High precise rotary movement/measurement

[simasa]

Hello,
I have a 50mm x 25mm x 4mm glass plate. It will rotate (CW and ACW) along its middle axis, about 20 times in a second, but only in a range of (+1°-1° to +20°- 20°). Very precise measurement upto 0.5 arc seconds accuracy is required.
To move the plate, I may use a high precise motor. To achieve high degree of precession level, I may possibly use backlash free high down gearing (upto 500 - 1000). Additionally, I may use a high resolution encoder (with interpolated output).
This is just an idea at the moment. One of the practical concerns could be the mechanical play in design assembly.

I may also use a plate, comparatively bigger in size. Instead of moving the mirror angularly, I may also be able to perform my desired operation by moving it linearly back and forth (+10mm to -10mm). But this would be the second preference.

I hope for some assistance in this forum regarding the followings.

Mechanical feasibility
Suggestions regarding high precise motors/encoders/gearboxes for such an application
Any other alternative suggestion/mechanism

Regards
Thanks in advance

Sims

 

[Skogsgurra]

This is a rather high precision. Less than 1 ppm counted over 360 degrees.

The weight of the plate is substantial, at least if you are going to move it 20 times per second. I doubt if any motor can do that with the required precision in the available time.

The limited movement can be positive factor since a two-wire galvanometer arrangement could be used for low friction and infinite resolution. But the weight of the glass plate is a problem.

In any case, there are no mechanisms that can produce the desired accuracy directly. You need to measure the angle and use the result in the outer position loop. Even the measurement can be very difficult because any displacement of the center of rotation ruins precision.

I have never been able to get better than a few arc-seconds. I did it with a modified (other, tighter, bearings) stepper motor that I calibrated and applied micro-stepping to. PWM made motor current resolution easy. And I didn't need to move as fast as you do.

A very interesting challange. But I am sure someone has done it. Watching answers with great interest.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

Skogs am I missing something? He seems to have described a microscope slide and you are saying "Heavy"?

Would seem a galvanometer is perfect for this and could be bought off-the-shelf.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[Skogsgurra]

Just try and hang a slab of thick glass on a galvo loop and see what happens. We are talking ppm here. And fast movement.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

http://cgi.ebay.com/General-Scannin...th-Mirror_W0QQitemZ160160100866QQcmdZViewItem

I don't know...
Some of these mirrors look like they mass way more than the described piece of glass. See the ones pictured in the PDF.

http://www.camtech.com

http://www.camtech.com/news/Choosing_A_Galvanometer.html

http://www.camtech.com/applications/ChoosingGalvo.PDF

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[waross]

You could keep the inertia of the measuring system down by using a beam of light reflected from the plate onto an array of photo detectors. Increase the radius and do the math to achieve any resolution you want. You will have a digital output from the detector array which should be easy to interface to the PLC.
respectfully

 

[Skogsgurra]

Have you two actually tried this? At 20 Hz? With a slab of glass being 4 mm thick? And two by one inch? And got a 1 ppm accuracy?

It is a lot more difficult than you think. And that is probably why the OP asked in the first place.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[IRstuff]

To achieve your stated requirement: 40º in 1/40th of a second, requires 4500 rad/s^2 acceleration, which is quite sporting.


General Scanning and Cabridge Technology are two companies that supply commercial scanners. Their specs are way worse than your requirements.

TTFN

FAQ731-376

 

[MikeHalloran]

For +/- 1 deg, I'd use a flexure and a voice coil.

For +/- 40 deg, I'd look into a constantly spinning optical polygon.



Mike Halloran
Pembroke Pines, FL, USA

 

[PNachtwey]

If the encoder provides 1 count per .5 arc seconds then the encoder counter needs to be able to count at over 18.1 MHz.

If you need only one then be prepared to pay $$$.

Peter Nachtwey

 

[blacksea]

PNachtwey: encoder with analog out (sin/cos) allow reaching high resolution (by external 4096 or even more interpolation factor) with lower frequency of analog signals.

 

[IRstuff]

0.5 arcsec ACCURACY is not going to be done through interpolation

TTFN

FAQ731-376

 

[blacksea]

IRstuff: I told about interpolation of analog sin/cos encoder - see Heidenhain encoders.

 

[sreid]

Canon also make some high resolution encoders (and some other interesting products).

http://www.usa.canon.com/html/industrial_encoders/lre_index.html

IRSTUFF is correct, 0.5 arc second accuracy is very difficult. But is it really 0.5 arc second resolution you need? If so, that can be done using, say, Heidenhain's flash/parallel output interpolator boards.

Forget gearboxes and think about direct drive and air bearings (New Way Air Bearings).

 

[simasa]

Thank for the all expert assistance.
The 0.5 arcsec accuracy and 20 Hz is presently required for my system. If I could be able to solve some other adjacent constraints, I may be able to come down to the requirements of 1-2 arcsec and 5 Hz.

@ Blacksea, @sreid
Use of Heidenhain's solutions seems to theoretical approaching the requirements. A concern could also be to find a drive with minimum play (repeatability/accuracy). Min. I found is +/- 10 arcsec. Some assistance in this regard would be really be desirable!

Due to inertia concerns (mass of the plate 8-10 g) “Galvo meter” seems to be a hard choice. If I could ensure a “Galvo meter” for my application, solution by “waross” would be the smart one. @waross; could you please further assist which kind of light beam you mean for such a solution.

Thanks and Reagrds
sims

 

[sreid]

For the linear implementation you describe in your first post, you could use a Laser Interferometer. See

http://www.renishaw.com/en/6482.aspx

These tend to be very accurate especially in a vacuum. In air you need to control temperature, pressure and humidity ( or the compensate for changes). High speed, high resolution and the retro reflector is light weight.

Heidenhain has a similar product.

 

[blacksea]

sreid: Renishaw has special compensation unit for laser:

http://www.renishaw.com/en/6486.aspx

But sin/cos 1Vpp encoder from Heidenhain maybe cheaper solution vs laser one.

 

[blacksea]

P.S. Single axis controller with build-in amp and sin/cos interface from Elmo would be suggested - see Cello for example (

http://www.elmomc.com).