Return to the Cloudy Nights Telescope Reviews home pageAstronomics discounts for Cloudy Nights members
· Get a Cloudy Nights T-Shirt · Submit a Review / Article

Click here if you are having trouble logging into the forums

Privacy Policy | Please read our Terms of Service | Signup and Troubleshooting FAQ | Problems? PM a Red or a Green Gu… uh, User

Equipment Discussions >> Mounts

Pages: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | (show all)
Starhawk
Space Ranger
*****

Reged: 09/16/08

Loc: Tucson, Arizona
Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5910765 - 06/09/13 01:48 AM

Is the plate attached to the back of the mount still the thin sheet with CCA standoffs? That should be beefed up as well.

What does it do if you unlock the axis and swing through 360° with no OTA attached?

-Rich


Post Extras: Print Post   Remind Me!   Notify Moderator  
orlyandico
Postmaster
*****

Reged: 08/10/09

Loc: Singapore
Re: Encoder-based PE Correction on the cheap new [Re: Starhawk]
      #5910781 - 06/09/13 02:13 AM

Yes there still is that thin sheet... but it's only there to prevent the encoder from rotating. All the physical support is provided by the extremely beefy shaft. I have rotated the mount 360* a couple times (to calculate the huge error). Have not seen how that sheet reacts...

Post Extras: Print Post   Remind Me!   Notify Moderator  
orlyandico
Postmaster
*****

Reged: 08/10/09

Loc: Singapore
Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5914358 - 06/11/13 12:19 AM

Update: the large periodic error is still there. I believe I've tracked it down to the torque mount on the encoder.

A torque mount of this type will invariably induce a tilt in the encoder. So a flange-mount encoder would be ideal.

The challenge is, there is a 3-month lead time if I order the Heidenhain flange mount encoder, and I'd have to re-fabricate the shaft coupling because Heidenhain doesn't make a 1" bore encoder. Having no lathe really bites at times like these..


Post Extras: Print Post   Remind Me!   Notify Moderator  
vdb
sage


Reged: 12/08/09

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5914399 - 06/11/13 01:03 AM

I did look at your first post with pictures, is this still the way the encoder is fixed? If you look at the TDM you see the adaption is really solid, so I don't think you need a new encoder but better adaption of the current encoder ...
Wild guess, would the CGEM TDM adaptor not be adaptable to your encoder?


Post Extras: Print Post   Remind Me!   Notify Moderator  
orlyandico
Postmaster
*****

Reged: 08/10/09

Loc: Singapore
Re: Encoder-based PE Correction on the cheap new [Re: vdb]
      #5914421 - 06/11/13 01:39 AM

Unlikely, the TDM uses a specific Heidenhain encoder (the one with a 3-month wait list if I order - and pay - now).

Also the CGEM adapter for the TDM costs $400. I'm not willing to pay that.


Post Extras: Print Post   Remind Me!   Notify Moderator  
SWR
member


Reged: 07/23/10

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5914494 - 06/11/13 03:45 AM

Here's a crazy idea for a low cost 2M PPR DIY digital encoder.

Use a Blu-ray disk as the encoder wheel and a PCB with three blu-ray pickups. Two pickups will generate the quadrature signal, and the last encoder will generate a coded index signal to give an absolute position after a few hundred counts in either direction.

The encoder disk will be generated by programming the three tracks on a Blu-ray disk. There's a description on Wikipedia under "Optical disk" where you can see the resolution of the Blu-ray disk. The shortest pulse is 130nm which will give a quadrature resolution of 65nm.

The inside tracks are 46mm diameter, which will give a resolution of 2.22M counts per revolution corresponding to a 0,58" resolution.

If it can be aligned with sufficient accuracy, the trimmed down Blu-ray disk could be mounted under the orange ring instead of your thin aluminum plate. The PCB can be mounted on the RA axis using a hollow adapter that will allow the feed through of a wire to the guide port routed internally in the CGEM.

