Accuracy of the NEMA 17 0.9°16micro-steps & THK KR2001A

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: ChrisR, Chris S., Pau, rjlittlefield

Adalbert
Posts: 1140
Joined: Mon Nov 30, 2015 1:09 pm

Accuracy of the NEMA 17 0.9°16micro-steps & THK KR2001A

Post by Adalbert »

Hello everybody,

I have just tested my new rail based on the THK KR2001A.

As far as I can see the carriage is moved by 1mm by one full rotation.
The controller of the NEMA 17 0.9 degree was set to the 16 micro-steps.
So, theoretically 6400 (16*400) micro-steps are needed for the full rotation and one micro-step should move the carriage by 0.000156mm (1/6400).
During my first test the carriage has been moved by 100mm and the step-counter has shown 638206.

I’m not sure if the difference of the 1794 micro-steps was caused by the inaccuracy of my measure-method with the folding yardstick or some steps were missed.

BR, ADi

mawyatt
Posts: 2479
Joined: Thu Aug 22, 2013 6:54 pm
Location: Clearwater

Post by mawyatt »

That's over a 1/4 turn of the stepper motor. This should be easy to see if you mark the stepper shaft (mark on the coupler to align with another mark on the fixture) like I do to evaluate the return to starting position.

I've run the KR20 and KR15 rails with my controllers over the full range at 1/8 micro steps for many many iterations, including many start and stop points in between and they always have returned to the start position as best I can determine. I even ran a rail testing session over almost an entire day (actually collectively over multiple days), back and forth without any apparent position error.

If you need a more accurate position indication, you might consider adding a micrometer to the stage setup or a tiny mirror and laser pointer off the motor shaft.

An interesting idea that would employ Zerene's ability to align images would be to take a high magnification & resolution image of a tilted subject with lots of detail at the start, then run the rail end to end for awhile, then return to the start and take another image. Use Zerene to stack the two images with alignments ON and look at the result log file which will show any alignment parameters. Ideally these parameters should not change if the rail returned precisely to the start position and the subject & lens/camera didn't move. I haven't tried this though, but would be an interesting experiment.

Best,
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

Adalbert
Posts: 1140
Joined: Mon Nov 30, 2015 1:09 pm

Post by Adalbert »

Hello Mike,

OK, firstly I only wanted to test the step size.

And yes, the best approach would be the taking of the photographs at the beginning of the way of the carriage and then the comparison of the offset between the runs (start->end->start).
1.) Move the carriage to the start point
2.) Take a photo perpendicularly to the movement of the carriage of a high-contrast point mounted on the side of the carriage
3.) Run the carriage to the end point (X-steps)
4.) Run the carriage from the end point to the start point (return X-steps)
5.) Take a photo (as in the point 2)
6.) Loop (n times)
7.) Evaluate the results of the Zerene-console

I have already tested my old rail in the common way:
http://www.photomacrography.net/forum/v ... c&start=30

BR, ADi

mawyatt
Posts: 2479
Joined: Thu Aug 22, 2013 6:54 pm
Location: Clearwater

Post by mawyatt »

I recall that thread, thanks.

Thought you were showing a error of 1794 micro-steps, ~0.28 a motor full turn, or 280 microns? That's a pretty big error for only a 100m run, and if it's accumulative it's even worse.

Have you found the source of the error?

Best,
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

ray_parkhurst
Posts: 2726
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

1/16 step has a fairly low current, so might be easy to miss the smallest steps. You may want to try increasing the current and try again.

Adalbert
Posts: 1140
Joined: Mon Nov 30, 2015 1:09 pm

Post by Adalbert »

Hello Mike, hello Ray,

I’m afraid I have suggested that some steps were missing but it wasn’t the case. Sorry for the confusion.

Theoretically 640000 (16 * 400 / 1 * 100) micro-steps should be done for the distance of 100mm.
- NEMA 17: 400 steps a 0.9° for 360° (full rotation)
- NEMA 17: 16 micro-steps a step
- THK KR2001A: 1mm a full rotation
- THK KR2001A: 100mm (distance)

But my rail has run this distance with 638206 micro-steps (less than necessary :-).
So, I would say that my method of the determination of the range was not good enough :-(
Image
Probably the paper has changed its position during the movement of the carriage.

BR,ADi

ray_parkhurst
Posts: 2726
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

Adalbert wrote:Hello Mike, hello Ray,

I’m afraid I have suggested that some steps were missing but it wasn’t the case. Sorry for the confusion.

Theoretically 640000 (16 * 400 / 1 * 100) micro-steps should be done for the distance of 100mm.
- NEMA 17: 400 steps a 0.9° for 360° (full rotation)
- NEMA 17: 16 micro-steps a step
- THK KR2001A: 1mm a full rotation
- THK KR2001A: 100mm (distance)

But my rail has run this distance with 638206 micro-steps (less than necessary :-).
So, I would say that my method of the determination of the range was not good enough :-(
Image
Probably the paper has changed its position during the movement of the carriage.

BR,ADi
Ahh, I was interpreting it the other way, ie that you made 640k and it only went 638.2k, ie it missed steps.

With your ruler method it would be difficult to tell if you went an extra 280um. The ruler may not even be accurately calibrated to 0.28% over that 100mm.

mawyatt
Posts: 2479
Joined: Thu Aug 22, 2013 6:54 pm
Location: Clearwater

Post by mawyatt »

I interpreted this the same as Ray.

Best,
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

Adalbert
Posts: 1140
Joined: Mon Nov 30, 2015 1:09 pm

Post by Adalbert »

Hallo Ray,
With your ruler method it would be difficult to tell if you went an extra 280um. The ruler may not even be accurately calibrated to 0.28% over that 100mm.
Actually the calculation of the step-size should be tested / confirmed by this test.
So, I may assume that the smallest step = 0.000156mm.
This must be verified :-) but this time with Zerene.
BR, ADi

Adalbert
Posts: 1140
Joined: Mon Nov 30, 2015 1:09 pm

Post by Adalbert »

Hello everybody,
The first test shows the behaviour of the rail with the 8 micro-steps a step.
In order to get 10 distances I have taken 11 photographs with the step-size = 5 micro-steps.
So, the expected step-size of the rail should be = 0,0015625 [mm].

Image
I haven’t found any reason why some steps are longer than expected/calculated.
Does anybody have any idea?

BR, ADi

ray_parkhurst
Posts: 2726
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

Adalbert wrote:Hello everybody,
The first test shows the behaviour of the rail with the 8 micro-steps a step.
In order to get 10 distances I have taken 11 photographs with the step-size = 5 micro-steps.
So, the expected step-size of the rail should be = 0,0015625 [mm].
...
I haven’t found any reason why some steps are longer than expected/calculated.
Does anybody have any idea?
Some will be longer, some will be shorter. Some may be 0, and then the system will catch up later with a big step. There are a few things going on that can cause this weird behavior. First, the stepper itself may not be rotating to the right position because even if the currents are correct by calculation, the stepper may not respond correctly to them. This can be caused by loading, insufficient drive current magnitudes, etc. The rail itself may not move as expected due to stiction or other mechanical effects. The errors you're showing seem systematic, so probably are not due to physical problems with the rail or motor bearings.

Adalbert
Posts: 1140
Joined: Mon Nov 30, 2015 1:09 pm

Post by Adalbert »

Hello Ray,
The errors you're showing seem systematic, so probably are not due to physical problems with the rail or motor bearings.
I have just tested the same scenario but with 16 micro-steps:
Image
As far as I can see every three steps comes a long one (with 8 micro-steps too).
BR, ADi

rjlittlefield
Site Admin
Posts: 20987
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

It is common that certain models of controllers and motors will give something like 1 long movement and 15 short ones, in each set of 16 micro-steps. Or similarly, 1 long and 7 short, in 8-microsteps mode.

An example is provided by the StackShot, when it is not in high-precision mode. See http://www.photomacrography.net/forum/v ... hp?t=16323 .

This very precise 1-in-16 behavior can easily be obscured when testing with units of multiple micro-steps. For example with 5 microsteps, a single long movement every 16 microsteps would be expected to appear as something very close to long/short/short/long/short/short.

--Rik

Adalbert
Posts: 1140
Joined: Mon Nov 30, 2015 1:09 pm

Post by Adalbert »

Hello Rik,
OK, I only will have to activate the high-precision mode :-)
BR, ADi

rjlittlefield
Site Admin
Posts: 20987
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

I was actually thinking that you might run a test with increments of 1 microstep, rather than 5. If you do have the problem that 1 microstep in every full step is long, then testing at each microstep should make make the pattern a lot more obvious.

For a long time the StackShot was believed to have irregular step sizes due to mechanical issues like stiction. But that interpretation was a mistake, caused largely by looking at step sizes that were multiples of several microsteps, which obscured the simple 1-in-N pattern. When somebody finally looked at each microstep separately, the real pattern became apparent immediately.

No way to know whether this will be the case for you also, but maybe it's worth running the test?

--Rik

Post Reply Previous topicNext topic