DIY Harmonic Drive Equatorial Mount

A home built equatorial mount for my telescope built mostly recycled components I have had lying around.

Similar projects worth following
As if I didn't have enough projects going on all at once I have started yet another, an equatorial mount for my telescope. This stems from trying to set up my 10" Meade LX200 for astrophotography. By the time I added the off axis guider, the filter wheel, focuser, and finally a camera on the scope I lost a lot of sky due to everything hanging off the back. Add a wedge to turn the alt/az form mount into an equatorial means I lost any chance of pointing north.

The solution seemed to be to take the OTA (Optical Tube Assembly) off the forks and mount it to something like a German Equatorial Mount (GEM). A GEM allows the scope to swing around without clearance issues for the most part. But, boy are they expensive. A low end one if $1500+!

So I figured it is time to build one myself. Probably 10 years ago I picked up an old PRI robot that had come out of intel or something, I think it had moved around wafer boat or something. Took most of it apart, gave the steppers away, and kept the useful stuff. Bunch of linear slides and stuff. The one gem was the big harmonic drive that the robot rotated on. The OD on the driven disc is 10" in diameter and it is supported by a nice, large ~8" bearing under it. All made out of aluminum so it is still reasonably light. This is going to be my Right Ascension (RA) drive.

Harmonic drives are very cool, they use a flexible spline on an eccentric bearing that wobbles inside another spline with a couple more teeth than the flexible one. Since there is alway hard contact between the flex spline and the outer splice the drive is truly zero backlash. They are used in a lot of places like robots, cnc and even the moon rover. It used one harmonic gear in each wheel. Here is a video of the one I am using, you can see the flexing of the spline. This is 102:1 gear ratio.

I have since taken it apart and cleaned it out, it runs much smoother now. The drive had a DC servo on it with a 500 count encoder. I tried setting it up with one of my little Elmo Harmonica servo drives like what I used with the laser cutter but it would just not tune. I dont thing the drive could supply enough current. Also after doing the calculations I found 500 lines is just not enough, that is less resolution than the mount I am already using. After messing around with various combinations of servos and gearboxes to drive it I have settled on a Mitsubishi MR-J2 200W brushless servo with a Bayside 10:1 gearbox. This drives the harmonic drive and gives me a resolution of 8,355,840 steps per rev. This puts me way above useful resolution.

For the Declination (Dec) drive I am using another Mitsubishi 200W servo, this one is a MR-J2S series with a 131,072 count encoder. This will drive another Bayside gearbox, this one 3:1 and then that will drive a 80:1 harmonic drive which will have the saddle for the telescope on it. This gives me stupidly high resolution of 31,457,280 steps per rev. I may try direct driving the harmonic drive bypassing the 3:1 gearbox, I am pretty sure it will have enough torque. The OTA is adjusted on the dovetail mount to balance it.

The Dec harmonic drive is pretty neat, it is basically a cross roller slewing bearing with a harmonic drive built in. I think these were pulled out of SCARA robots, I got one from S. Korea shipped for $180. More than I would have like to pay, but oh well. Considering I have had all the rest of the parts lying around I cant complain too much!

Im now drawing up the design for the tilt mechanism for setting the angle of the RA drive and figuring out how I am building the motor mount for the Dec drive.

This is the back of the main harmonic drive with a servo and the gearbox attached. There are commercial GEM setups available but they start around $20k!

  • 1 × Harmonic Drive Base Came out of a PRI Robot
  • 1 × Harmonic Drive SHF-17-80-2UJ, 80:1 Ebay, from SCARA Robot
  • 1 × Bayside 3:1 Planetary Gearbox Ebay
  • 1 × Bayside 10:1 Planetary Gearbox Ebay
  • 1 × Mitsubishi MR-J2S-20A1 Servo Drive and Motor Ebay

View all 14 components

  • Finally working on this darn thing....

    Jerry Biehler06/23/2017 at 04:55 0 comments

    So two years have gone by and I have not done a damn thing with this, just sitting in my living room collecting dust. Now that I am gainfully unemployed again I have some time to mess around with it.

    So, where did I leave off.. Oh yeah, filtering the PWM. Well, that was a mess. I couldn't get a stable drive signal out and the controller started acting up, it was replaced and I never touched it.

    I decided a few weeks ago to mess with it some more, I went with plan B this time, directly coupling an encoder to a small Escap motor to a CUI 2500 line shaft encoder act as a kind of integrator. I then tapped into the single ended encoder signals back to the controller with a 26LS31 line driver. This sends a differential signal to the mitsubishi servo drives which are set up for encoder follow mode.

    It works! Now I need to finalize the encoder tap and mount everything down as well as program in all the parameters specific to the mount.

    In the image below you can see the motor-encoders on the left, the mess of wires on the breadboard is the encoder tap and the black box below that is the SiTech telescope controller.

  • Low Pass Filter for PWM

    Jerry Biehler03/09/2015 at 21:41 0 comments

    I used LTSpice last night to do some circuit simulation. I hooked up the SiTech controller to the scope and found it uses a 23.45khz base frequency. Popping this info into LTSpice allowed me to come up with a pretty simple 2 pole low pass filter that should give me reasonable ripple and good step response with a 10v max output.

  • Back to work: Tripod and SiTech controller

    Jerry Biehler03/09/2015 at 04:00 0 comments

    I have finally started working on this mount again. I got an old Quick-Set Hercules 5302 tripod from a friend. It is rated for 150lbd but I am betting it will hold a bit more.

    I took another piece of the scrap 17-4 stainless and machine a spud that adapts the 1/2-13 threaded hole in the center of the mount to the 1-1/4" spud on the tripod column. A set screw in the spud locks in a steel socket in the column.

    The OnStep controller never really worked how I hoped so I have decided to go with a SiTech controller. A couple of days ago my SiTech Servo 1 controller showed up. First thing I am going to try is making a 2nd order filter to attach to the motor outputs of the controller to smooth out the PWM and then scale it to the +/-10v analog input the servo drives have. Then the encoder feedback from the drives will connect to the encoder inputs on the SiTech. If that does not work I will use the SiTech to drive a couple small dc servo motors to and then use another encoder to drive the Mitsubishi servo drives through their pulse inputs, they can be configured to take quadrature signal inputs as well as CW/CCW and step/dir modes.

    I used LTSpice to design a filter, we will see how well it actually works out. I put the controller on the o-scope and found it runs at a PWM freq of 23.45khz.

  • Control Box

    Jerry Biehler07/20/2014 at 05:52 0 comments

    One thing about a lot of japanese stuff is they really like to use more of the odd-ball connector. In the case of this it uses 20 pin MDR connects for all of the IO for the drive. So I ordered some of those off ebay and ordered some mating connectors for the servo motors from mouser. Or so I thought. In the manual it shows the part number for Molex MiniFit Jr connectors but they are really AMP Mate-n-lok connectors. They look the same but they keying is different and they use different pins. So I ordered the right ones and was able to get the servo cables made up.

    For the box to hold all the drive electronics and controller I used an old box from a communications package for a ScanEagle drone. Nice aluminum case with connectors already on the side. I got the wiring finished tonight, now I need tie it all together and see if it works.

  • Mechanically Finished

    Jerry Biehler07/01/2014 at 01:20 0 comments

    The OTA popped off the forks very easily, I did it in about 10 minutes last night. I went down to the hardware store and picked up the bolts for the saddle today and put it together. Some pics of it assembled.

    At this point I would say it is mechanically finished. Now I need to put together the controls.

    A bit overkill for a 10", I think. It weighs in at 96lbs without the scope. It has gone from portable to luggable.

  • Powdercoating and the OnStep Controller

    Jerry Biehler06/30/2014 at 03:19 0 comments

    hjd1964, a member of the CloudyNights forum, has been working on the code to get his OnStep software package working on a Teensy 3.1. It is working for the most part so far.

    I powder coated the last batch of parts for the scope mount last night and mounted the dovetail today. Now comes the fun part of pulling the OTA off the old forks. Whee...

    Here is the breadboarded Teensy 3.1, line driver, and bluetooth module:

    I did some gizmo triage and recycled a bunch of stuff, I now can see some of my desktop:

    Here are the parts just after being coated and ready to bake. 400F for 30mins.

    The dovetail mounted on the OTA. Funny thing, I went to bolt it down and I found my holes to be 1/16" off. I milled out one end turning the holes into slots and reinstalled it. Now the bolts fall in to where they were supposed to before. I have no idea what happened. Oh well.

  • Telescope Dovetail and Saddle

    Jerry Biehler06/23/2014 at 03:52 0 comments

    I made the dovetail and saddle for the telescope today. Figuring out where to put the holes to match the existing accessories hole on the scope was kind of a pain, no documentation anywhere on the hole spacing and angle. The dovetail should be compatible with Losmandy type accessories.

    I am probably just going to paint these. We will see. Now it is time to de-fork the telescope.

    Machining the dovetail on the telescope side.

    The four parts I made today:


  • Declination adjuster added

    Jerry Biehler06/16/2014 at 04:23 0 comments

    I got the declination fine adjuster installed last night. It came out of an old Sperry Gyroscope Tool Set I got a couple years ago. A little overkill but it should work fine. I also made up the clamps for the Dec adjuster as well.

    Next is the dovetail saddle for the scope and the mechanical portion of the scope will be done. I need to start ordering some parts for the control system, a couple AM26LS31 line drivers for the step/direction inputs of the servo drives and the bluetooth module for the controller. A guy at the local Dorkbot meeting gave me one of the Orbital LK204 serial LCD/button modules. I might try to use it as a control pendant.

    The adjuster, should barely clear the scope, it came out bulkier than I had hoped:

    The kit the parts came out of. I had just bought it for the case but figured the rest might come in handy someday.

  • Parts back and more assembly

    Jerry Biehler06/13/2014 at 04:04 0 comments

    I got my parts back from anodizing today and assembled what I have so far. 

    I think I have created a monster.

    I inserted the bearing cup into the base, a bit of a tight fit but it pressed into place and trimmed the spud to it's final length. I lubed up the two bearings with Krytox 240AC, a nonreactive teflon based grease, and put the base together. I glued the spud into the post mount with some Hysol 1C epoxy. Hopefully that will keep it from turning. I may have to pin it in place. 

    Everything went together pretty well, one spot I ran into interference between the bottom plate of the main harmonic drive mount and the Az fine adjust bar, it will only get in the way at low altitude settings which will never happen so I am not too worried right now. 

    Next up, figure out the angle lock and fine adjustment. I have a section of worm gearing that might work nice. I also need to make the dovetail saddle and buy the dovetail kit for my telescope, that will attach the scope to the mount. Then I need to get working on the control.

    The taper bearing in place and lock nut:

    The fine adjuster in place:

    Side pieces on:

    Another view:

    On the tripod, this is a standard Meade Field Tripod

  • Another batch of parts...

    Jerry Biehler05/28/2014 at 07:01 0 comments

    Another batch of parts finished for the telescope. I have decided to just buy a saddle and dovetail kit from ADM, I dont want to mess with that.

    Needed a piece of aluminum rod for the bearing spacer/axle. Couldn't find anything but I did have a piece of titanium, close enough...

    Making the spud that the base rotates around. I used some 17-4 TGP stainless I had left over from an old job years ago. Found an insert that would actually break chips with this stuff which just created tiny super hot little chips flying everywhere. Ouch!

    I am using a large ball thrust bearing for support and a taper bearing on top of that. The spud ties it all together and also serves as what the fine adjust presses against. I single pointed 5/8-18 threads for a nylok to tighten up the bearings.

    This piece is the fine adjustment arm, two Vlier fixture screws will push on either side and allow it to move around it's axis. It is slit down the middle to clamp on the spud and can be loosened for rough alignment.

    Everything ready for anodizing:

View all 15 project logs

Enjoy this project?



Tiago Damian wrote 04/26/2015 at 19:47 point

Very expensive steup;

  Are you sure? yes | no

Jerry Biehler wrote 04/27/2015 at 18:09 point

Only if you buy new. Or don't have the stuff to build it. Even at that it is way cheaper than a commercial harmonic mount. Though the trend now is going to direct drive telescope mounts with super high res encoders.

  Are you sure? yes | no

steven.sagerian wrote 10/14/2014 at 13:11 point
Does the harmonic drive contribute any vibration? Seems like with the large moving parts it would not be suitable for astro photography..but very cool none the less...

  Are you sure? yes | no

Jerry Biehler wrote 10/30/2014 at 19:38 point
Not really, not any more than an existing worm drive and there are commercially available harmonic drive mounts available.

  Are you sure? yes | no

Jasmine Brackett wrote 08/06/2014 at 20:40 point
Hello macona, I've left a comment on letting you know what's required for your projects to be considered for the next round of The Hackaday Prize. They apply to this project too. Thanks, Jasmine

  Are you sure? yes | no

Ken Meyer wrote 07/08/2014 at 08:18 point
I love the project! It's been a long time since I pulled out my old telescope, but I've got little kids now and I can't afford to miss sleep!

It looks like you entered all your tags into one field (separated by spaces). Consider breaking up the terms into separate tags so people can find your project more easily!

  Are you sure? yes | no

Jerry Biehler wrote 07/09/2014 at 03:56 point
Thanks, it looks like they changed the tag system since I had entered them. Before it was all on one line.

  Are you sure? yes | no

Adam Fabio wrote 06/18/2014 at 01:59 point
Holy cow. I'm speechless. Anyone who's turned a medium powered telescope toward the heavens - and (then tried to keep their objective in view) knows how important a drive is. I love the machining you're doing! Keep up the good work - and Thanks for entering The Hackaday Prize!

  Are you sure? yes | no

Jerry Biehler wrote 06/18/2014 at 05:07 point
Thanks. I am mostly hoping this does not become the biggest Hackaday Fail in history. ;-)

  Are you sure? yes | no

vibrolax wrote 05/11/2014 at 23:42 point
I work with harmonic drives for a precision optical positioning application. I don't know what your dynamic pointing accuracy needs to be, but the ones we are using show about a 1 mrad peak-to-valley systematic error. It's sinusoidal with one period every 120 degrees. Harmonic drives don't have backlash, per se, but they do exhibit a windup-type error motion that can easily be seen when you reverse directions. For equatorial tracking, this may not bother you. We use a flexible linear encoder wrapped around the outside radius of the axis for feedback, but we need to change directions frequently. BTW, mad props to your very nice project.

  Are you sure? yes | no

Jerry Biehler wrote 05/12/2014 at 21:39 point
Thanks. I have been looking at the flexible strip encoders from Renishaw. I am hoping to find a strip and reader on ebay. A friend had a couple Micrion PZT XYT stages that had them on the Theta axis. Of course that means I need to handle an external encoder loop.

  Are you sure? yes | no

Will wrote 05/02/2014 at 09:09 point
I've been talking to my Dad about building one of these... unfortunately the stuff I have "lying around" is in a slightly different league :) Looks great!

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates