Jump to content


Photo

Encoder-based PE Correction on the cheap

  • Please log in to reply
306 replies to this topic

#126 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 02 April 2013 - 10:04 AM

I think $500 is optimistic. :tonofbricks: It can be done for $500 if you assemble everything yourself (as the encoder is $325).

If it's a partial kit or something, someone would have to do the integration and that drives up cost because labor ain't free. :)

#127 Alph

Alph

    Surveyor 1

  • -----
  • Posts: 1769
  • Joined: 23 Nov 2006
  • Loc: Melmac

Posted 02 April 2013 - 12:17 PM

I don't think people would pay $500 to take a CGEM to +/- 4" PE. But they might if it was +/- 2" or better.



I don’t view this project as limited to the CGEM only. I personally would be interested in a simple device that could be used to accurately measure PE of a mount in-house. To limit the cost and the hassle of building electronics board, the device should be operated/controlled from PC. What would it take to make the output from an encoder readable by PC? The data feed from an encoder could be consumed e.g by Pempro. That would be a very nice product.

#128 Starhawk

Starhawk

    Space Ranger

  • *****
  • Posts: 5614
  • Joined: 16 Sep 2008
  • Loc: Tucson, Arizona

Posted 02 April 2013 - 12:56 PM

It would take an electronics box if you want to be inexpensive.

Otherwise, there are evaluation board setups which use software like Labview to use a PC. So, besides losing portability, that approach would also be quite expensive and entail a new software suite.

The good news is that stuff is all off the shelf, now- just go buy it.

-Rich

#129 Mert

Mert

    Skylab

  • *****
  • Posts: 4231
  • Joined: 31 Aug 2005
  • Loc: Spain, Pamplona

Posted 02 April 2013 - 01:26 PM

I have another question for you Orlando:

On the drift item you have demonstrated, did you check
if the sum of all the corrections done over 1 worm cycle
adds up to 0 aprox??? :question:

#130 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 02 April 2013 - 11:21 PM

Alph, the box as it is today can measure PE when attached to a PC via USB.

Now as we've seen, there is a good but not perfect correspondence between the PE measured by the box, and by PEMPro. I would not consider PEMPro to be the "reference" because it is affected by seeing, while the box is not.

I moved the whole box over to the PICMX32, since I will need more I/O pins going forward and the PICMX costs about the same as an Arduino Mega2560. Discovered that MPIDE and the Arduino IDE aren't 100% identical, and had to re-write some code.

Hopefully this will solve the timing issue (the Max32 has a TCXO) and the 128K of RAM will allow me to do the statistical correction in Kavanagh's article. The only challenge is I have introduced a bug in the code... :tonofbricks: the ADC reads out properly during calibration but in the main loop outputs -1.....

Mert, yes I did check and the average error over 1 cycle was close to zero (it was 0.05").

#131 Pinbout

Pinbout

    Cosmos

  • *****
  • Posts: 8291
  • Joined: 22 Feb 2010
  • Loc: You can't see me...

Posted 02 April 2013 - 11:36 PM

hi Chris,

No I haven't tried taking a picture.... but I kind of determined that I want perfection. Besides this would cost $500. I don't think people would pay $500 to take a CGEM to +/- 4" PE. But they might if it was +/- 2" or better. :D


Oh yes they would and I suspect that the well tuned mount might do better in the first place since it would be starting with lower and smoother PE and random error. If it can be done for $500, that would be great.


that's cheaper than Explorer Scientific's high end encoder TDM $1799.

iOptron suppose to have a IEQ45 gt version that's a lot cheaper than their first version which with a 3rd encoder that made the mount cost $5k

#132 Alph

Alph

    Surveyor 1

  • -----
  • Posts: 1769
  • Joined: 23 Nov 2006
  • Loc: Melmac

Posted 02 April 2013 - 11:57 PM

Alph, the box as it is today can measure PE when attached to a PC via USB.


Yes, I am aware of it. Let me rephrase my question. What are the bare minimum hardware requirements to get the output from the encoder to the PC? I see a lot of potential in having just that.

Now as we've seen, there is a good but not perfect correspondence between the PE measured by the box, and by PEMPro. I would not consider PEMPro to be the "reference" because it is affected by seeing, while the box is not.



Sure. However, you still need to sync up with the worm index and upload PEC to the mount and that’s where PEMPro comes in.

#133 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 03 April 2013 - 01:09 AM

The bare minimum is what I have right now:

1) an Arduino of some sort (an Uno will work fine if reading the PEC is all that's required) - this is about $30, and connects to the PC via USB

2) an Arduino daughterboard with an AD converter on it (total cost $10)

3) the encoder itself, that needs to be wired up to the Arduino

The vast majority of the cost would be fabricating the mount-specific adapter, and the encoder itself.

As for sync'ing the worm index and uploading PEC... that's a job best left to PEMPro or something. It might be possible to produce a PE file from the encoder that PEMPro can understand.

As for the TDM.. it's $1800 but you need to budget another $400 for the mount-specific adapter.

I have a whole lot more respect for the Germans who built that thing now (you know the old saying.. the ignorant sometimes attempt really difficult problems because they don't know how hard the problem is.. well that was me). I still think the price can be less.

The last iEQ45 I checked that had a Renishaw encoder was $4800. Which is idiotic. Nobody will buy a $4800 China mount with 1/2 the capacity of a Mach1 when you can get a Mach1 used for a few hundred bucks more.

If there was a $2K encoder-equipped mount with the capacity of an iEQ45 or Atlas/CGEM, then that would fly off the shelves.

Avago has a cheap 20-bit (1 million tick) digital encoder (this one - http://www.avagotech...cs/AV02-0865EN), that if combined with something like the SiTech tick management (possible if the RA encoder is integrated with the mount drive electronics) would enable this price point.

#134 vdb

vdb

    Messenger

  • -----
  • Posts: 432
  • Joined: 08 Dec 2009

Posted 03 April 2013 - 03:44 AM

Problem is that Renishaw encoder is already around 1K so cheapest I can see for such a mount is around 3k ...

#135 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 03 April 2013 - 03:46 AM

The iEQ45 in pinbout's link is a Renishaw RGH22 (or similar) linear read head. These have built-in interpolation and output a digital signal. They are cheaper than rotary encoders and you don't need as high a resolution because you would put an encoder tape around the circumference of the RA housing, which is a much larger radius, so the grating pitch on the tape can be less. I believe the tape is $12/inch or something.

the Gurley 320K tick encoders used on SiTech are about $600 and have built-in interpolation. The resolution is not high enough by itself, but in conjunction with tic management using the motor encoders, SiTech does a very good job. I cannot see how the iOptron guys could not use a similar approach..

(and.. adding a $2200 TDM to a base iEQ45 results in $3700 - which is still well over $1000 less than the Renishaw-equipped iEQ45)

#136 freestar8n

freestar8n

    Skylab

  • *****
  • Posts: 4062
  • Joined: 12 Oct 2007

Posted 03 April 2013 - 04:16 AM

What are the bare minimum hardware requirements to get the output from the encoder to the PC? I see a lot of potential in having just that.



Years ago in my P.E. days, I was involved in other ways to measure PE. Why do you think it's so valuable or desirable to measure PE without a star? Are there others that share your interest?

I suppose for people making mechanical adjustments to their mounts it would be helpful to have prompt feedback, or when iterating to find an optimal PEC curve to apply.

Frank

#137 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 03 April 2013 - 06:04 AM

Frank, there was a long discussion of this before.

Basically, having a mechanical method of measuring PE would allow a mount manufacturer to do QA cheaply. They would be able to ensure that no stinker mounts got out the door, and perhaps charge a premium for a "certified" mount. Surely a certification would be worth another $100 to $200?

But if you have to use a star to measure the PE, the labor alone of the guy doing the measuring will wipe out your profit.

#138 freestar8n

freestar8n

    Skylab

  • *****
  • Posts: 4062
  • Joined: 12 Oct 2007

Posted 03 April 2013 - 06:10 AM

That makes sense - but would only apply to a mount manufacturer... And presumably they can afford an expensive and high-res encoder in the first place since they would use it in production.

My own work involved other methods for measuring PE - but I lost interest. If there is a general need or desire for such things then I can take a look.

Frank

#139 vdb

vdb

    Messenger

  • -----
  • Posts: 432
  • Joined: 08 Dec 2009

Posted 03 April 2013 - 06:17 AM

(and.. adding a $2200 TDM to a base iEQ45 results in $3700 - which is still well over $1000 less than the Renishaw-equipped iEQ45)


But you loose the polarscope ... which for a portable setup is nice, especially the iOptron as it's very accurate, I could do 30 min subs without field rotation.

#140 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 03 April 2013 - 06:19 AM

One method I thought that a manufacturer could use to measure PE.. (not useful for astronomers)

Rigidly affix a laser pointer to the mount, so that the dot shines on a curved surface (like the inside of a dome, or a curved piece of white-painted board) some distance away.

Run the mount while taking a movie of the laser dot. Then post-process each frame of the movie to quantify the laser dot's movement. A perfect mount would show the dot moving at a fixed rate, any variation in the dot speed is PE.

QED. No need for a fancy encoder.

#141 Pinbout

Pinbout

    Cosmos

  • *****
  • Posts: 8291
  • Joined: 22 Feb 2010
  • Loc: You can't see me...

Posted 03 April 2013 - 06:45 AM

(and.. adding a $2200 TDM to a base iEQ45 results in $3700 - which is still well over $1000 less than the Renishaw-equipped iEQ45)



last year when I was speaking to the same guy at NEAF [i think he's the president], he said they're suppose to come out with a gt this year thats a lot cheaper. They're gonna make their own encoder, what kind I don't know but I'm sure gonna bug him about it this year at NEAF so do you have any specific question I should ask him that's not too over my head? :grin:

#142 cn register 5

cn register 5

    Viking 1

  • -----
  • Posts: 760
  • Joined: 26 Dec 2012

Posted 03 April 2013 - 07:28 AM

There's a program called Star Guiding Emulator that puts a dot on a PC screen and moves it at the correct rate. You can use it to train PEC and practice imaging indoors.

http://sweiller.free...ng-emulator.htm

Chris

#143 Starhawk

Starhawk

    Space Ranger

  • *****
  • Posts: 5614
  • Joined: 16 Sep 2008
  • Loc: Tucson, Arizona

Posted 03 April 2013 - 08:48 AM

One of the biggest benefits of Orly's encoder system is it fixes a pec problem peculiar to the CGEM- the 8/3 resonance in the drive motors whic makes it impossible to get a true pec loop in one worm rotation.

The other is it's a bolt on and forget setup- no looking for a guide star and training. Just turn it on.

So, for those of us who have to flee city lights, this makes a lot of sense.

-Rich

#144 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 03 April 2013 - 09:01 AM

Well I'm kinda stuck right now, so... :tonofbricks:

#145 Alph

Alph

    Surveyor 1

  • -----
  • Posts: 1769
  • Joined: 23 Nov 2006
  • Loc: Melmac

Posted 03 April 2013 - 10:49 AM

I have a whole lot more respect for the Germans who built that thing now (you know the old saying.. the ignorant sometimes attempt really difficult problems because they don't know how hard the problem is.. well that was me).



The Telescope Drive Master, or TDM, is the product of Hungarian amateur astronomers, Istvan Papp & Attila Madai of MDA Telescoop and it is marketed and sold by Meade Europe.

#146 Alph

Alph

    Surveyor 1

  • -----
  • Posts: 1769
  • Joined: 23 Nov 2006
  • Loc: Melmac

Posted 03 April 2013 - 11:44 AM

but would only apply to a mount manufacturer...



Many mount owners would like to know if their mounts perform to specifications. I have noticed that many folks struggle with periodic error measurements and analysis, and the results are quite often inconclusive. Flexure, seeing, and other unknown factors obscure the real PE graph. Other mounts like Paramount ME/MX have over 1000 PEC table entries. The PMX control board issues PEC corrections every 0.25s. You can't measure PE with such a precision just by tracking a star.

#147 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5720
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 03 April 2013 - 12:53 PM

I have also wondered why the AP GTO CP3 has over 1000 PEC table entries.

When you look at the way PEMPro programs the PEC on a GTO CP3, there is a "minimum move" command of 10ms. Because the error is so low to begin with, and the minimum move causes a 10ms correction, probably only one in 20 PEC entries has something in it.

Oh, and I stand corrected about the TDM. I always thought they were Germans...

On a related note, I finally figured out what was causing the weird ADC readings on my box. It wasn't bad code or running out of memory - so maybe moving to the PIC32 was not such a great move - but noise with my rat's nest wiring. This never popped up when I was bit-banging the SPI because the speed was low, but now that I'm clocking the ADC at 2MHz it is an issue.

I rewired (and soldered) everything onto an Arduino shield and the ADC has settled down; it still (very rarely) glitches but I can work around that in code. The PIC32 is very fast - and actually has an onboard 1 Msps 8- or is it 16-channel ADC. So in theory one could implement the PEC reader with a PIC32 alone. But the ADC inputs aren't differential, and I don't feel like mucking around with op amps..

I also found that the Lissajous pattern of the encoder is almost perfectly circular (0.9995), that there is zero offset on both A and B channels, and 99.7% of the radius readings (3 sigma) fall within a +/- 5% range.

So on startup I take several thousand readings (two slots worth @ 17 seconds/slot) and calculate the average radius. From this I can set a low (0.95X) and high (1.05X) bound for acceptable radius, which makes it trivial to throw out bad data.

I haven't implemented the algorithm described by Kavanagh yet, but now that the ADC issues have been sorted out that's next on my list.

#148 tjugo

tjugo

    Apollo

  • -----
  • Posts: 1013
  • Joined: 06 Nov 2007

Posted 03 April 2013 - 02:16 PM

Hi Orly,

Keep it going! This is a nice project, I hope you get to your goals 2 arcsec p-p under US$500. This will be a very nice addition to EQMOD.

Cheers,

Jose

#149 vdb

vdb

    Messenger

  • -----
  • Posts: 432
  • Joined: 08 Dec 2009

Posted 03 April 2013 - 02:44 PM

last year when I was speaking to the same guy at NEAF [i think he's the president], he said they're suppose to come out with a gt this year thats a lot cheaper. They're gonna make their own encoder, what kind I don't know but I'm sure gonna bug him about it this year at NEAF so do you have any specific question I should ask him that's not too over my head? :grin:


Just ask how much it costs and when it ships

#150 OzAndrewJ

OzAndrewJ

    Apollo

  • *****
  • Posts: 1034
  • Joined: 30 Nov 2010

Posted 03 April 2013 - 02:58 PM

Gday Orlando

Rigidly affix a laser pointer to the mount, so that the dot shines on a curved surface (like the inside of a dome, or a curved piece of white-painted board) some distance away.

Run the mount while taking a movie of the laser dot.



I tried this once, but i used a tape measure on a fence over the road
and took webcam snaps every second.
Didnt work well, so i did some numbers.
One arcsec error at 100metres is approx 0.5mm,
and my fence was only 30m away.
Considering tracking is at 15 arcsec/sec and you also have
camera timing issues etc, you have no hope of getting
any real accuracy with this mechanism.
( with a laser, you also have the problem of getting a correct centroid )

Andrew Johansen Melbourne Australia






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics