Jump to content


Photo

Harmonic Drive Telescope Mount

  • Please log in to reply
23 replies to this topic

#1 Ed Wiley

Ed Wiley

    Apollo

  • -----
  • Posts: 1045
  • Joined: 18 May 2005
  • Loc: Texas, USA

Posted 06 February 2014 - 05:30 PM

Anyone used one of these mount?

http://www.gototeles...ic mount 1.html

125 lb capacity and way under 10K. That might make a nice entry into harmonic drives, but I know nothing about such mounts in terms of PE, tracking and etc.

Thanks, Ed

#2 David Pavlich

David Pavlich

    Transmographied

  • *****
  • Administrators
  • Posts: 27932
  • Joined: 18 May 2005
  • Loc: Mandeville, LA USA

Posted 06 February 2014 - 06:11 PM

This is the only mount I've seen with harmonic drives, but as you can see, I can't think of a reason why someone would want to spend that kind of money for a 250lb capacity mount. It does have the advantage that you can image from horizon to horizon without a flip, but still, that's a lot of astrobucks!

David

#3 Skunky

Skunky

    Ranger 4

  • -----
  • Posts: 357
  • Joined: 16 Sep 2013

Posted 06 February 2014 - 06:51 PM

Nice.. would like to hear more about this mount.

#4 David Pavlich

David Pavlich

    Transmographied

  • *****
  • Administrators
  • Posts: 27932
  • Joined: 18 May 2005
  • Loc: Mandeville, LA USA

Posted 06 February 2014 - 10:43 PM

Here is an excellent tutorial on the inner working of a harmonic drive. Note the emphasis on "zero backlash", hence the reason that it makes for a good telescope mount.

David

#5 Ed Wiley

Ed Wiley

    Apollo

  • -----
  • Posts: 1045
  • Joined: 18 May 2005
  • Loc: Texas, USA

Posted 06 February 2014 - 11:19 PM

Thanks, David, very informative. But, what about this particular mount? Anyone have one or knows anything about it? 6K is quite a price break compared to 20K.

Ed

#6 austin.grant

austin.grant

    Viking 1

  • *****
  • Posts: 768
  • Joined: 18 Oct 2010
  • Loc: Shreveport, LA

Posted 06 February 2014 - 11:33 PM

Did you watch the tracking and PE videos? Looks impressive.

#7 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5719
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 06 February 2014 - 11:57 PM

Harmonic drives have considerable PE but it is long period (the period is the entire revolution). However most harmonic drives are 50:1 to 100:1 reduction so you need another gearhead to drive them. I suspect the 14.4 minute period of the chronos hd32 implies it has a 100:1 final gear ratio... (1440 minutes/day).

There's a guy on the Sitech yahoo group who built his own harmonic drive mount. The drives can be had for $500 to $1000 per axis on eBay. Of course list price is much higher.

#8 OzAndrewJ

OzAndrewJ

    Apollo

  • *****
  • Posts: 1031
  • Joined: 30 Nov 2010

Posted 07 February 2014 - 12:03 AM

Hmmm, shows really low numbers for tracking and PE.
Wonder where the encoders are mounted and how fast the feedback loops are???
Looking at the pipe its fitted in, the motor gear could only be a max of say 6" dia. At that dia, 1 arcsec of error equates to a tangential error of 36um at the gear pressure point. Not many gears will be made to that accuracy for the full 360 deg of rotation, so it would be interesting to see how it gets accounted for.
Being a fixed system, it could easily be factory modelled??
Intriguing :-)

Andrew Johansen Melbourne Australia

#9 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5719
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 07 February 2014 - 12:06 AM

The chronos specs 7.8" of PE. Which is not particularly good. The period is slow however so its easy to guide. Or not guide, by keeping the exposure to less than 1/2 of the drive period.

Harmonic drives don't need encoders. They are just another variety of gear reduction.

#10 OzAndrewJ

OzAndrewJ

    Apollo

  • *****
  • Posts: 1031
  • Joined: 30 Nov 2010

Posted 07 February 2014 - 12:10 AM

Gday orlyandico

Harmonic drives have considerable PE but it is very long period


I hoped you might post here based on yr recent "playing" with stuff at this accuracy level.
Its been 30 years since i did gear design at uni, but the harmonic drives still rely on teeth changing position under load, which means any tooth form errors on individual teeth will also come into the mix.

Based on my prev post, as this gear is directly on the output shaft ???? then tooth form errors may result in higher freq PE per tooth on top. Dunno???

Andrew Johansen Melbourne Australia

#11 OzAndrewJ

OzAndrewJ

    Apollo

  • *****
  • Posts: 1031
  • Joined: 30 Nov 2010

Posted 07 February 2014 - 12:20 AM

Gday orlyandico

Harmonic drives don't need encoders. They are just another variety of gear reduction.



I fully understand how the harmonic drives work, i was more interested in how the feedback loop would work to give the quoted numbers, as unless there is an absolute encoder on the output axle, you are still using dead reckoning for position.
Sure it can be guided, but i wonder what the base tracking is like over multiple teeth locations.
All good fun.

Andrew Johansen Melbourne Australia

#12 TxStars

TxStars

    Apollo

  • *****
  • Posts: 1391
  • Joined: 01 Oct 2005
  • Loc: Lost In Space

Posted 07 February 2014 - 12:43 AM

With multiple teeth contacting at all times it would be the average error of the worst fit of the two rings.
Over time you could adjust the base start position of each motor to reduce the error and larger drives could be used.

#13 OzAndrewJ

OzAndrewJ

    Apollo

  • *****
  • Posts: 1031
  • Joined: 30 Nov 2010

Posted 07 February 2014 - 12:56 AM

Gday TxStars

Ahh, but in this scenario, i suspect they wont all fit and average that smoothly at the micron level, hence my query of how much other feedback is reqd.

I agree there will be some form of mechanical averaging going on, but i suspect the primary contact points when transitioning teeth will still be individual teeth, esp when tooth spacing and form errors come into the mix.
Need to read up some more on the math behind exactly how the teeth would transition in this design, but it obviously can be made to work.
Just be interesting to see how much electronic adjustment is reqd for unguided tracking.


Andrew Johansen Melbourne Australia

#14 Hilmi

Hilmi

    Soyuz

  • *****
  • Posts: 3813
  • Joined: 07 Mar 2010
  • Loc: Muscat, Sultanate of Oman

Posted 07 February 2014 - 12:58 AM

What would put people off is that the web page looks unprofessional and the description of the product talks about inconsistencies as if they were part of the design.

#15 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5719
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 07 February 2014 - 01:43 AM

I believe the "value proposition" is that you are contacting 1/3 of the teeth at all times. Hence averaging.

As to whether that would work at a micron level... well regular worm gears only contact a few teeth and they work fine. I'd expect harmonic gears to be better. How much better? Based on the hd32 spec, not even an order of magnitude.

And while there's no backlash, there is some sort of stiction / starting torque due to the flexing of the spline. That I suspect would also look like backlash from an electronic perspective.

#16 TxStars

TxStars

    Apollo

  • *****
  • Posts: 1391
  • Joined: 01 Oct 2005
  • Loc: Lost In Space

Posted 07 February 2014 - 02:00 AM

Don't forget the harmonic drive only replaces the worm and ring.
You still need a drive motor to make it work.
And with no positional feedback you are "shooting in the dark". :p

#17 orlyandico

orlyandico

    Fly Me to the Moon

  • *****
  • Posts: 5719
  • Joined: 10 Aug 2009
  • Loc: Singapore

Posted 07 February 2014 - 02:26 AM

well all encoder-less mounts are shooting in the dark...

as i mentioned the harmonic drive replaces the worm and ring gear. but you still need a smaller gearhead (whether also harmonic or a regular spur or planetary) to lower the motor speed. that of course would also introduce errors.

i'm somewhat curious though. conventional worm and ring gear mounts suffer when the motor reduction gearhead is of poor quality (e.g. CGEM 8/3) because the gearhead harmonics get transmitted to the OTA.

but with a main harmonic drive.. would it be possible that the inherent damping in the spline would suppress gearhead harmonics?

this is one of those things I'd like to experiment with, but with the prices of harmonic drives......

#18 TxStars

TxStars

    Apollo

  • *****
  • Posts: 1391
  • Joined: 01 Oct 2005
  • Loc: Lost In Space

Posted 07 February 2014 - 03:14 AM

A large HD driven by a small HD directly connected to the drive motor would be interesting.

