Automated Stack & Stitch System Robots

[RDolz]

Mike, may not explain it well

X axis two rails 500 mm, 5 mm / rev, direct drive, each shaft an motor.
Y axis one rail 500 mm, 5 mm / rev, direct drive
Z axis one rail 250 mm, 4 mm / rev, drive with reduction 10: 1

The X and Y axes of my system have a step by revolution of 5mm.
The Z axis has a 4mm revolution step, but it has a 10: 1 reduction.

therefore the precision in the X and Y axes is 0.025 (5mm / 200 step motor) and 0.002 (4 / (200 * 10) micrometers in the Z axis.

Accuracy X axis = 5/200 = 0.025m
Accuracy Y axis = 5/200 = 0.025mm
Accuracy Z axis = 4 / (200 * 10) = 0.002mm

I advise you to install these reducers in all axes, in fact I am about to do it. Not because more resolution is necessary, but because you will have less vibrations.

Although the vibrations only occur when they move from one point to another. My program after each displacement, and before taking the picture, stops the time I want in order to eliminate the residual vibration of the movement.

I have more problems with the trains that pass near my house than with the vibrations of my system.

In my program, I always attack movements in the same direction in order to eliminate backlash problems in X and Y axes. But for these axes is not important: what does a change in the overlap of a few microns or so in a field of 5 or 6 mm matter ?.

With your setup do you notice any backlash or wobble (side to side tilt or motion) of the rails, and what pitch are the screw threads?

No, I have not detected that kind of problem. Yes on the Z axis, but I think it's because I had an accident when I was assembling the robot. The table, in which the robot was, collapsed and the Z axis was hit against a nearby shelf that started the bearings of their guides ... and the same I think the screw was deformed.Despite this I can work perfectly in the range of the magnifications in which I work.


Here I show you the error that produces the displacement of the Z axis. What you see is the trace that makes a bad pixel in the set of photos of the stacking of the Bryozoa. The HFOV is 8mm. They are 25 pixels in one direction and 68 pixels in the other. this represents an error of 0.04mm and 0.11mm. On the right side you can see the artifact generated by Zerene when making the automatic adjustments (X, Y)



Since the robot can withstand relatively high weights, you may feel more comfortable if you mount a double guide on the Z axis.

Put in any search engine "cnc Z axis slide" and you will see many options.

Just as an example:

[mawyatt]

Ray,

I'll be using a KR20 or KR26 as the vertical rail for focus stacking, but considering Fugu as a possible XY stage to move the subject. I've asked Fugu about the thread pitch, backlash and rail wobble.

Have some other rails I could use (another surplus KR20, Wemacro or MJKZZ) but they are mounted by cascading ARCA clamp/plates.

Best,

So based on Fuyu's numbering scheme, are you looking at the FLS40XY-H2? That one would allow you to mount your own Z-motor. You could still make the improvements I mentioned above for both stiffness and coupling.

Edited to fix typo should be FLS40XY-H2

Further edit to add link:

https://www.ebay.com/itm/Linear-Motion- ... 1800712164

This looks like quite a deal for $455 and free shipping.

Another edit...I thought when I selected the FLS40XY-H2 and copied the link it would revert to that model, but you still need to select it from the drop-down menu.

Ray,

Here's an example of the Fugu XY stage I am thinking about.

https://www.ebay.com/itm/283026117642

I should add that I don't think these rails are as good as the THK KR series, but for moving the subject for stack & stick efforts maybe good enough.

Best,

[ray_parkhurst]

Ray,

Here's an example of the Fugu XY stage I am thinking about.

https://www.ebay.com/itm/283026117642

I should add that I don't think these rails are as good as the THK KR series, but for moving the subject for stack & stick efforts maybe good enough.

Best,

How do you plan to mount the KR20, or do you plan to move the subject?

Edit...oops, that's what you said! For sure this should be OK for moving subject, but I would still recommend adding the idler per my original drawing.

[mawyatt]

Ray,

The KR20 (or 26) is directly bolted to the ThorLabs 95mm Sliding clamp which mates to the Thorlabs 95mm precision verticle bar. This is about as rigid as you can get, best secure clamping system I've ever come across, yet still easy to slide and remove. Simple, elegant design sliding clamp and rail support system from Thorlabs. The camera/lens will attach to the vertical KR focus rail.

Where I'm thinking of using the Fuyu XY stage placed on the Thorlabs horizontal 95mm precision one sided rail (silver colored extrusion below), directly bolted to another 95mm sliding clamp. The subject (chip) holding fixture (small ball head & attachment) will mount directly to the top Fuyu rail. I can already do this with a THK KR20, or Wemacro or MJKZZ focus rails used as XY positioners.

Here's a KR20 holding a MJKZZ on top for the XY stage. This works very well, so I'm not sure how much better the Fuyu solution would be. With a pair of KR20s I think I can get the vertical height under 100mm, the MJKZZ shown on top is higher than a KR20 and this combo is ~130mm, with a Wemacro on top (or bottom) this is ~150mm, the Fuyu setup is 144mm.

I'll probably hold off on the Fuyu now since I have a mechanically working solution and need to get the stepper drivers setup and the software completed, or to a usable state. Just don't want to tie up all 3 of my THK rails if I can get by with something else, that's not too expensive. I haven't tested the MKJZZ nor Wemacro for this setup yet either, and they may be perfectly fine.

Anyway lots of choices to consider.



You can see the Thorlabs verticle 95mm bar on the top of the image.

Here's another image showing the camera with your Nikon PN105 mounted to the focus rail and attached to another Thorlabs 95mm sliding rail clamp.





Best,

[ray_parkhurst]

Ray,

The KR20 (or 26) is directly bolted to the ThorLabs 95mm Sliding clamp which mates to the Thorlabs 95mm precision verticle bar. This is about as rigid as you can get, best secure clamping system I've ever come across, yet still easy to slide and remove. Simple, elegant design sliding clamp and rail support system from Thorlabs. The camera/lens will attach to the vertical KR focus rail.

Where I'm thinking of using the Fuyu XY stage placed on the Thorlabs horizontal 95mm precision one sided rail (silver colored extrusion below), directly bolted to another 95mm sliding clamp. The subject (chip) holding fixture (small ball head & attachment) will mount directly to the top Fuyu rail. I can already do this with a THK KR20, or Wemacro or MJKZZ focus rails used as XY positioners.

Here's a KR20 holding a MJKZZ on top for the XY stage. This works very well, so I'm not sure how much better the Fuyu solution would be. With a pair of KR20s I think I can get the vertical height under 100mm, the MJKZZ shown on top is higher than a KR20 and this combo is ~130mm, with a Wemacro on top (or bottom) this is ~150mm, the Fuyu setup is 144mm.

I'll probably hold off on the Fuyu now since I have a mechanically working solution and need to get the stepper drivers setup and the software completed, or to a usable state. Just don't want to tie up all 3 of my THK rails if I can get by with something else, that's not too expensive. I haven't tested the MKJZZ nor Wemacro for this setup yet either, and they may be perfectly fine.

Anyway lots of choices to consider.
...
You can see the Thorlabs verticle 95mm bar on the top of the image.

Here's another image showing the camera with your Nikon PN105 mounted to the focus rail and attached to another Thorlabs 95mm sliding rail clamp.
...
Best,

Mike...I did not mean using an idler with the Z-motor, but in conjunction with the Y-motor. In the pic it looks like you're using a KR20 as Y-motor? For sure this is very rigid, and you can probably get by with it as-is if it seems stiff enough. Even for my KR15's I added an idler to improve stiffness, but the KR20 is larger and may give a stiff enough platform for essentially zero-load (just the X-motor and light subject).

Edited to add: Glad to see the 105PN in use!

[RDolz]

https://www.ebay.com/itm/Linear-Motion- ... 1800712164

This looks like quite a deal for $455 and free shipping.

The FSL40, with other dimensions, is exactly the system that I buy.
It is a very oversized system to move samples. The Nema 23 and the mechanics are designed to support and move more than 5 kg of weight. As you can see the axis that supports the Z axis has a double rail to avoid twisting.
Mike, Fuyu has responded regarding the possibility of supplying rails with a lower screw pitch ?. When I bought it last year, there was no such possibility, because of that I assembled a 10: 1 gearbox on the Z axis.

[mawyatt]

Ramon,

They have replied but not indicated any details yet. I'll inform as soon as I get the information.

Best,

[ray_parkhurst]

The FSL40, with other dimensions, is exactly the system that I buy.
It is a very oversized system to move samples. The Nema 23 and the mechanics are designed to support and move more than 5 kg of weight. As you can see the axis that supports the Z axis has a double rail to avoid twisting.

There is always room for improvement in these systems, and there are ways to essentially eliminate unwanted movements. Assessing susceptibility is not too difficult:

- How long do you need to wait for the system to settle before taking a pic?
- After settling, how much movement do you still notice when viewing Live at 200%?
- How much movement happens when you push down on the camera with a small added force (maybe to push a button) when viewing Live at 200%?

[mawyatt]

Ray,

The KR20 (or 26) is directly bolted to the ThorLabs 95mm Sliding clamp which mates to the Thorlabs 95mm precision verticle bar. This is about as rigid as you can get, best secure clamping system I've ever come across, yet still easy to slide and remove. Simple, elegant design sliding clamp and rail support system from Thorlabs. The camera/lens will attach to the vertical KR focus rail.

Where I'm thinking of using the Fuyu XY stage placed on the Thorlabs horizontal 95mm precision one sided rail (silver colored extrusion below), directly bolted to another 95mm sliding clamp. The subject (chip) holding fixture (small ball head & attachment) will mount directly to the top Fuyu rail. I can already do this with a THK KR20, or Wemacro or MJKZZ focus rails used as XY positioners.

Here's a KR20 holding a MJKZZ on top for the XY stage. This works very well, so I'm not sure how much better the Fuyu solution would be. With a pair of KR20s I think I can get the vertical height under 100mm, the MJKZZ shown on top is higher than a KR20 and this combo is ~130mm, with a Wemacro on top (or bottom) this is ~150mm, the Fuyu setup is 144mm.

I'll probably hold off on the Fuyu now since I have a mechanically working solution and need to get the stepper drivers setup and the software completed, or to a usable state. Just don't want to tie up all 3 of my THK rails if I can get by with something else, that's not too expensive. I haven't tested the MKJZZ nor Wemacro for this setup yet either, and they may be perfectly fine.

Anyway lots of choices to consider.
...
You can see the Thorlabs verticle 95mm bar on the top of the image.

Here's another image showing the camera with your Nikon PN105 mounted to the focus rail and attached to another Thorlabs 95mm sliding rail clamp.
...
Best,

Mike...I did not mean using an idler with the Z-motor, but in conjunction with the Y-motor. In the pic it looks like you're using a KR20 as Y-motor? For sure this is very rigid, and you can probably get by with it as-is if it seems stiff enough. Even for my KR15's I added an idler to improve stiffness, but the KR20 is larger and may give a stiff enough platform for essentially zero-load (just the X-motor and light subject).

Edited to add: Glad to see the 105PN in use!

Ray,

The THK KR20 are very good, the KR26 probably even better since it's bigger.

The MJKZZ rail has some backlash and side to side movement, which is to be expected with a single support guide, the Wemacro and Stackshot are slightly better in the side to side movement because they have dual guides, again as to be expected. The MJKZZ, Stackshot & Wemacro are all pretty good though, and with some external loading (strategic placement of loading rubber bands) might just be "good enough" for the X & Y subject positioning when used in an overall vertical setup since gravity is not pulling against X or Y movements. However I don't think this will be sufficient for a horizontal setup, where the focus rail is horizontal. I'll need to do some extensive tests to find out what works OK and what doesn't, and report what I find.

I'm not sure the Fuyu system will be good enough for precision work, but probably perfectly fine for general purpose use. Again some testing will be required.

Now I'm buried in the software effort for the stack and stitch controller. Think I've got the hardware figured out, including the intricate details of the stepper motors and drivers for precision micro-stepping. Lots of very detailed stuff going on in the modern stepper controllers, including the pulse width modulation of the coils to allow current mode operation with higher supply voltages (higher instantaneous & holding torque) and the clever use of a complex commutating timing sequence of the H bridge to "recirculate" some of the coil current back to the supply. This creates a kind of switch-mode converter concept with the stepper coil current, and that current isn't just wasted. Thus the supply current is actually less than the motor coil currents!! The motor current induced magnetic fields are sustained and very smoothly controlled with this technique, which produces a smooth motor operation in rotation when operated correctly (I'm experimenting with one now). Motor torque does not behave as a constant function with micro-stepping, and can become quite weak at certain micro-step angles, which creates the opportunity for missed micro-steps. So another benefit of these advanced control techniques is the less likely skipping of motor steps, especially during micro-stepping.

All these advanced stepper motor control concepts can be attributed to the widespread precision CNC & printer use, and we should be able to directly benefit from such......if I could only be better at programming

But I'm learning, abet slowly

Best,

[ray_parkhurst]

Ray,

The THK KR20 are very good, the KR26 probably even better since it's bigger.

The MJKZZ rail has some backlash and side to side movement, which is to be expected with a single support guide, the Wemacro and Stackshot are slightly better in the side to side movement because they have dual guides, again as to be expected. The MJKZZ, Stackshot & Wemacro are all pretty good though, and with some external loading (strategic placement of loading rubber bands) might just be "good enough" for the X & Y subject positioning when used in an overall vertical setup since gravity is not pulling against X or Y movements. However I don't think this will be sufficient for a horizontal setup, where the focus rail is horizontal. I'll need to do some extensive tests to find out what works OK and what doesn't, and report what I find.

I'm not sure the Fuyu system will be good enough for precision work, but probably perfectly fine for general purpose use. Again some testing will be required.

Now I'm buried in the software effort for the stack and stitch controller. Think I've got the hardware figured out, including the intricate details of the stepper motors and drivers for precision micro-stepping. Lots of very detailed stuff going on in the modern stepper controllers, including the pulse width modulation of the coils to allow current mode operation with higher supply voltages (higher instantaneous & holding torque) and the clever use of a complex commutating timing sequence of the H bridge to "recirculate" some of the coil current back to the supply. This creates a kind of switch-mode converter concept with the stepper coil current, and that current isn't just wasted. Thus the supply current is actually less than the motor coil currents!! The motor current induced magnetic fields are sustained and very smoothly controlled with this technique, which produces a smooth motor operation in rotation when operated correctly (I'm experimenting with one now). Motor torque does not behave as a constant function with micro-stepping, and can become quite weak at certain micro-step angles, which creates the opportunity for missed micro-steps. So another benefit of these advanced control techniques is the less likely skipping of motor steps, especially during micro-stepping.

All these advanced stepper motor control concepts can be attributed to the widespread precision CNC & printer use, and we should be able to directly benefit from such......if I could only be better at programming

But I'm learning, abet slowly

Best,

All the signal processing development for steppers is extremely interesting. A lot of this development was done long ago for photolith, which has now gone to much more advanced stuff while CNC is benefiting from continued work. I wish I had the motivation to learn it.

My first effort at the automated XY stage used a surplus XY system which I had great hopes for: 2.5mm pitch screws with 400 step dual shaft motors seemed very promising. It was at this time I was having issues with the mjkzz controller on Win7, and I also had a looming project that required a total range of >30mm, which was just a bit larger than I could get with one of the axes. However, it was the annoying backlash that finished it for me and pushed me to use the KR15's. While backlash seems "acceptable" for XY stitching use, I find that it forces having to set a lot of alignment points, and makes the process much slower and more manual.

How are you doing the camera/lens mounting for the Z-Motor? As you know I am very obsessed with eliminating moments in my systems, and am curious what you are doing since you referenced high stiffness in your system.

Regarding the microstepping and reduced torque, I see this in a couple of steppers I intended to use to build a vertical stage like the one offered by Sigma Koki. That stage unfortunately has a 5-phase (0.72-deg) motor, which none of the readily-available/inexpensive controllers can drive, so I thought about building my own, but all the small motors I've tried have pitiful torque. What I would do if I were you (WIWDIIWY) and doing a complete software and motor drive solution, is to take advantage of the weight of the vertical setup in reducing the required torque for microstepping. If you always arrange your stacks to go from top to bottom, then you can make gravity your friend and minimize/eliminate any missed steps. Make the "up" movements at 1/4 step, and any extreme microstepping only happen on the "down" movements. Just my WIWDIIWY $0.02.

Since you're building your own controller, you might even consider going 5-phase. You get an immediate benefit to step angle, and from my understanding microstepping is easier to implement (but I don't really know why). I have a few Sigma Koki rails with 5-phase motors if you want to experiment...

[mawyatt]

Ray,

Yes agree, I suspect the semiconductor folks were originally behind some of this with the mask steppers and such.

I'll probably use the mount I have now for the THK KR20 or 26 which is directly bolted to the Thorlabs Sliding rail clamp. I drilled & tapped the clamp to accept either KR20 or 26 rails and also for an ARCA clamp. The PN105 is mounted with 3 MJKZZ lens clamps to a 160mm ARCA plate which is clamped to the KR20 or 26 platform. This produces a rigid lens/camera mount setup, better than just using the camera mount, although with the quality camera brackets like the RRS & Sunwayfoto do a good job also...but not as good as this lens clamped setup.

Good point doing the focus stacking from top to bottom, that's generally the way I stack in simple vertical only mode....highest subject point at the stack beginning and lowest at the end. Take advantage of every little detail you can for precision work!!

It's very interesting how the controllers are using the current thur the motor (which is what produces the torque) to augment the overall motor performance. Some folks are using 24 and even 32 volt supplies to drive small steppers to achieve high torque and fast acceleration, and are even "tuning" the motor/controller combo. Since I don't need fast acceleration nor massive torque I'll just use the standard (1.6 ohm, 2.4 amp) NEMA 17 motors with a 12 volt supply, but under current mode control. I've got a simple test setup running the motor now and it's quite smooth using the current commutating controlled decay mode.

Thanks for the offer on the Sigma, but I'll have to pass I've already got way too much on my plate now, especially the software effort.

Best,

[ray_parkhurst]

I'll probably use the mount I have now for the THK KR20 or 26 which is directly bolted to the Thorlabs Sliding rail clamp. I drilled & tapped the clamp to accept either KR20 or 26 rails and also for an ARCA clamp. The PN105 is mounted with 3 MJKZZ lens clamps to a 160mm ARCA plate which is clamped to the KR20 or 26 platform. This produces a rigid lens/camera mount setup, better than just using the camera mount, although with the quality camera brackets like the RRS & Sunwayfoto do a good job also...but not as good as this lens clamped setup.

Your arrangement eliminates the movements of a bellows setup very effectively, but the entire moment of the camera/extensions/lens is imparted across the relatively short length of the KR20/26 carriage. This is the situation I was in as well with my KR20 "A" rail, and it resulted in significant unwanted movements. I ended up buying some "B" rails (with 2nd "idler" carriage for reducing torque in these situations), but ultimately decided to move the idler well beyond/outside the rail in order to build a much longer and more stable, moving optical platform for the camera and lens. You might want to consider doing something similar.

[Smokedaddy]

You guys are way over my head with all this but it's very interesting to read. I'd like to know how you'll eventually quantify/measure the accuracy or precision of your rails?

[ray_parkhurst]

You guys are way over my head with all this but it's very interesting to read. I'd like to know how you'll eventually quantify/measure the accuracy or precision of your rails?

It's actually pretty easy. I simply watch the Live View 100% view to see how the system reacts when stacking. Ideally, what I want to see is:

- no vibrations between shots
- no x or y movements between shots

How well the system works depends on how far from the ideal I observe in the Live View. I will usually see some amount of vibration, though with very small microsteps sometimes there is almost none, and if I use a voice coil motor I generally see none at all.

For the movements, this depends on the quality of the rail and especially the screw and nut drive unit. Misalignment our un-centeredness can cause both x and y movements. How well the carriage is constrained makes a big difference as well. For the THK rails I use, I see almost no movement, though at 40x I can see a small amount, and I expect this would become visible (>1 pixel) at higher mags. Again, the voice coil shows no movements at 40x and I would expect none visible at 100x as well.

It's important to make sure the optical axis is aligned with the rail axis to eliminate systematic x/y offsets as the rail moves.

[Smokedaddy]

As usual I wasn't clear in my question. For example, how do you quantify that you've actually made a 5 micron step on the rail.