The pin on the guide port that is not connected, should be wired to 5V inside the CGEM. This way you have power supply and guide signals in the same wire.

The small microcontroller on the encoder PCB should implement the guide pulses as open-collector so that a guide camera can be connected in parallel.

If the alignment issues can be worked out, I think this will be a very low cost solution.


Post Extras: Print Post   Remind Me!   Notify Moderator  
orlyandico
Postmaster
*****

Reged: 08/10/09

Loc: Singapore
Re: Encoder-based PE Correction on the cheap new [Re: SWR]
      #5914495 - 06/11/13 03:51 AM

That's the thing. The alignment issue is the biggest issue, I'm starting to realize...

I've tried to allow the torque mount to have a degree of freedom parallel to the encoder shaft (so it just slides along rather than tilt the encoder).

This doesn't work. I've noticed that the TDM uses the through-hole version of the Heidenhain encoder, but the difference between the Heidenhain and the Baumer I'm using, is that the H. has compression collars on both ends of the bore, whereas the Baumer only has a compression collar on one end.

I imagine the two compression collars help align the shaft more accurately.


Post Extras: Print Post   Remind Me!   Notify Moderator  
SWR
member


Reged: 07/23/10

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5914535 - 06/11/13 05:16 AM

The advantage of using the mount as alignment is that you have no bearings that have to align. The bearings on the RA axis are already aligned. The only fiddely thing would be to push the small piece of Blu-ray disk under the orange ring around, until it is perfectly aligned with the RA axis. The PCB with the Blu-Ray pickups is automatically aligned with the RA-axis because it is mounted on this axis. Alignment of the PCB adapter is not important. It will just change the diameter of the pickup circle.

I suppose that this can be aligned by having several similar tracks side by side on the Blu-ray disk. Which one it catches is not important. This way the only alignment is the centering of the Blu-ray disk. I admit that this can be difficult, but at least it's only one thing that must be adjusted.

Perhabs an accurate hole with the right diameter can be made in the disk using a lathe? Otherwise it must be aligned manually.


Post Extras: Print Post   Remind Me!   Notify Moderator  
orlyandico
Postmaster
*****

Reged: 08/10/09

Loc: Singapore
Re: Encoder-based PE Correction on the cheap new [Re: SWR]
      #5914762 - 06/11/13 10:15 AM

I think you underestimate how accurate the centering should be..

In my case, the encoder has a 1.000" bore. My original shafts were 0.992" (a difference of 0.008") and the resulting encoder swash and errors were huge. The replacement shafts are 0.998" (a difference of 0.002") and there still is significant error.

Also if the disk is not exactly parallel to the pickups (i.e. it is not perfectly perpendicular to the RA shaft) this will also result in encoder swash and errors.

There is a very good treatment of the numerous errors that afflict encoders here -

http://resources.renishaw.com/en/download/white-paper-the-accuracy-of-angle-e...

in particular check out the error induced by bearing wander, encoder swash, and eccentricity (all these errors are of the same magnitude):

Angular measurement error (arc seconds)
= bearing wander (µm) x 412.5/D
where D is the diameter of the encoder scale in mm

So if you have a 100mm disk, a bearing wander of 10 micrometers would result in 41 arc-seconds of error. So the 2 million ticks (~ 0.6 arc-second) is empty resolution, and is completely swamped by bearing wander/eccentricity/encoder swash.

And 10 micrometers is 0.0003937 inches (0.01mm). I don't think a blue ray disk could even be dimensionally stable enough to meet that spec..


Post Extras: Print Post   Remind Me!   Notify Moderator  
SWR
member


Reged: 07/23/10

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5914888 - 06/11/13 11:30 AM

Thank you for the link. I will study this in detail.

Does it matter that the accuracy is 41" as long as the resolution is 0.6"? If the error is a slow changing error once per revolution, it can be removed in software by PEC training the encoder. If higher frequency cyclic errors from the gearbox and non repeatable errors from wind etc. can be removed, it should be possible to calibrate the slow moving cyclic encoder errors with a guide camera.

One idea that springs to my mind is this: One way to completely remove all alignment issues, would be to mount a blank disk and burn the pattern on the mount using the pickups when both disk and PCB has been mounted securely. The only requirement would be that the RA axis is spun at an accurate speed. Maybe a small/weak motor and a flywheel on an unloaded RA axis that is allowed to stabilize over several minutes could be used? The software could start by burning a single index pulse, and monitor the frequency of this pulse with great accuracy. This could then be used as an accurate timing reference.

Just some loose thoughts.


Post Extras: Print Post   Remind Me!   Notify Moderator  
orlyandico
Postmaster
*****

Reged: 08/10/09

Loc: Singapore
Re: Encoder-based PE Correction on the cheap new [Re: SWR]
      #5915044 - 06/11/13 01:06 PM

Yes that's what I'm planning to do.. PEC-train the encoder, because I can't get the eccentricity out of the system.

This however requires that you have an additional, absolute indexer - a normal PEC indexer is a single slot, but a worm rotates once every 4-10 minutes. Since the RA shaft only rotates once in 24 hours, you can't determine the PEC index with a single slot: you need an absolute encoder.

You can't calibrate the disk by spinning the RA shaft at a fixed speed. Because how can you ensure a fixed speed when any motor or geartrain you'll use to spin the axis would also have its own periodic errors.

Many commercial encoder manufacturers calibrate their higher-end encoders using the concept you've suggested.. but they use a laser interferometer with sub-arcsecond resolution, or a high-resolution master encoder, to provide the "known good" readings. By mapping the "known good" angular readings with the (distorted) output of the encoder under test, they can construct an error map specific to that encoder.

The problem is... the error map only takes into account the eccentricity of the encoder "stand alone." Once that encoder is installed, bearing wander, axle swash, etc. will induce its own periodic error which the "canned" map in the encoder can't compensate for. So for the really high-end positioning systems like IC etching machines, they do calibration with an interferometer on the completed system.

Here are some more encoder references I've coughed up in the course of this project.. (most are available online, but some require IEEE membership and payment)


Selection of a tape encoding system for the main axis of the Gemini telescopes, John D. Wilkes & Martin Fisher, Proc. SPIE 3112, Telescope Control Systems II, 30 (September 18, 1997); doi:10.1117/12.278836, From Conference Volume 3112, in Telescope Control Systems II, ed. Hilton Lewis, July 27, 1997


Yang, Y., Rees, N. P., & Chuter, T., Application of a Kalman Filter at UKIRT, Astronomical Data Analysis Software and Systems IX, ASP Conference Proceedings, Vol. 216, edited by Nadine Manset, Christian Veillet, and Dennis Crabtree. Astronomical Society of the Pacific, ISBN 1-58381-047-1, 2000., p.279

Watanabe, T., Fujimoto, H., & Masuda, T. Self-Calibratable Rotary Encoder, Journal of Physics: Conference Series 13 (2005) 240-245

Erm, T. & Sandrock, S. Adaptive Correction of Periodic Errors Improves Telescope Performance, 2005 American Control Conference, 8-10 June 2005

Erm, T. & Sandrock, S. Adaptive periodic error correction for the VLT telescopes, Proc. SPIE Large Ground-Based Telescopes, Vol. 4837-106, 2002

Ulich, B., Montgomery, J., Poyner, A. & Janes, C. Correcting Periodic Error in Telescope Absolute Encoders, MMTO Tech Memo 84-17, 11 June 1984.

Encoder Tests, Gemini Mount Control System Report, MCSJDW11 (Issue 1), 10 July 1996

“Determination and correction of quadrature fringe measurement errors in interferometers” Peter L. M.
Heydemann, APPLIED OPTICS Vol. 20 No. 19, 1 October 1981.