#19 OzAndrewJ

OzAndrewJ

    Apollo

  • *****
  • Posts: 1031
  • Joined: 30 Nov 2010

Posted 07 February 2014 - 05:33 AM

Gday Orlyandico

well all encoder-less mounts are shooting in the dark...



I postulate that even drives with encoders are still shooting in the dark if the encoder isnt on the output axle, hence my intrigue as to what the raw unguided tracking of these mounts is like.

but with a main harmonic drive.. would it be possible that the inherent damping in the spline would suppress gearhead harmonics?



I am sure with a harmonic drive that you wont get any gearhead "harmonics" like the Celestron or Meade 8/3 error but you may get localised errors on a tooth by tooth basis that cant be addressed by a simple PEC model.
Be fun to get a gearhead unit to test what it looks like across multiple teeth.


Andrew Johansen Melbourne Australia

#20 schluterdude

schluterdude

    Vostok 1

  • -----
  • Posts: 174
  • Joined: 05 Aug 2013
  • Loc: Centerville, OH

Posted 07 February 2014 - 05:36 PM

Many machines I work on (high accuracy cylindrical grinders) use harmonic drives to swivel the wheel head, and will repeat a position down to a 0.001* degree accuracy... That being said, they also seat to a hurth coupling...

BUT, in a very nice, high accuracy setup, they are tough to beat for their size and weight. PE can be finessed out of the equation with good software driving it. The other alternative, as we all know, is nice worm gears/wheels, which can be comparatively heavy...

Has anyone used torque motors? Some machines I work with (tool grinders) use torque motors for their rotory axis, and produce mirror finishes on corner radii. But, again, big and heavy... Strong like bull though!

#21 Per Frejvall

Per Frejvall

    Messenger

  • *****
  • Posts: 424
  • Joined: 28 Sep 2012
  • Loc: Saltsjöbaden, Sweden

Posted 07 February 2014 - 05:43 PM

0.001° = 3.6 arc-seconds... Well above average seeing.

/per

#22 OzAndrewJ

OzAndrewJ

    Apollo

  • *****
  • Posts: 1031
  • Joined: 30 Nov 2010

Posted 07 February 2014 - 06:14 PM

Gday schluterdude

and will repeat a position down to a 0.001* degree accuracy



So about 3.6arcsecs. ( I assume thats with properly calibrated software driving them as well?? )
How much do they cost???
ie the initial mount linked to quotes $6000 for the whole mount.
I have been looking for specs/prices on these sort of drives and so far i am seeing results like +/-1arcmin for larger units down to +/-10arcsec for the micro units.
( For anyone interested, the micro units are amazing. Basically the size of a bee.
http://www.orlin.co....onic_drives.htm )

I guess like worms and anything else mechanical, there are ranges of quality in the drives and the price goes up with the accuracy.

PE can be finessed out of the equation with good software driving it.



Again, thats part of my earlier query. ie what is the native PE of a gear design like this and how much effort goes into software / feedback loops to get the last bits of accuracy???
I have found a range of absolute positioning specs, but nothing on tooth PE yet.
I guess for most robotic operations, its only final position that matters, so they dont post specs on that side of things.
Interesting topic.

Andrew Johansen Melbourne Australia

#23 schluterdude

schluterdude

    Vostok 1

  • -----
  • Posts: 174
  • Joined: 05 Aug 2013
  • Loc: Centerville, OH

Posted 07 February 2014 - 06:56 PM

I would have to dig up engineering drawings of the system for their specs...

The system is run through a FANUC CNC control though... Costly!

Though, that's what we use for the entire machine. One can always use a different control. Pointing at and tracking a spot in the sky using two axis is MUCH different than using 5 to make a part! Easier in some respects, tougher than others...

#24 Odell

Odell

    Explorer 1

  • *****
  • Posts: 94
  • Joined: 29 Apr 2012
  • Loc: Florida

Posted 07 February 2014 - 09:44 PM

Last time I corresponded with Leonard at GoTo & Tracking Systems he was closing his shop. I was going to have him do a conversion on my LX. Don't know if he's fired up his operation or not. I just had a Sitech conversion performed on my CGE. Haven't gotten out from under the clouds yet,so I don't have any solid accuracy numbers but it isn't a stock CGE anymore.

Ian






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics