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)
galaxy_jason
Vendor


Reged: 05/22/07

Loc: Texas
Re: Encoder-based PE Correction on the cheap new [Re: Stew57]
      #5760678 - 03/27/13 04:00 PM

Quote:

The PEC on the CGEM can not take out the infamous 8/3 error as it is non integer. Either a 3 worm rotation PEC would have to be instituted (not a 3 rotation average) or new motors. For some mounts the non integer componet is larger than the first harmonic.





Skip the encoder. If you can use the existing, or install your own worm position sensor that's all you need. Setup PEMPro to capture however many worm rotations are needed. Save the PEMPro data then write Arduino code to play back the curve .

I did something similar back in 1990 with an LX5 before PEC was available. I used a wiper and a piece of wire attached to the worm. I cut into the clock line and ran one of three different counters into the motor drive circuit depending on the position of the worm measured photographically. Essentially a three "bin" PEC. It cut the PE from 60" down to about 15". With a micro processor you could do the same (through the autoguider port or with ASCOM) with a much finer resolution.

With non-integer ratios you may need to have a second sensor on a reduction gear to tell the Arduino what cycle you are on. Would have to be worked out somehow but since the title of the thread is "on the cheap", software and a couple sensors are cheaper than a high-res encoder.


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

Reged: 08/27/07

Loc: Deep in the heart of Texas
Re: Encoder-based PE Correction on the cheap new [Re: galaxy_jason]
      #5760998 - 03/27/13 06:25 PM

Skywatcher just announced that their API is now open source. I wonder if the Atlas/EQ-G mount has enough memory to incorporate this software into the mount controller? Is it necessary to use the autoguider port or can it use autoguider commands as if it were coming from a computer?

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: StarmanDan]
      #5761245 - 03/27/13 08:31 PM

1) Jason - I believe the Celestron folks are trying to figure out how to do the 3-cycle PEC with the existing infrastructure. Since the CGEM only has a single-cycle worm indexer, this is a bit.. challenging. Also the CGEM controller only has an 88-cell PEC, so correcting for fast errors is out. Adding a 3-cycle sensor isn't something I can think of doing elegantly.. there's not a lot of space inside the mount.

In any case - if getting "good" PEC was all I was after, I'd just wait for the CGEM AVX (or retrofit motors) which has all-integer ratios and would get rid of the 8/3. My aim is a bit more than just getting the PEC to work.

2) Dan - I believe no. The Atlas motor controller uses some fairly primitive PIC microcontrollers. I'm fairly certain they don't have the smarts to calculate arc-tangents. The Arduino may be cheap and cheesy, but it's a 16MHz CPU.


Post Extras: Print Post   Remind Me!   Notify Moderator  
OzAndrewJ
scholastic sledgehammer
*****

Reged: 11/30/10

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5761319 - 03/27/13 09:07 PM

Quote:

Adding a 3-cycle sensor isn't something I can think of doing elegantly.. there's not a lot of space inside the mount.




Just do what Meade did for the RCX400, and later the LX200s
( assuming it isnt patented )
Very elegant, and the worse the 8/3 error, the better it works

Andrew Johansen Melbourne Australia


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: OzAndrewJ]
      #5761329 - 03/27/13 09:14 PM

Andrew, I'm not familiar with that approach?

Post Extras: Print Post   Remind Me!   Notify Moderator  
OzAndrewJ
scholastic sledgehammer
*****

Reged: 11/30/10

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5761480 - 03/27/13 10:49 PM

It was discussed in great detail when it first came out, but its basically a simple patternmatch, and you use the PE "inside the gearbox" to do it.
You have a sensor on the worm itself and an encoder on the motor.
One rev of the motor output shaft "should" be a set no of encoder ticks.
If the gearbox has the 8/3 error, what you do know is after 3 revs of the worm, everything is back synchronised, but inbetween, it can ( and does ) vary.
Thus you count encoder transitions between each firing of the sensor on the worm, and repeat for at least three full firings of the sensor.
Now because there IS PE in the gearbox, we know the encoder count per sensor trip will vary over those three turns, but the sum of the three should equal the perfectcount * 3.
Ie You will end up with 3 numbers that can be validated for consistency.
We dont really care what they are,
as when you next start the scope, you do the spin again and read the values.
Then find which position gives the best patternmatch to the master stored prior to doing the PEC model.
Its fully self correcting, but it does rely on a relatively repeatable sensor.

Andrew Johansen Melbourne Australia


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: OzAndrewJ]
      #5761598 - 03/28/13 12:24 AM

The goal here isn't just to overcome the 8/3 resonance- the idea is to true up the output tracking rate with the encoder. This was to get much better performance from the mount- as in an order of magnitude better than the stock mount.

-Rich


Post Extras: Print Post   Remind Me!   Notify Moderator  
OzAndrewJ
scholastic sledgehammer
*****

Reged: 11/30/10

Re: Encoder-based PE Correction on the cheap new [Re: Starhawk]
      #5761629 - 03/28/13 01:08 AM

Gday Rich

Quote:

The goal here isn't just to overcome the 8/3 resonance-




Fully understood.
My comment was merely a reply to the sidethread re why doing an "offmount" PEC model inside the Arduino should or should not be considered.

Andrew Johansen Melbourne Australia


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


Reged: 12/08/09

Re: Encoder-based PE Correction on the cheap new [Re: Raginar]
      #5761923 - 03/28/13 08:47 AM

RESPECT!

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]
      #5762575 - 03/28/13 03:13 PM

Andrew, thinking about your solution... it cannot be done with a bolt-on, because the internal motor encoder on the Celestron is already used by the Nexstar hand controller. Short of ripping out the Nexstar and implementing mount modeling, GoTo, etc... not worth doing.

Celestron could do this fix, since they have control of the motor board firmware, but the fact they haven't... maybe Meade does have a patent on that approach.

What I want to achieve is a dumb bolt-on solution that will work on any darn mount, even a non-GoTo G11 or Atlas. So long as it has an ST-4 port.


Post Extras: Print Post   Remind Me!   Notify Moderator  
OzAndrewJ
scholastic sledgehammer
*****

Reged: 11/30/10

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5762707 - 03/28/13 04:52 PM

Quote:

it cannot be done with a bolt-on, because the internal motor encoder on the Celestron is already used by the Nexstar hand controller




Understood,
but it could be done easily enough if you have the ability to tap into the encoder lines between the motor and motor card.
You only need to read them whilst doing the synch process.
That said, your current "guide only" process bolts to the output axle, so even the much simpler one turn PEC synchronisation would require a sensor on the worm itself, hence is not a simple "bolt on" for all mounts.

Andrew Johansen Melbourne Australia


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: OzAndrewJ]
      #5762720 - 03/28/13 04:59 PM

Actually the CGEM does have an indexer for the PEC. But again, attempting to tap into the indexer and motor encoder is much more involved than bolting something to the polar scope bore and plugging an RJ-12 into the ST-4 socket.

My feeling is (given Celestron's release of the AVX) they have decided to just replace the gearbox, and not try to fix the existing 8/3-plagued one. There is some talk that Celestron will sell AVX type motors and a motor board for us poor unfortunate souls with CGEMs. I estimate it will cost upward of $300 to do this retrofit. For about the same price I can buy a new Heidenhain encoder and get to have some fun developing something while it's cloudy (which is very often).

The bottom line is, there will always be random or periodic but fast gear noises on every rotation (on non-premium mounts), and PEC can't get rid of that, whereas an external encoder correcting fast enough (say 5-10 times per second) has a better shot.


Post Extras: Print Post   Remind Me!   Notify Moderator  
OzAndrewJ
scholastic sledgehammer
*****

Reged: 11/30/10

Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5762887 - 03/28/13 06:23 PM

Quote:

For about the same price I can buy a new Heidenhain encoder and get to have some fun developing something while it's cloudy




Just dont forget the golden rule here.
Unless it takes more than 10 times longer and costs more than 5 times as much as an off the the shelf item, its not worth doing

Andrew Johansen Melbourne Australia

Hmm, if a motor / Gbx could be found that reset itself properly every rev of the output shaft, and could take an encoder, there would also be a retrofit market for the Meade LX200s as well


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: OzAndrewJ]
      #5763199 - 03/28/13 09:13 PM

Dunno if 10X as long is available in this case. He's already built something...

-Rich


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

Reged: 08/27/07

Loc: Deep in the heart of Texas
Re: Encoder-based PE Correction on the cheap new [Re: Starhawk]
      #5763225 - 03/28/13 09:29 PM

Another question: Why use an Arduino? Couldn't a computer perform the same task? Some of these encoders already incorporate a serial output.

Post Extras: Print Post   Remind Me!   Notify Moderator  
Jim McGee
member


Reged: 04/21/07

Re: Encoder-based PE Correction on the cheap new [Re: StarmanDan]
      #5763270 - 03/28/13 09:59 PM

First, I think this is a great idea. I'm looking forward to seeing the results of testing with actual on-the-fly adjustments sent to the mount.

Second, does anyone know where I can actually purchase an appropriate encoder? I'm in the US and I have a CGE that desperately needs something like this. My web searches have not turned up anything that shows an actual price of less than $1000.

Edited by Jim McGee (03/29/13 12:21 AM)


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: Jim McGee]
      #5763528 - 03/29/13 01:43 AM

Dan - you are right, a PC can do it. I'm not aware of encoders with PC-style serial ports, but the Gurley 320K tick interpolating encoder that's used by many SiTech users is over $600. And it's a Russian encoder. I think the extra cost is due to the interpolation electronics.

My whole scheme (scam?) is to make the interpolation electronics as cheap as possible. If the electronics can also do the mount corrections then that's just icing.

Jim, as I've stated above, I contacted the Heidenhain office where I live. You could try these guys..

http://www.heidenhain.us/contact.php

The encoder I'm planning to use going forward is ERN 480 x 5000 ID: 385480-17

Although of course if you want really cheap, go scrounge on eBay. Make sure the encoder has sine-cosine output, and that you can get a data sheet for it (so you know how to wire it up).


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]
      #5763707 - 03/29/13 06:13 AM

Tentatively, success has been achieved.

Note: the periodic error displayed here is from the encoder, not from the sky! hence I don't have PEMPro data yet, it's raining hard (and still daylight).

Hopefully within the week I'll be able to get PEMPro data.

I did not quite achieve 1" peak-to-peak. More like 2" peak-to-peak. This is with corrections every 0.5 second. Seems the CGEM is not responsive to guide pulses much shorter than 100ms.



Notice the large error (first graph) that suddenly goes to zero. That's when I plugged the ST-4 cable into the CGEM's autoguider port.

The second graph is the tracking error with ST-4 correction enabled. It is about 4" peak-to-peak (the worst excursions) although majority of the error is within 2" peak-to-peak.

And the RMS error is 0.54" (the numbers on the graph are deci arcseconds (i.e. 10 = 1") because I'm using floating point as little as possible.

The only difference between today's update and last week's, is I have added the code to calculate the tracking error and control the ST-4 output. Needless to say I have also wired an optocoupler to the Arduino to talk to the CGEM.

There is no capability to accept ST-4 input yet (I will need to wire a jack to the Arduino, and I don't have such a jack).

Raw putty data from the Arduino is here - https://dl.dropbox.com/u/63497702/encoderdata-2ch-withST4.log

the columns are:
time (milliseconds)
channel A ADC count
channel B ADC count
calculated theta (interpolated value, from 0.. 2592)
slot number (0.. 5000)
raw encoder angle
measured angle
expected angle
error
required pulse width
RA+ pulse width
RA- pulse width

all angles are in deci-arcseconds.

The Excel spreadsheet above - https://dl.dropbox.com/u/63497702/encoderdata-2ch-withST4.xls

and the source code - https://dl.dropbox.com/u/63497702/read_encoder.7z

Note that at this time, there are a lot of hard-coded stuff

1) it assumes a 5000-tick encoder
2) it assumes you are in the northern hemisphere
3) it assumes your ST-4 is wired correctly (some ST-4 jacks reverse the RA+ and RA-) - in my first attempt, the errors got bigger and bigger as the Arduino tried to correct, so I just swapped the RA+ and RA- wires.

Obviously in a mass-market build, things like these should be automatic, or there should be a switch and a red LED to tell you things are screwing up. Right now there is nothing, only the serial port output to tell you how things are going.


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]
      #5763746 - 03/29/13 07:21 AM

By increasing the correction rate to 200ms (5X per second, instead of 2X) and decreasing the tolerable error from 0.6" to 0.3" I have been able to reduce the peak-to-peak PE to about 1.2" and the RMS PE to 0.23"



That's at the encoder.. still have to update with what PEMPro says.


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

Reged: 10/19/10

Loc: Rapid CIty, SD
Re: Encoder-based PE Correction on the cheap new [Re: orlyandico]
      #5763947 - 03/29/13 09:15 AM

Wow, cool Orly .

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
29 registered and 46 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: 14602

Jump to

CN Forums Home


Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics