54 replies to this topic

[benula]

I bought an old CG-11 a couple of years ago with the intent of doing astrophotography. It came with the Celestron 492 digital drive which allowed me to pursue that goal though it lacked GOTO abilities and all of the automation features that come with an intelligent controller. Getting up at 2am to spend 30 minutes pointing the scope and framing a new target grew old pretty quick and it wasn't long until I convinced myself that I needed a new mount controller. I weighed my options and decided that OnStep was the best solution for me.

 

As with all of my projects, life got in the way and it ended up taking almost a year for me to finish this (this is a hobby though, right?). If someone was really motivated, building OnStep could probably be done in a single weekend. I think I probably spent about $150 on the parts not including the use of a 3D printer (which is another one of my hobbies), though that really isn't necessary. Not bad when compared to what it would cost to buy an off-the-shelf solution.

 

I have used OnStep under the stars probably 5 times now and it was worked flawlessly. The GoTos are spot on, the tracking is perfect, and the INDI (Linux equivalent of ASCOM) driver and Androind app work great. I would whole heartedly recommend OnStep to anybody that is considering looking for a GoTo solution (you are in the ATM/DIY section, after all).

 

-Ben

[benula]

I built the OnStep controller by adding components to a piece of perfboard. Here are the components that I used:

 

* Teensy 3.1

* Trinamic TMC2130 stepper drivers

* Trinamic SilentStepStick protectors

* boost converter to convert 12V input power to 24V for stepper drivers

* HC05 bluetooth module

* ESP-01 ESP8266 wifi module

* 400 step NEMA 17 68 oz-in stepper motors 

 

The motor connectors are 6-pin DIN connectors though only 4 pins are used for the steppers. At some point I might use the additional pins to implement a worm indexer for PEC. The male DIN connectors are tricky to solder and I spent quite a while chasing motor issues that were caused by an intermittent connection in my cable.

 

The case was designed and 3D printed to fit in the footprint of the original 492 drive so it connects nicely to the mount. It turned out OK, but it's functional so I don't plan on redoing it.

 

Edited by benula, 09 October 2017 - 12:50 PM.

[benula]

I made the motor mountings to connect the NEMA 17 steppers to the existing out of a piece of aluminum using only a table saw, drill press, and an inkjet printer. I drew up the adapter schematic in FreeCad and then (2D) printed it to scale and glued it to the aluminum plates. Then, a machinist punch was used to create a dimple in the aluminum so that the holes could be accurately drilled on the drill press.

 

After a lot of search on ebay, I found some suitable motor couplings to connect the stepper motor shaft to the worm. The trick here was finding couplings that had an outer diameter less than 18 mm (I think) since there is limited clearance on the mounting plate that holds the worm blocks. Most couplings for shafts of this size have an OD of closer to 25mm. This wouldn't be a problem if you have access to a lathe, though.

 

I designed and printed some friction fit covers for the motors that the DIN connector could be screwed to so the stepper wires couldn't get yanked around. This works pretty well, but I've found that the long stepper motors can torque the 2 screw motor mount on the G-11 enough to flex quite a bit.

 

[benula]

Another shot of the motor mounts.

[benula]

 

[orlyandico]

interesting.. direct drive from the steppers to the worm? no gear reduction?

 

you'd need some pretty serious micro-stepping and even then, is there no stepper jitter at high power visually?

[benula]

interesting.. direct drive from the steppers to the worm? no gear reduction?

 

you'd need some pretty serious micro-stepping and even then, is there no stepper jitter at high power visually?

Yes, there is no gear reduction. I was a little worried about this before I actually tried it. The steppers are 400 step/revolution rather than the more common 200 step/revolution variety. The TMC2130s are used in 32 microstep mode with 256 microstep interpolation which are incredibly smooth (and silent). I'm aware of the pitfalls of microstepping but I have not noticed any vibrations at high power viewing, though I don't really do that often.

Edited by benula, 09 October 2017 - 11:57 AM.

[RAC]

interesting.. direct drive from the steppers to the worm? no gear reduction?

you'd need some pretty serious micro-stepping and even then, is there no stepper jitter at high power visually?

Have you ever seen Trinamic stepper drivers in action? My tmc2100 drives are far more smooth than a friends $500 applied motion drives. The world needs to totally convert to Trinamic and forget every other drive out there. My stepper motors turn like servos.

[hjd1964]

interesting.. direct drive from the steppers to the worm? no gear reduction?

 

you'd need some pretty serious micro-stepping and even then, is there no stepper jitter at high power visually?

I also have a OnStep G11 mount w/ TMC2130 stepper drivers and Nema17 400 step motors direct into the worms.  I don't do much planetary/lucky imaging and perhaps another 2:1 or 3:1 reduction would be advisable if I did, not sure.  We haven't exhausted the TMC2130 bag of tricks either, like programming a custom micro-step table to match the stepper motor.  Looks like a complicated thing to do though and since we're 1:1 into the worm a higher speed PEC might be able to do essentially the same thing.  You might have a tough time measuring anything (against the sky) at this level though...

 

The following star image was taken with my G11/10" F/4 Newtonian/Baader MPCC and an ASI178MM camera.  It's a stack of about 84 x 1 second exposures unguided @ just a hair under 0.5 arc-sec/px (RA is aligned left/right.)  There was a slight haze, not sure what the seeing was like (don't recall.)

 

 

I haven't had this camera for long and the above data was from testing during collimation... stars are almost as tight at this scale with longer exposures and I think the difference can be attributed to the challenges of guiding, 48 x 4 minute exposures:

 

Edited by hjd1964, 10 October 2017 - 11:59 AM.

[PrestonE]

I am so glad to hear this...

 

As I just ordered TMC2130 stepper drivers and Nema17 400 step motors with

Teensy 3.2's for 3 different mounting systems 2 weeks ago after spending the

better part of a month reviewing and trying to understand everything on the

OnStep Yahoo group...

 

Looking forward to trying these soon.     

 

Best Regards,

 

Preston

[orlyandico]

ok I get it. 400 steps/rev * 360 teeth on the G11 worm wheel gives 9 arc-seconds per step.

 

I don't really believe one can get 256 micro-steps, but even at a conservative 16 micro-steps that's 0.6" per micro-step so should be good..

[ron scarboro]

I'm just wandering into this arena of astronomy, so apologies if this is a naive question.

 

What slew rate do you get with the motors?

 

Thanks,

 

Ron

[TonyStar]

ok I get it. 400 steps/rev * 360 teeth on the G11 worm wheel gives 9 arc-seconds per step.

 

I don't really believe one can get 256 micro-steps, but even at a conservative 16 micro-steps that's 0.6" per micro-step so should be good..

The Teensy outputs 16 u-steps, the driver interpolates to 256 u-steps for a super smooth motion:

 

http://blog.trinamic...tors-explained/

[orlyandico]

I may have to check out those Trinamic's for myself.. in any case doing away with the gearhead gets rid of one major source of periodic error. Assuming the motor output is indeed "super smooth" then the only periodic error should be from the worm itself (and any de-centering of the shaft coupler).

[kbahey]

Oh, what a beauty of a project! Really well done ...

 

I am embarking on doing the same thing, with Teensy, TMC2130, and also directly attach the motors to the worm gear to avoid Non-Period Error from the gear train.

 

I may be mistaken, but aren't the 400 step motors just 200 step motors with gear reduction built in?

 

@PrestonE and @benula, can you please point to specific models of stepper motors, and TMC2130, and where you got them? That would help a lot.

[benula]

I'm just wandering into this arena of astronomy, so apologies if this is a naive question.

 

What slew rate do you get with the motors?

 

Thanks,

 

Ron

During GoTos my mount is probably slewing 2-3 degrees/second. I know it can slew at twice that rate without breaking a sweat with my imaging load (which is very modest for a G11). I think Howard has experimented quite a bit so I'm sure he could offer more insight.

[benula]

Oh, what a beauty of a project! Really well done ...

 

I am embarking on doing the same thing, with Teensy, TMC2130, and also directly attach the motors to the worm gear to avoid Non-Period Error from the gear train.

 

I may be mistaken, but aren't the 400 step motors just 200 step motors with gear reduction built in?

 

@PrestonE and @benula, can you please point to specific models of stepper motors, and TMC2130, and where you got them? That would help a lot.

0.9 degree steppers actually have twice the number of rotor teeth than 1.8 degree steppers.

 

The motors that I ordered are these ones from sparkfun.

 

I ordered my TMC2130s from filastruder but later found they are much cheaper from digikey. You should take some care in ordering the 5V or the 3-5V version of the SilentStepStick. The 3-5V versions have specific power sequencing requirements. I used the 3-5V version and used an N-channel MOSFET with the gate controlled by Vmot so that Vio wouldn't come up until Vmot was available. I'm not sure it's really a big deal though if Vio and Vmot come up at approximately the same time. I think you can also buy PCBs from someone in the yahoo group and then you don't have to worry about any of this.

 

-Ben

[kbahey]

@benula

 

Thanks a lot for all this info.

 

The board for sale on the Yahoo Group (Steve's board) is the older design, based on ArduinoMega2650 and DRV8825 drivers. The Teensy and the TMC2130 are superior.

 

I wish someone would make such a board. I would buy it instead of all the soldering.

[RAC]

There's not alot of soldering if you use pre made female-female terminal wires that lots of places sell to suit teensy's ect.... they clip into plugs made for the job. I use little stepper driver shields and just connect it all up with those wires.

[gregj888]

There are board files on the Yahoo website,  There are several versions now.  I just had 3 rev 1 board made up ( Express PCB, 3 for ~$65).  If you make up a few more and do some shopping the boards can be a lot less than $22 each.

 

I OnStepped an old Meade 8" SCT and it's working well.  The PCBs will replace the hand wired board shortly for higher reliability.

 

A note on the motors.  I'm using 8825 drivers and 32 step micro steps.  The motors are geared stepper motors.  These have lower step counts and higher gear ratios.  The system pulls <120 ma tracking and 350ma with both motors slewing.   I can run all night on a 12v LIPO pack I fit into the fork base.  14500 micro steps/degree...  Slews aren't "snappy" but are plenty fast.

 

http://www.ebay.com/...EAAAOSwARZXkjL2

 

BTW, I very suspicious of the listed gear ratio...  but all seems to work well. 

[benula]

It looks like there are two PCB design for the Teensy versions on the Yahoo group and one of them is designed by Howard. Even if nobody is selling them you could buy batch from a board house like seeed studio for probably less than $10.

[hjd1964]

There's not alot of soldering if you use pre made female-female terminal wires that lots of places sell to suit teensy's ect.... they clip into plugs made for the job. I use little stepper driver shields and just connect it all up with those wires.

Yes those should be fine for a SSS TMC2100 and are a good choice in many cases.  You can then jumper the drivers for the one micro-step mode/decay mode that works best overall.  Sometimes there is no real advantage to the TMC2130 since one of the TMC2100 modes just works so well for both tracking and slewing. 

 

Other times more flexibility is nice and the TMC2130 can help, but this requires wiring in the SDO/SDI/SCK/CS connections from each driver (and not to the default hardware SPI port but we need not get into those details.)  Here are some of the things you can do: 

1. Switch between decay modes for tracking vs. slewing.
2. Ramp power up for slews and back down for tracking.
3. Have finer than 16X micro-stepping granularity (16x for a G11 would be 0.6" resolution for guiding.)  You can use any mode from 1x to 256x (with or without 256x interpolation.)
4. Etc. etc. etc. read the datasheet to get an idea of just what is possible, I've only scratched the surface here.

[hjd1964]

It looks like there are two PCB design for the Teensy versions on the Yahoo group and one of them is designed by Howard. Even if nobody is selling them you could buy batch from a board house like seeed studio for probably less than $10.

You really need to use ExpressPCB for my OnStep Mini PCB, their file format is propitiatory.  It would be nice to break free and convert into Gerbers especially for folks not in the US.  There was a program CopperConnection that could, from what I understand, do a good job converting ExpressPCB format to Gerbers... want to guess who bought them?  You can still get the Gerbers, but only for a fee.

 

Annoying but don't get me wrong, I understand that ExpressPCB is running a business and their service/price/quality have been fine for me and I really don't have much interest in learning another PCB design tool.

[hjd1964]

 

I'm just wandering into this arena of astronomy, so apologies if this is a naive question.

 

What slew rate do you get with the motors?

 

Thanks,

 

Ron

During GoTos my mount is probably slewing 2-3 degrees/second. I know it can slew at twice that rate without breaking a sweat with my imaging load (which is very modest for a G11). I think Howard has experimented quite a bit so I'm sure he could offer more insight.

 

In general users report this G11 configuration works well at 2 or 3 degrees/second with payloads to the limits of a G11.  I routinely run with just over 40 lbs. on my G11 and the stealthChop mode is fine at 2.5 deg/s.  In spreadCycle I'll sometimes use higher speeds up to 3.5 deg/s.  These are solid performing, never an issue, examples.  Plenty of room for when things go wrong like re-configuring my scope/payload this spring I released the lock on the RA axis and the tube swung around quickly due to being seriously out of balance (whoops.)  This had no affect on goto's.

[marcosbaun]

I am using the TMC 2100 driver on my platforms. Tmc 2100 is fantastic. Extremely smooth. It has a very good motor current control.
Marcos