“William Herschel Telescope Inductosyn Tape Encoder Project” C. S. Amos, J. E. Churchill, M. Fisher, R. A. Laing. September 1992. RGO Report - RGO-N-008.

“A High Resolution Incremental Tape Encoder On The William Herschel Telescope” Martin Fisher, ADVANCED TECHNOLOGY OPTICAL TELESCOPES V, SPIE Vol. 2199, March 1994.

“Comments on Report RGO-N-008 - William Herschel Telescope Project” Frank Ruhle, Inductosyn International. EN7032. January 1996.

“Encoder Considerations” Martin Fisher and John Wilkes, GEMINI MOUNT CONTROL SYSTEM REPORT, November 1995.


Tan, K.K., Zhou, H.X., & Lee, T.H. New Interpolation Method for Quadrature Encoder Signals, IEEE Transactions on Instrumentation and Measurement, Vol. 51, No. 5, October 2002

Kavanagh, R.C. Probabilistic Learning Technique for Improved Accuracy of Sinusoidal Encoders, IEEE Transactions on Industrial Electronics, Vol. 48, No. 3, June 2001


Cheung, N.C. An Innovative Method to Increase the Resolution of Optical Encoders in Motion Servo Systems, IEEE 1999 International Conference on Power Electronics and Drive Systems, PEDS '99, July 1999, Hong Kong

Nanowave, Inc. Using Jump Error to Assess the True Performance of Position Encoders, White Paper #4, 2012


Lee, W.Y. Implementation of a DSP-based 20-bit High Resolution Quasi-Absolute Encoder for Low-Speed Feedback Control Applications, Avago Technologies White Paper.


Texas Instruments, TMS320F240 DSP-Solution for High-Resolution Position with Sin/Cos Encoders, Application Report SPRA496

EXTRACTION OF HIGH RESOLUTION POSITION INFORMATION FROM SINUSOIDAL ENCODERS, J. Burke, J. F. Moynihan, K. Unterkofler

Ultra Precise Position Estimation of Servomotor using Analog Quadrature Encoder, Ju-Chan Kim*, Seon-Hwan Hwang*, Jang-Mok Kim†, Cheul-U Kim* and Cheol Choi**, JPE 6-2-6


Post Extras: Print Post   Remind Me!   Notify Moderator  
Mert
Post Laureate
*****

Reged: 08/31/05

Loc: Spain, Pamplona
Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5915253 - 06/11/13 03:18 PM

Really interesting Orlando, I'd be more then curious
to know the runout you can measure with a feeler gauge
when rotating the RA axis and measuring on the new axis
where the encoder will be mounted.

If you could mount the encoder on your aluminum end-plate
with three screws, maybe you could align it better??
Just an idea...


Post Extras: Print Post   Remind Me!   Notify Moderator  
SWR
member


Reged: 07/23/10

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5915563 - 06/11/13 05:58 PM

Quote:

This however requires that you have an additional, absolute indexer - a normal PEC indexer is a single slot, but a worm rotates once every 4-10 minutes. Since the RA shaft only rotates once in 24 hours, you can't determine the PEC index with a single slot: you need an absolute encoder.


It is easy to make an absolute encoder with an extra track besides the quadrature signal. Instead of just a single pulse in one position, you can make a coded signal that doesn't repeat anywhere on the circle. This way you can determine the absolute position after a small fraction of a revolution.
Quote:

You can't calibrate the disk by spinning the RA shaft at a fixed speed. Because how can you ensure a fixed speed when any motor or geartrain you'll use to spin the axis would also have its own periodic errors.


How about using a weak motor and a large flywheel on the axis? This way the rotational speed will be stabilised because the errors of the motor doesn't have enough force to move the large inertia within a single revolution. The downside is that the motor will need some time to stabilize the speed, because the rotating mass will be very slow to bring up to speed, but it should be quite accurate.
Quote:

Many commercial encoder manufacturers calibrate their higher-end encoders using the concept you've suggested.. but they use a laser interferometer with sub-arcsecond resolution, or a high-resolution master encoder, to provide the "known good" readings. By mapping the "known good" angular readings with the (distorted) output of the encoder under test, they can construct an error map specific to that encoder.


Yes, that is a very good solution. The weak motor/flywheel mass was just to suggest a low cost alternative.
Quote:

The problem is... the error map only takes into account the eccentricity of the encoder "stand alone." Once that encoder is installed, bearing wander, axle swash, etc. will induce its own periodic error which the "canned" map in the encoder can't compensate for. So for the really high-end positioning systems like IC etching machines, they do calibration with an interferometer on the completed system.


Yes, that will be a limitation of this approach, so these errors must be kept to a minimum, but as I see it no system can remove bearing wander and axle swash. Those errors will have to be adjusted to a minimum in the mount assembly.
Quote:

Here are some more encoder references I've coughed up in the course of this project.. (most are available online, but some require IEEE membership and payment)



Thank you for the comprehensive reference list.

Post Extras: Print Post   Remind Me!   Notify Moderator  
orlyandico
Postmaster
*****

Reged: 08/10/09

Loc: Singapore
Re: Encoder-based PE Correction on the cheap new [Re: SWR]
      #5915841 - 06/11/13 09:09 PM

actually there is a solution - the Renishaw ring encoders.

There is a lot of prior art (patents even) and some of the articles I've listed also discuss this. The use of two read-heads on opposite sides of the encoder disk or ring basically eliminates encoder eccentricity, bearing wander, and swash related errors - this is because the periodic error seen by each read head is the complement of the other, and averaging the readings from the two read heads will cancel these slow one-cycle errors. 4 heads are better than 2, but >4 gives no benefit (this is also discussed in the Renishaw article linked above).

Unfortunately multiple read heads are not available on the sanely-priced encoders. You'd have to buy a Renishaw ring and two read heads, and mount them yourself. I strongly suspect this is what AP is doing, which explains the large and bulky axes on the 1100 and 1600 (for accommodating the ring).


Post Extras: Print Post   Remind Me!   Notify Moderator  
SWR
member


Reged: 07/23/10

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5916260 - 06/12/13 02:11 AM

That is indeed a very good solution, but also a very costly one. The idea with an optical disk with high resolution plus PEC was an attempt at suggesting a low cost alternative. One of the few low cost technologies working with sub micron resolutions are the optical drives. The lasers and disks are cheap, because they are mass produced.

Another low cost alternative to an accurate encoder as source for PEC calibration is low frequency guiding. The prerequisite is that there are no high frequency encoder errors, to allow for longer guide exposures that will eliminate atmospheric disturbances.

I agree that the easiest solution is to buy an accurate encoder, but it is an expensive solution.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Starhawk
Space Ranger
*****

Reged: 09/16/08

Loc: Tucson, Arizona
Re: Encoder-based PE Correction on the cheap new [Re: SWR]
      #5917675 - 06/12/13 08:35 PM

The optical disk would have to be accurately mounted. Otherwise, what happens if you jump tracks? So, it isn't quite so easy since the encoder has to be internally consistent. Now, if you could trick a disk burner into printing a perfectly radial set of alignment data, then you could have one which would be able to survive jumping tracks, and then you could conceivably make one out of blu-ray player parts.

On the issue of indexing the PEC of the encoder, I had an idea: Provide multiple unique indexing points. So, for example, you could have one every 45°, and the mount would be assured to run over various ones periodically just in the alignment routine. Notionally, there could be a pinhole in a disk, and then have 8 detectors, so at the moment it runs over any of the detectors, the alignment of the disk is known.

-Rich


Post Extras: Print Post   Remind Me!   Notify Moderator  
orlyandico
Postmaster
*****

Reged: 08/10/09

Loc: Singapore
Re: Encoder-based PE Correction on the cheap new [Re: Starhawk]
      #5918128 - 06/13/13 02:03 AM

Rich, my idea was to put a 128-tic ABSOLUTE encoder on the mount, in parallel with the 5000-ppr incremental.

The 128 absolute counts gives 128 separate PEC index marks. In fact I wonder why this isn't done on regular mounts instead of a homing switch. The 128-ppr absolute encoder is $10. And it's rated for 50,000 rotations. Better than a switch that you must rotate to seek the home position.

But.... I measured the encoder error with the encoder disk at a fixed position (used a digital level) and the error varies. So it's not eccentricity. The SiTechServo folks are helping me with this and I have some ideas...

On the other hand the new Renishaw LM10 encoder can be had cheaply, and the magnetic tape (Bogen brand) is cheap as well. You can get started for only about $150 (tape and encoder, both from ebay). The problem is, the accuracy is low (about 300K digital tics over an entire revolution).


Post Extras: Print Post   Remind Me!   Notify Moderator  
vdb
sage


Reged: 12/08/09

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5918170 - 06/13/13 03:01 AM

The problem is, the accuracy is low (about 300K digital tics over an entire revolution).
But if you use the sitech, one can use the geared motor tics to do interpolation ... hmmmm


Post Extras: Print Post   Remind Me!   Notify Moderator  
orlyandico
Postmaster
*****

Reged: 08/10/09

Loc: Singapore
Re: Encoder-based PE Correction on the cheap new [Re: vdb]
      #5918179 - 06/13/13 03:24 AM

At this point interpolation doesn't matter. I am only using the gross quadrature ticks (20K ticks per revolution) and even there I am seeing drift. Until I can fix that, the fine angle doesn't matter.

Also, the TDM works well as everyone knows, and they are also using a 5000-ppr analog sin/cos encoder, and they don't have servo tics available either.

These 5000-ppr analog sin/cos encoders are about +/- 10" accuracy, but somehow the TDM manages to obtain +/- 1" out of it, so it's definitely possible. I imagine reading the encoder at high speed and Kalman-filtering the output is an approach.

I have tried this, but at the end of the day the gross drift is still stopping me. I have to fix that first before I even need to worry about the fine accuracy.


Post Extras: Print Post   Remind Me!   Notify Moderator  
SWR
member


Reged: 07/23/10

Re: Encoder-based PE Correction on the cheap new [Re: Starhawk]
      #5918228 - 06/13/13 05:35 AM

Quote:

The optical disk would have to be accurately mounted. Otherwise, what happens if you jump tracks? So, it isn't quite so easy since the encoder has to be internally consistent. Now, if you could trick a disk burner into printing a perfectly radial set of alignment data, then you could have one which would be able to survive jumping tracks, and then you could conceivably make one out of blu-ray player parts.


How about first securely fixing an empty writable disk and the reader PCB on the mount, and then burn the track signals with the same fixed lasers that are used for reading the signals? This will remove all alignment errors that are not caused by temperature drift and bearing tolerances. The lasers doesn't have to move because they only have to burn and read a single track.
Quote:

On the issue of indexing the PEC of the encoder, I had an idea: Provide multiple unique indexing points. So, for example, you could have one every 45°, and the mount would be assured to run over various ones periodically just in the alignment routine. Notionally, there could be a pinhole in a disk, and then have 8 detectors, so at the moment it runs over any of the detectors, the alignment of the disk is known.


An even better way is to have a non repeating code all the way round on the index track. This way you can tell the absolute position when you have moved a few counts in either direction.

Post Extras: Print Post   Remind Me!   Notify Moderator  
Pages: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | (show all)


Extra information
22 registered and 36 anonymous users are browsing this forum.

Moderator:  Dave M, richard7, bilgebay 

Print Thread

Forum Permissions
      You cannot start new topics
      You cannot reply to topics
      HTML is disabled
      UBBCode is enabled


Thread views: 14748

Jump to

CN Forums Home


Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics