High resolution rotary shaft encoders

[perdrix]

I'm looking for high resolution absolute rotary shaft encoders.   The Renishaw Resolute encoders with RESA 150mm diameter scales provide accuracy of 1.91 arc-seconds, and resolution in 26 bit mode of 0.019 arc-second both of which are in the right ball park for the application.   They are also "simple" to fit.   The down side is the very high cost of over £1200 plus tax for one encoder ring and sensor.

That's why I am looking for alternatives.  So far I have found the Celera Motion Mercury II 600 encoders with a special disk that allows absolute encoding, but that has drawbacks in terms of fitment.   Another option I found this morning was the Vishay Sfernice RAMK090 which looks good, but I've been completely unable to find any availability or pricing data.

Are there any other options out there that I need to consider?

Thanks
David

[TopQuark]

How about the magnetic rotary encoder chips? Something like the MPS MA730 or the AMS AS5047. They are absolute positioning and outputs up to 14 bits. Both the AMS and MPS parts work fine in my experience.

[langwadt]

How about the magnetic rotary encoder chips? Something like the MPS MA730 or the AMS AS5047. They are absolute positioning and outputs up to 14 bits. Both the AMS and MPS parts work fine in my experience.

there's a long way from 14 to 29 bit ...

[perdrix]

Indeed it is a *very* long way from 14 bit ... I would be happy with anything much over 22 bits (0.34 arc-seconds).  23 bits (0.152 arc-seconds) or 24 bit (0.075 arc-seconds) would be fine.

D.

[CaptDon]

With that many bits and such a high resolution you will need some really fast logic for tracking. Is this part of a homebrew CNC type machine? You may have to approximate the position with the high order bits then slow the shaft and 'home-in' taking advantage of the lower order bits and slower shaft speed. Reminds me of the typical 'servo' sound of pick-n-place machines.
 

[perdrix]

Maximum rotation speed is pretty low (about 1 rpm) with normal movement speed being a fraction of that.  So even slow logic should be able to keep up in this application.

David

[jmelson]

I make a device that converts the proprietary Fanuc "serial pulse coder" format to quadrature.  Fanuc makes some VERY nice encoders in this series.  You can often get them on eBay.  They do have encoders that go over 1 million counts/rev.  The ones with an A in the model number have absolute data that allows them to be used as an absolute encoder, but the high resolution data may not be accurate unless a backup battery is provided.  So, an (alpha) I 64 would have 65536 counts/rev, but no absolute data, the (alpha)A64 would have the absolute data.
Feel free to ask if you need more info.    The position can be interrogated at up to 8 KHz.
Jon

[H.O]

Hengstler makes absolute encoders with various interfaces and resolutions (up to 22bits by the looks of it). I don't know of any prices.
https://www.hengstler.de/en/s_c100301/Rotary_encoder/Absolute_encoder/

[Benta]

I'm looking for high resolution absolute rotary shaft encoders.   The Renishaw Resolute encoders with RESA 150mm diameter scales provide accuracy of 1.91 arc-seconds, and resolution in 26 bit mode of 0.019 arc-second both of which are in the right ball park for the application.   They are also "simple" to fit.   The down side is the very high cost of over £1200 plus tax for one encoder ring and sensor.

At that kind of resolution, I think you'll have to live with the price. This is extreme precision stuff.

[Conrad Hoffman]

Gurley makes some superb encoders, both rotary and linear. They also have a nice USB interface that they don't promote as much as they should. I've used the 7700 series rotary encoders in custom test equipment because the price wasn't too bad and interfacing was easy.
https://static1.squarespace.com/static/5b9183b8fcf7fdac4d3a1067/t/5d23444a0b9e9c00019889ab/1562592334798/7700_datasheet.pdf

https://www.gurley.com/rotary-motion

The company goes back a very long ways, back to surveying transits in the 1800s. https://www.gurley.com/history

[jbb]

Uhh, I’m no expert and could be about to make a fool of myself, but at those angular resolutions you might be in Mechanical Engineering Trouble land. Things like slop and backlash and no doubt other things I don’t know about.

——

Some years ago I used an AMS encoder and found it worked largely as expected BUT if your axis of rotation, centre of magnet and sensitive zone of the chip don’t line up, you get a (repeatable and periodic) wobble in the output angle.

[jbb]

Oh, some further questions…

Do you need high resolution and absolute positioning at the same time?

I’m imagining something like a telescope tracker. That would call for high resolution to maintain smooth motion at low speeds, plus absolute positioning to mitigate risk of crashing into end stops.

Could it make sense to use a compound solution, eg a high resolution incremental encoder for smooth control and a medium resolution absolute encoder for ‘where the heck am I?’

[branadic]

Just curious, are you able to provide an eccentricity, on whatever cinematic you are using, low enough to benefit from such a high resolution and accuracy? Renishaw provides just a kid (encoder ring + read head), so whatever you apply the encoder ring to needs extrem low eccentricity. Just calculate the accuracy error of 10 µm eccentricity on 150 mm diameter. Immediately all your arcsec accuracy is gone.

-branadic-

[jmelson]

Uhh, I’m no expert and could be about to make a fool of myself, but at those angular resolutions you might be in Mechanical Engineering Trouble land. Things like slop and backlash and no doubt other things I don’t know about.

——

Some years ago I used an AMS encoder and found it worked largely as expected BUT if your axis of rotation, centre of magnet and sensitive zone of the chip don’t line up, you get a (repeatable and periodic) wobble in the output angle.

0.019 arc-second??!!??  Yes, that is in the region of insane measurement.  I have a Taylor Hobson Talyvel electronic level that reads out in tensth of an arc second.  When I put it on the bed of my 3500 Lb lathe and walk from one end of the lathe to the other, it shifts about one whole arc second.  That is the effect of my weight shifting the level of the concrete floor!  Juts touching anything in the measurement apparatus, or even breathing in the room will likely shift the readings by a few arc seconds.
Jon

[Conrad Hoffman]

You need excellent mechanics, but it's doable. Either preloaded precision bearings or air bearings. I went with the latter by New Way. https://www.newwayairbearings.com/catalog/air-bushings/

[thm_w]

Uhh, I’m no expert and could be about to make a fool of myself, but at those angular resolutions you might be in Mechanical Engineering Trouble land. Things like slop and backlash and no doubt other things I don’t know about.

——

Some years ago I used an AMS encoder and found it worked largely as expected BUT if your axis of rotation, centre of magnet and sensitive zone of the chip don’t line up, you get a (repeatable and periodic) wobble in the output angle.

Yeah, if this goes on a motor, does OP expect to be able to control the motor to that level of precision? Needs more actual details about the intended application.

For the AMS stuff, depending on the motor, you can do a full rotation and save all the data for calibration. I believe some ICs had a built in function for this, can't recall the part though.

[Someone]

Uhh, I’m no expert and could be about to make a fool of myself, but at those angular resolutions you might be in Mechanical Engineering Trouble land. Things like slop and backlash and no doubt other things I don’t know about.

——

Some years ago I used an AMS encoder and found it worked largely as expected BUT if your axis of rotation, centre of magnet and sensitive zone of the chip don’t line up, you get a (repeatable and periodic) wobble in the output angle.

0.019 arc-second??!!??  Yes, that is in the region of insane measurement.  I have a Taylor Hobson Talyvel electronic level that reads out in tensth of an arc second.  When I put it on the bed of my 3500 Lb lathe and walk from one end of the lathe to the other, it shifts about one whole arc second.  That is the effect of my weight shifting the level of the concrete floor!  Juts touching anything in the measurement apparatus, or even breathing in the room will likely shift the readings by a few arc seconds.
Jon

For sure, even an autocollimator on an optical bench is challenging to make repeatable measurements of fractions of an arcsecond (having used one to verify a rotary encoder!).

[perdrix]

Just curious, are you able to provide an eccentricity, on whatever cinematic you are using, low enough to benefit from such a high resolution and accuracy? Renishaw provides just a kid (encoder ring + read head), so whatever you apply the encoder ring to needs extrem low eccentricity. Just calculate the accuracy error of 10 µm eccentricity on 150 mm diameter. Immediately all your arcsec accuracy is gone.

-branadic-

Don't worry I'm not (quite) insane - anything at or below and arc-second resolution will be fine.  The BISS serial protocol only offers certain bit-depths and 26 bits is simply "more then adequate".