Jump to content


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Direct drive motor

mount
  • Please log in to reply
48 replies to this topic

#26 James74

James74

    Vostok 1

  • -----
  • Posts: 119
  • Joined: 29 Jan 2012
  • Loc: Sacramento, CA

Posted 28 December 2014 - 10:23 AM

I think it would work fine with a powertank, as long as the scope is well balanced and you don't rapid slew everywhere. I worked out the math and 3-5A would work for an 8 or 10 inch scope. When you are tracking at sidereal rate the power draw should be pretty small.

Edited by James74, 28 December 2014 - 10:25 AM.


#27 don clement

don clement

    Vendor (Clement Focuser)

  • -----
  • Vendors
  • Posts: 821
  • Joined: 02 Feb 2011
  • Loc: Running Springs, California

Posted 28 December 2014 - 05:23 PM

I think it would work fine with a powertank, as long as the scope is well balanced and you don't rapid slew everywhere. I worked out the math and 3-5A would work for an 8 or 10 inch scope. When you are tracking at sidereal rate the power draw should be pretty small.

 

 

3-5A @ what voltage?

 

Don C.



#28 Oberon

Oberon

    Messenger

  • -----
  • Posts: 445
  • Joined: 24 Feb 2013
  • Loc: Gunnedah NSW Australia

Posted 28 December 2014 - 07:10 PM

High current means high levels of heat dissipation means hot air means poor seeing means lousy telescope.



#29 steveastrouk

steveastrouk

    Ranger 4

  • -----
  • Posts: 332
  • Joined: 01 Aug 2013
  • Loc: State College, Pa.

Posted 29 December 2014 - 12:20 PM

High current means high levels of heat dissipation means hot air means poor seeing means lousy telescope.

 

Only if a.) Its high current all the time and b.) its high resistance. Unlike stepper motors, these motors only draw what they need as they need it.


Edited by steveastrouk, 29 December 2014 - 03:07 PM.


#30 James74

James74

    Vostok 1

  • -----
  • Posts: 119
  • Joined: 29 Jan 2012
  • Loc: Sacramento, CA

Posted 29 December 2014 - 05:34 PM

The MOSFETs won't put off enough heat to effect seeing at all. Especially at a max of 5A.

I think 12V would be a good voltage, since you will only find larger batteries and powertanks in 12V. Voltage shouldn't matter for the motor coils since current is what generates torque.

#31 KevinS

KevinS

    Mariner 2

  • -----
  • Posts: 254
  • Joined: 27 Dec 2011

Posted 29 December 2014 - 05:38 PM

I recall Dan Gray was working on a direct drive motor. As I recall they were building some small units for university scopes. 

 

Here's a link...

http://siderealtechn...kInProgress.pdf


Edited by KevinS, 29 December 2014 - 05:39 PM.


#32 James74

James74

    Vostok 1

  • -----
  • Posts: 119
  • Joined: 29 Jan 2012
  • Loc: Sacramento, CA

Posted 29 December 2014 - 05:49 PM

Yup, check 4 posts up.

The only problem is the controller costs $3500 and is way overspecified for small scopes, since it was designed for 1m telescopes.

#33 KevinS

KevinS

    Mariner 2

  • -----
  • Posts: 254
  • Joined: 27 Dec 2011

Posted 29 December 2014 - 06:11 PM

Guess I missed that post sorry to repeat.



#34 Oh Happy Nights

Oh Happy Nights

    Lift Off

  • -----
  • Posts: 3
  • Joined: 27 Dec 2014
  • Loc: Utah

Posted 30 December 2014 - 01:27 PM

I like the idea of making my own motors which is very doable as KevinS mentioned above-- http://siderealtechn...kInProgress.pdf
http://www.wvi.com/~...8/calpoly18.htm

 

During the past few years, we have seen the inception of the pancake radial flux direct drive washing machine motor.  They have become very obtainable at a very low price on ebay and even free if you can find an old washer.  I am going to pick one of these up to see how much torque I can get out of it.

 

Obtaining a motor is surely within a DIYer’s realm, but I, like many other amateurs, will have a hard time coming to terms with dishing out $6,500 for a controller and encoders (although it may be something that I must eventually do), and I was interested if there were other options.


