Support Request For A New Macro Rail Project

[Saul]

...it is simple to get very small focus steps by using a microscope focus block and motor control of the fine focus knob...

Fully agree with Rik. In my studio setup I'm using Labophot base and 400 stepper with WeMacro controller. So if for the 50x I'm using 20 steps in the slow mode, it gives me 1um (if my calculations are correct). If I'll use 1 step ... Did not try this mode yet.
And you can find labophot base on the ebay at the very good price

[Smokedaddy]

...

[mawyatt]

How would you measure and know sub-micron performance without very expensive equipment unlikely to be in the hands of us amateurs?

The measuring part is actually not difficult. You just watch from the side with high magnification optics and use the image alignment functions of Zerene Stacker to track movement of a finely detailed target. See the thread "Tiny focus steps: how to make them, how to measure them" for movements & measurements summarized as "This is a sequence of 132 moves, the largest of which measures a bit under 0.14 microns (average 0.0625 microns)."

But that thread also illustrates another fact: it is simple to get very small focus steps by using a microscope focus block and motor control of the fine focus knob. See also the recent post by nanometer for an effective setup that doesn't even involve any machining.

I applaud lonepal's enthusiasm for pushing the envelope by mastering the Parker servo. But to make sense of that enthusiasm, I have to consider the effort as an educational and entertaining exercise in its own right. If the goal were only to get small and stable focus steps, there are much simpler approaches.

--Rik

Rik,

This looks like it would show sub micro measurement performance at least to the levels I mentioned (0.625 and 0.3125 microns), but I would hesitate to say that it can achieve accurate performance measurements at levels below this.

Best,

[sushidelic]

Need 5nm steps? Here ya go:



Nice gadget, I only needed the "coarse" dials till now.

[Smokedaddy]

...

[sushidelic]

Just forget about the piezo actuators (copper ports), no one needs that precision. And those coaxial micrometers are manual only.
If you really want to use the piezos (and if only to impress someone at the bar), that ain't too complicated either.

[Smokedaddy]

...

[mawyatt]

You do realize you are talking about 17 Aluminum atom diameters. If I breathed on that it would move more than 5nm

Best,

[rjlittlefield]

This looks like it would show sub micro measurement performance at least to the levels I mentioned (0.625 and 0.3125 microns), but I would hesitate to say that it can achieve accurate performance measurements at levels below this.

I understand your skepticism, but I have no hesitation in saying exactly that.

0.1 micron at 50X is 5 microns on sensor. That's more than a full pixel shift on any modern DSLR, and from extensive experiments I know that the alignment methods used by Zerene Stacker are good to roughly an order of magnitude smaller than that.

The key insight is that the alignment method works by minimizing an error term that is integrated over all the pixels in the image -- typically 20 million of them. Using a random pattern target, any issues there might be with subpixel interpolation tend to disappear in the averaging.

By far the biggest difficulties that I've had in making the measurements have been associated with mechanical stability. The one that had me flummoxed for the longest time eventually turned out to be a bellows front bayonet mount that was shifting laterally by a small amount on every exposure due to mirror/shutter shock. A few drops of superglue, and the data cleaned right up!

--Rik

[mawyatt]

Rik,

Integrating over the entire image (20MP) certainly should provide a better result, but what about the optical effects introducing errors. Say I want to step 20.15625 microns (8 and 1/16 step at 2.5 microns per step). Could Zerene actually interpret this larger step with this level of precision?

Agree, the mechanical (and thermal) requirements become very challenging, even more so for repeatability. Anything that moves must be carefully evaluated, and the same that goes for things that aren't supposed to move...as you found with the bellows bayonet (that must have been difficult to track down

Anyway, I can't see any need for my use beyond what the THK rails can provide with the Stackshot controller (unless I can get some of these new metamaterial lenses...other thread

Best,

[sushidelic]

Anyway, I can't see any need for my use beyond what the THK rails can provide with the Stackshot controller

+1

Best regards,
Michael

[rjlittlefield]

Say I want to step 20.15625 microns (8 and 1/16 step at 2.5 microns per step). Could Zerene actually interpret this larger step with this level of precision?

I'm not sure exactly what question you're asking, so let me provide an analogy.

The optical measurement is like a nearly frictionless spring scale with major tick marks from say 0 to 500 and 10 minor ticks per major tick -- so 5000 ticks total. The spring is interchangeable so you can have pretty much any range you want, but since it's a spring, you're never quite sure what its spring constant is, or even whether it's really constant. Spelling out the analogy, "major tick marks" are full pixels on sensor, "minor tick marks" are sub-pixel interpolation, and "spring" is the lens with its attendant magnification and possible distortions.

So, if you want to measure 20 pounds (or ounces, or grams, or microns), first you have to pick the appropriate spring, and then you can measure to a precision of 20/5000 = 0.004 pounds (or ounces, or grams, or microns). But you don't know anything about accuracy or linearity unless you have calibrated the spring.

In the optical system, there is the additional wrinkle that the alignment step is really trying to match the phase of two complex waveforms. It cannot determine matched phase with perfect accuracy, so there's also some lower limit that depends on the bandwidth of the signals. In other words, if the image is sufficiently fuzzy, then the alignment cannot be determined accurately to 1/10 of a pixel due to pixel noise, vignetting, and so on. In the spring scale, a corresponding problem would be some slight remaining friction and environmental noise that would establish a lower limit on reproducibility.

You've asked specifically about 20.15625 microns. Because of the "5000 ticks" issue, the measurement could distinguish between 20.152, 20.156, and 20.160 units, but those last 2 1/2 digits of your number would be out of reach.

Specifically with microns, and assuming 400 nm light at NA 0.8, the signal to be phase matched would have wavelength 500 nm = 0.500 microns on subject. I don't really know where the limit is for this sort of computational phase matching, but if it's say 1/100 cycle, then that would be another limit around the 5 nm level.

Whether the limit is 5 or 10 or 20 nm I don't know, but I'm quite confident that it's way smaller than the 312 nm that you pushed back against. Take a look at the "Actual Movement at Finest Microsteps" graph, image 6 at http://www.photomacrography.net/forum/v ... hp?t=27549, and note that the scale of that graph is 500 nm per coarse grid interval.

Anyway, I can't see any need for my use beyond what the THK rails can provide with the Stackshot controller

+1

+1 again. Microstep control of the fine focus knob of a microscope at 100 microns/turn is definitely overkill. On the other hand, it's simple and cheap overkill, especially if you can live with the limitations of a rig like the one that nanometer shows.

--Rik

[Macro_Cosmos]

Short answer is no, the Wemaco controller and Stackshot will not be able to control that rail. Both controllers are only able to handle two-phase bipolar stepper motors, not servo motors.

Servo motors require a position encoder to communicate with the controller, and they are extremely difficult to drive. The advantage of a servo motor is its ability to drive heavy items at high speed due to its large torque, hence the size of that thing. To slow it down, often or not a gear box is needed, adds to the already complicated system. You'll be better off purchasing an industrial rail that is controlled by a two-phase bipolar stepper motor and connect it to a stackshot/wemacro controller. Wemacro uses a GX16-4 aviation plug: https://www.ebay.com/sch/i.html?_from=R ... 6&_sacat=0

[mawyatt]

Say I want to step 20.15625 microns (8 and 1/16 step at 2.5 microns per step). Could Zerene actually interpret this larger step with this level of precision?

I'm not sure exactly what question you're asking, so let me provide an analogy.

The optical measurement is like a nearly frictionless spring scale with major tick marks from say 0 to 500 and 10 minor ticks per major tick -- so 5000 ticks total. The spring is interchangeable so you can have pretty much any range you want, but since it's a spring, you're never quite sure what its spring constant is, or even whether it's really constant. Spelling out the analogy, "major tick marks" are full pixels on sensor, "minor tick marks" are sub-pixel interpolation, and "spring" is the lens with its attendant magnification and possible distortions.

So, if you want to measure 20 pounds (or ounces, or grams, or microns), first you have to pick the appropriate spring, and then you can measure to a precision of 20/5000 = 0.004 pounds (or ounces, or grams, or microns). But you don't know anything about accuracy or linearity unless you have calibrated the spring.

In the optical system, there is the additional wrinkle that the alignment step is really trying to match the phase of two complex waveforms. It cannot determine matched phase with perfect accuracy, so there's also some lower limit that depends on the bandwidth of the signals. In other words, if the image is sufficiently fuzzy, then the alignment cannot be determined accurately to 1/10 of a pixel due to pixel noise, vignetting, and so on. In the spring scale, a corresponding problem would be some slight remaining friction and environmental noise that would establish a lower limit on reproducibility.

You've asked specifically about 20.15625 microns. Because of the "5000 ticks" issue, the measurement could distinguish between 20.152, 20.156, and 20.160 units, but those last 2 1/2 digits of your number would be out of reach.

Specifically with microns, and assuming 400 nm light at NA 0.8, the signal to be phase matched would have wavelength 500 nm = 0.500 microns on subject. I don't really know where the limit is for this sort of computational phase matching, but if it's say 1/100 cycle, then that would be another limit around the 5 nm level.

Whether the limit is 5 or 10 or 20 nm I don't know, but I'm quite confident that it's way smaller than the 312 nm that you pushed back against. Take a look at the "Actual Movement at Finest Microsteps" graph, image 6 at http://www.photomacrography.net/forum/v ... hp?t=27549, and note that the scale of that graph is 500 nm per coarse grid interval.

Anyway, I can't see any need for my use beyond what the THK rails can provide with the Stackshot controller

+1

+1 again. Microstep control of the fine focus knob of a microscope at 100 microns/turn is definitely overkill. On the other hand, it's simple and cheap overkill, especially if you can live with the limitations of a rig like the one that nanometer shows.

--Rik

Stated another way. I want to return (repeatability) to a previous position that was 4 and 1/16 steps away, with a step size of 2.5 microns.

An example might be returning to the starting point of a stacking session and moving to a stack point where the image wasn't captured properly (maybe the flash didn't fire, someone slammed the door or walked near the setup). But to get to that very point you must first go to the starting point with some degree of precision, then make the proper number of steps & micro steps with some degree of precision, or you could just step & micro step from the present position to the desired position without returning to the start. This would be a case where good repeatability & precision would be important.

Anyway, not to distract the OP I'll repeat what I first noted.

"This looks like a lot of work and requires specialized skills in programming and electrical engineering as well as control theory. If you have those skills or willing colleagues this might be an interesting project to attempt to tackle, however if your intent is to get a really good and stable focus stacking rail system capable of 20X or better other options might be worth considering."

Best

[lonepal]

Hi;

Thanks everybody for the informations and advices.
I made some investigations and decided that using this type of linear rails are not easy.

Lots of time+programming needed so I gave it up.

May be I will try it when I have time and energy to make it usable for macro works