble would depend on the concentricity of the idlers. If they are perfectly round and concentric, there should be no wobble.

Edited to add: Does the belt drive one of the idlers, or is it hard-attached to the carriage?

Further edit...I realized that the motors move with their respective carriages, so you must be driving an idler with the belt, correct?

[ray_parkhurst]

Ray,

Though, I am reluctant to disassemble it, this is what I meant . . . mounting Z-axis on the table, cut the top aluminum short (at red line) or remove it completely.

Sure, you might be limited by how far your lens extends out from the Z rail.

The system I'm contemplating needs something a bit different unfortunately. This system should work well for moving the subject in XY in a vertical stacking setup, but I'm trying for a vertical sensor-pan system. It's going to be interesting!

[mjkzz]

I am not sure about distortion. Distortion comes in when things are not squared up (not parallel), wobble can be one cause of this.

For X and Y, step size and accuracy is not that important (not to say it is NOT important :-)), it is the stability (wobble), backlash that are key. I find belt driven setup is perfect for this (backlash), provided you have the right belt tension (else backlash might be there) and the OpenBuilds system has excellent stability, very little or NO (so I claim under 10x) wobble.

I suppose wobble would depend on the concentricity of the idlers. If they are perfectly round and concentric, there should be no wobble.

Edited to add: Does the belt drive one of the idlers, or is it hard-attached to the carriage?

Further edit...I realized that the motors move with their respective carriages, so you must be driving an idler with the belt, correct?

The idler is more or less a position holder so that the belt can go through that tunnel, it is not driven.

[mjkzz]

Ray,

Though, I am reluctant to disassemble it, this is what I meant . . . mounting Z-axis on the table, cut the top aluminum short (at red line) or remove it completely.

Sure, you might be limited by how far your lens extends out from the Z rail.

The system I'm contemplating needs something a bit different unfortunately. This system should work well for moving the subject in XY in a vertical stacking setup, but I'm trying for a vertical sensor-pan system. It's going to be interesting!

If you take a look at the second image I posted, you will see that you can mount the Z on the XY table.

[ray_parkhurst]

Ray,

Though, I am reluctant to disassemble it, this is what I meant . . . mounting Z-axis on the table, cut the top aluminum short (at red line) or remove it completely.

Sure, you might be limited by how far your lens extends out from the Z rail.

The system I'm contemplating needs something a bit different unfortunately. This system should work well for moving the subject in XY in a vertical stacking setup, but I'm trying for a vertical sensor-pan system. It's going to be interesting!

If you take a look at the second image I posted, you will see that you can mount the Z on the XY table.

I did see that but I don't think it would work for my system. Perhaps if I mounted it upside down. I guess there would be no problem doing this.

[mjkzz]

Ray,

Though, I am reluctant to disassemble it, this is what I meant . . . mounting Z-axis on the table, cut the top aluminum short (at red line) or remove it completely.

Sure, you might be limited by how far your lens extends out from the Z rail.

The system I'm contemplating needs something a bit different unfortunately. This system should work well for moving the subject in XY in a vertical stacking setup, but I'm trying for a vertical sensor-pan system. It's going to be interesting!

If you take a look at the second image I posted, you will see that you can mount the Z on the XY table.

I did see that but I don't think it would work for my system. Perhaps if I mounted it upside down. I guess there would be no problem doing this.

I was going to suggest that -- upside down .

[ray_parkhurst]

Ray,

Though, I am reluctant to disassemble it, this is what I meant . . . mounting Z-axis on the table, cut the top aluminum short (at red line) or remove it completely.

Sure, you might be limited by how far your lens extends out from the Z rail.

The system I'm contemplating needs something a bit different unfortunately. This system should work well for moving the subject in XY in a vertical stacking setup, but I'm trying for a vertical sensor-pan system. It's going to be interesting!

If you take a look at the second image I posted, you will see that you can mount the Z on the XY table.

I did see that but I don't think it would work for my system. Perhaps if I mounted it upside down. I guess there would be no problem doing this.

I was going to suggest that -- upside down .

I was planning on stacking by moving the coin up/down, not the camera. This gets back to the "best way to stack" ideas, except now the additional layer of stitching is at work. Probably with the super long extensions I'm thinking about, the issue of mag changes will be fairly small. The lens I'll be using is not telecentric, so the question is if the mag changes due to WD/perspective shift of only moving the coin are larger or smaller than the changes due to moving only the sensor? I'll probably need to experiment to see.

So, where can I find more info on this XY table? Is it offered by OpenBuilds?

[mjkzz]

Yes, www.OpenBuilds.com should have all of them. Mine was built by a vendor, so I do not have individual part list.

Just checked OpenBuilds, they have a mini CNC for a good price, maybe, it is a good idea for you, just replace the spindle with camera :-)