Edited by Oh Happy Nights, 30 December 2014 - 02:52 PM.


#35 James74

James74

    Vostok 1

  • -----
  • Posts: 119
  • Joined: 29 Jan 2012
  • Loc: Sacramento, CA

Posted 30 December 2014 - 02:28 PM

When you choose a motor, consider the number of poles. You want to have an odd number of poles to avoid cogging.
You can wrap your own coils, then mold resin around them to secure the wire. The Sidereal guys did this and it looks pretty good.

the controller is the expensive part. Is anyone interested in an open source project to create a controller for lighter scopes and field mounts?

#36 brave_ulysses

brave_ulysses

    Mariner 2

  • -----
  • Posts: 208
  • Joined: 19 Apr 2009
  • Loc: far outside the wire

Posted 30 December 2014 - 05:12 PM

hi james,

 

what are you thinking about for the open source controller?

 

mcu?

physical size?

voltage/current?

encoder?

hall sensor?

etc?

 

a couple of other links that might present some ideas:

 

http://www.instructa...-Motor-and-You/

 

https://groups.yahoo...-torquemax/info



#37 James74

James74

    Vostok 1

  • -----
  • Posts: 119
  • Joined: 29 Jan 2012
  • Loc: Sacramento, CA

Posted 30 December 2014 - 08:03 PM

hi james,

 

what are you thinking about for the open source controller?

 

mcu?

physical size?

voltage/current?

encoder?

hall sensor?

etc?

 

a couple of other links that might present some ideas:

 

http://www.instructa...-Motor-and-You/

 

https://groups.yahoo...-torquemax/info

 

MCU - unsure; needs to be fast enough to read encoders during reposition slews and have a frequency above the resonant frequency of most small scope setups. Other than that, I have no idea

Physical size - smaller than a tennis shoebox, 6x4x4" would be optimal for a portable mount, but the small the better, of course. I think the need to keep the MOSFETs cool will drive this one

Voltage/current - 12V is a common voltage for power supplies, so we'll roll with that. 3-5A through the MOSFETs should generate 60-100 in*lb of torque, which should be plenty for a small scope during fast slews

Encoder - Renishaw linear encoder with read tape placed on a disk on the RA axis; not sure for Dec (Renishaw is good, but expensive even if bought used; I want to look into other encoders)

Hall sensor - not needed if you are using high-res encoders

 

There are a ton of other considerations, I'm sure, but I don't know electronics are well as I need to for this project (trying to learn, though). I do know how to program and I know controls pretty well.


Edited by James74, 30 December 2014 - 08:24 PM.


#38 brave_ulysses

brave_ulysses

    Mariner 2

  • -----
  • Posts: 208
  • Joined: 19 Apr 2009
  • Loc: far outside the wire

Posted 30 December 2014 - 09:48 PM

sounds like a good start.

 

 

i like the stm32 family. the timers can be setup for 3 phase pwm output and quadrature encoder input. the f4 discovery boards are cheap and powerful:

http://www.mouser.co...PRSvEN8XDBeCtgD

 

 

there are several online projects with this chip being used for driving a brushless motor

 

 

this project uses the stmf4 chip and has some instructions for setting up a development environment:

https://pixhawk.org/



#39 gregj888

gregj888

    Viking 1

  • -----
  • Posts: 711
  • Joined: 26 Mar 2006
  • Loc: Oregon

Posted 01 January 2015 - 01:44 PM

Hard to get around the encoders, and they cost as much as a couple of 12"- 15" worms.  For a small scope I don't see that direct drive has much advantage over a worm/servo (Scitech1/Sitech2) system at least at the increased cost. 

 

I see an advantage at 20"... cost wise, but just.  Depends a lot on the cost and availability of a quality worm of the right size.



#40 James74

James74

    Vostok 1

  • -----
  • Posts: 119
  • Joined: 29 Jan 2012
  • Loc: Sacramento, CA

Posted 01 January 2015 - 05:28 PM

You are right, encoders are expensive, no two ways about it.
Encoders really are what gives you the ability to point well, which is why the Sidereal servo controllers are so good. That and they probably have good controls software.

Direct drive has zero backlash, effectively zero periodic error, and great responsiveness in windy conditions.
If we can keep the cost below 2k for motors, encoders, and controller we'd have a killer travel mount on our hands.

#41 Oh Happy Nights

Oh Happy Nights

    Lift Off

  • -----
  • Posts: 3
  • Joined: 27 Dec 2014
  • Loc: Utah

Posted 01 January 2015 - 08:44 PM

I would be tripping over myself to get to those motors, encoders, and controller for $2,000.

 

I had a chance to talk to a friend in the automation dept. at work, and he said that the electronics would not be so bad but the programing would be a time consuming nightmare.  Maybe this is why Dan Gray has taken 5 years to get the ST controller on the market.



#42 brave_ulysses

brave_ulysses

    Mariner 2

  • -----
  • Posts: 208
  • Joined: 19 Apr 2009
  • Loc: far outside the wire

Posted 02 January 2015 - 03:56 AM

i'm cheap, so i'd be looking for an alternative to the renishaw encoders

 

my plan might look something like an image sensor reading a series of encoded tracks on the edge of a disk

 

use as much of the sensor as needed to determine absolute position and a reduced window for the deltas

 

some sort of light source/lens/focusing is needed...

 

 

 

http://www.aptina.co.../mt9p031i12stm/

 

some back of the napkin numbers (better check the math):

 

2.2um pixel pitch and 12bit adc resolution on the image sensor

let's say 6 bits are reliable for determining the track feature edges

detectable motion ~ (2.2/64) u

 

sub arc second resolution is desirable, so ~10,000,000 divisions

2.2u / 64 * 10000000 ~ .34 m circumference,  11 cm diameter  disk

 

 

 

the beaglebone black has a camera cape and can run linux/opencv

 

the larger packaged stm32f4 chips have camera and external ram interfaces



#43 steveastrouk

steveastrouk

    Ranger 4

  • -----
  • Posts: 332
  • Joined: 01 Aug 2013
  • Loc: State College, Pa.

Posted 02 January 2015 - 08:48 AM

i'm cheap, so i'd be looking for an alternative to the renishaw encoders

 

my plan might look something like an image sensor reading a series of encoded tracks on the edge of a disk

 

There are plenty of mag tape reading systems with resolutions down to 10um, and with the ability to interpolate, which are a fraction of Renishaws. That said, I've specced a very high end Renishaw optical angular encoder into a system, with 0.1uRad resolution for around 500 bucks.



#44 James74

James74

    Vostok 1

  • -----
  • Posts: 119
  • Joined: 29 Jan 2012
  • Loc: Sacramento, CA

Posted 02 January 2015 - 01:16 PM

The only reason I specified Renishaw is because a lot of mounts are using them now and they have high resolution. They are also less susceptible to mounting error.

I know that Gurley makes good rotary encoders, but their resolution isn't high enough to place directly on the axis. I did see a project out there that uses a cheap rotary encoder connected to the axis with a gear to count in between the Gurley ticks. The Gurley encoders are $600, though.

What other encoders have y'all seen?

#45 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 6377
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 04 January 2015 - 01:04 PM

1) the CalPoly guys wound their own direct-drive motors

 

2) for my encoder project, I used Renishaw RLS LM10IC encoders, these are cheap ($100 - $200 per read head) and use cheap Bogen magnetic tape.

 

https://dl.dropboxus...landoAndico.pdf

 

The downside is, they are only 1um resolution with an interpolation error of +/- 3.5um.

 

To get say 1 arc-sec pointing, the circumference of the magnetic tape surface must be large enough.  Since 1 rotation is 1296000 arc-seconds, if you had a 2-meter circumference that would give you 2M ticks or about 0.5 arc-sec per tick, and +/- 2.5 arc-sec error. By using two read heads you can cut the interpolation error roughly in half.

 

A 2-meter circumference encoder disk would be about 24" in diameter - useful on say a large fork mount, but otherwise not.

 

Renishaw does make Signum optical encoders with RESM rings, and these are much finer than 1um resolution, but they also cost a lot more.


Edited by orlyandico, 04 January 2015 - 01:19 PM.


#46 LarsMalmgren

LarsMalmgren

    Lift Off

  • -----
  • Posts: 11
  • Joined: 22 Jan 2010
  • Loc: Denmark

Posted 04 January 2015 - 04:16 PM

How about using a DVD or Blue-Ray as an encoder?

 

With a writable disc a 010101 pattern could be written to the track which then could be read back when in operation as an encoder.

I've been googling for info about how small "holes" can be burnt into the disc and of-cause it varies a lot between different optical disc systems (Laser Disc, CD-ROM, DVD, Blue-Ray, etc.)

But it seams 0.5 um is doable - at least according to this source >> http://www.cdrfaq.org/faq02.html#S2-43

 

The disc is about 12 cm in diameter, so do we have enough resolution??

What about placing several read-heads on the same disc and interpolating their data !?

 

I assume a custom built "drive" with custom firmware/arduino-controller but build from scrap Blue-Rays...



#47 gregj888

gregj888

    Viking 1

  • -----
  • Posts: 711
  • Joined: 26 Mar 2006
  • Loc: Oregon

Posted 05 January 2015 - 07:48 PM

Lars,

 

The Blue-Ray would be a ground up effort.  As far as I know they still use a sector format so it's not just bits around the track.  There are blank spaces and sector numbers...  then there's the whole issue of what you reference too, the motor on a Blue-Ray isn't that precise.  0.5um is about as good as you can do with a laser even with blue light, seems pretty small.  In any case, you're in the same resolution regime as the Renishaw's.

The Gurley's can get you to 7m ticks but those are probably real expensive.  There are linear encoders that are > 0.1um, again, probably expensive.

In round numbers you want 4m plus ticks per revolution for direct drive.  In theory, you could gear up a lower CPR encoder with a 360k tick encoder on the telescope axis.  The geared encoder would give enough ticks for the direct drive system while the telescope controller would do the actual pointing.

 

The Gurley's use interpolation, not sure how they do that, but might be an option with some clever work.  Maybe the analog signal from the encoder inputs?

 

The motor isn't hard, the encoder is the real key.



#48 steveastrouk

steveastrouk

    Ranger 4

  • -----
  • Posts: 332
  • Joined: 01 Aug 2013
  • Loc: State College, Pa.

Posted 06 January 2015 - 10:45 AM


The Gurley's use interpolation, not sure how they do that, but might be an option with some clever work.  Maybe the analog signal from the encoder inputs?

 

Yes. Fed into circular interpolators from the likes of icHaus. The Renishaw Atom range do that too - they're the ones I'm using on a professional project right now.

 

One way I saw in a research paper of interest was the use of a very modest camera and a laser to create speckle on shot blasted wheel. The scatter pattern was correlated in real time, and absolute nano-arc second resolutions was achieved. That was before the era of optical mice, and I have mused to myself that one could create a very high resolution incremental encoder, after extending the optics of the usual mouse.



#49 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 6377
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 08 January 2015 - 04:52 AM

The Gurley's use arctan interpolation.  There is significant interpolation error.  The paper I wrote (linked above) has a ton of references on how interpolation is done.  There are a variety of techniques to correct for the interpolation error, e.g. best-circle fitting, or Heydemann interpolation, but the Gurley's don't do this (which is why SiTech has to implement tick management - to remove the rather massive interpolation error)

 

https://sites.google...erpolation/Home

 

Tick management is a rather clever hack that uses the ticks of the motor encoder to index the Gurley, and build up a correction table in real-time.  This is because there are "many" motor ticks, per tick of the Gurley.  In a direct-drive system, tick management wouldn't work since you don't have a motor encoder separate from the main axis encoder.

 

Note that Dave Rowe is also mentioned in the context of Gurley tick management.

 

The speckle pattern and camera technique is - as far as I can guess - the subject of patent EP2275782A1 (attributed to David Rowe, Filippo Riccio, Ivan Mariotti, and Thomas Baader) - which, also as far as I can guess, is the technology used in the encoders inside 10micron mounts.

 

The new optical mice also use Avago chips to read the surface the mouse is on - but sadly the  resolution on these things is poor, 19200 DPI at best, and no word on linearity.


Edited by orlyandico, 08 January 2015 - 04:57 AM.



CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.



Recent Reviews


Imaging/Sketching Contest





Also tagged with one or more of these keywords: mount



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics