Celestron conversion to stepper-pulley/belts with ONSTEP

[ChrisV]

A while ago I built the onstep controller (http://onstep.groups.io/g/main/wiki/home) for an EQ5 mount. It has been going well with a 200mM Rokinon lens with an ASI071 for AP, and an 80mm refractor for live imaging.

So now I’m going to have a go at my celestron CGEM mount which I use for AP – on the 80mm refractor and a 200mm F5 newt. Two reasons. It’s a noisy coffee grinder and the RA error when guiding is always much larger than the DEC error. I thought this might be due to the terrible gears. Sometimes it results in eggy stars. On the ONSTEP forum someone has said that they’ve managed to use a belt/pullies inside the CGEM with stepper motors. So I thought I’d start a thread to provide a bit more detail. For this onstep build I’ve got:
- 17HM19-2004S (59N.cm, 2.8V & 2A) steppers. These are 0.9degree per full step.
- TMC2209 drivers. These are nice smooth drivers that can handle more current than the TMC2130s.
- I’ve got MXL pulleys & belts – 15 tooth (5mm bore for the stepper motors) & 48 tooth (12mm bore for the worm drive shaft) pulleys. I’m using neoprene belts with Kevlar reinforcement. I’ve also ordered 60 tooth pulleys, but I’m concerned that might not fit inside the mount. There’s not much space inside the housing (first pic – shows the original cog on the left and the 48T pulley installed on the right).

Using 32 microsteps with the CGEM 180 tooth, the 15-48T (3.2 gear ratio) combination gives a 0.176 arc-sec per microstep resolution (the 15-60T combination would give 0.141 arc-sec).

To mount the stepper motors I’m starting with 3D printed brackets and connecting it externally to an onstep controller (second pic). If it works I will eventually make the stepper mounting plates out of aluminium and install the controller PCB inside the mount, attached to the cover plate.

[ChrisV]

With the 3D printed brackets the 15T pulley on the stepper lines up with the 48T stepper on the worm drive shaft. One thing, the 48T pullies are wider than the original cogs and rub against the bearings on the worm drive shaft. So they need to be filed down a bit (see first pic). Now there’s a gap to the narrow part of the shaft which sits inside the bearing. I also replaced the bearings with some nice new 608 sealed bearings.

Next will be the first test run.

[Sunfish]

Looks excellent. The choice of Kevlar reinforced belts sounds like a good thing.

A big step to to take with your main mount. Very interesting to see.

A very challenging operation , even with your experience, but looks like you have it all sorted admirably.

I have recently added onstep with wireless connection to an old Celestron fork mount as an experiment and it is tracking reasonably with 32 microsteps on 0.9 degree per step motors. I have not got the pulse guide rate sorted out yet. Luckily with 360 teeth on the wheel , the worm gear can be direct driven as I can not fit any RA belts and pulleys in easily. Not so easy with 180 teeth.

[ChrisV]

After realising that it can actually be done, the most challenging thing was getting everything the correct size - went through a lot of 3D print prototypes. Then a mini USB camera allowed me to look inside and confirm all was lined up correctly.

With the gear ratio of 3.2:1 I didn't have to modify anything on the mount because the pulleys are a reasonable size. Just took out the cogs and the faceplate/PCB/motors - they can be put back in later if I want. I could go to a higher gear ratio but would have to do a bit of filing on the inside of the mount - rather not tackle that.

I'm just finishing off a new PCB that can be attached inside the mount faceplate. There's plenty of room there.

Be interesting to see how you go with the fork. 360 teeth is the same as a G11 and quite a few have gone direct drive with that using 0.9 degree steppers / 32 microsteps.

[ChrisV]

So the setup I used to test was the onstep miniPCBv1 with TMC2209 drivers. I’ve set these to run at 900mA which is about 40% of the current for the 17HM19-2004S stepper. I’ve chosen this as it reduces the midband resonance of the stepper – avoids noise and stalling. I’m running them in spreadCycle during GOTOs (full step mode) and switch to stealthChop (32 microstepping) during tracking. I’ve added a 12-24V DC converter to improve control. On the bench this seemed to run smoothly and quietly with no excessive vibration.

I set up the mount with an 80mm refractor. The mount coped with slews at 3.5 deg/sec. I started on chamaeleon-I, a dark nebula. The attached pic shows the PhD log from that night, after the auto meridian flip, from about 1.30am to 5am. It got a bit hairy at the end but you can see from the star SNR that it got a bit noisy with the onset of twilight. But overall I was pretty happy with a total error of under 0.7arc.sec over 3 hours when dithering every 5th 90sec sub.

So given it works so well, I’m going to make up a new pcb to install inside the mount. That will come soon

[ChrisV]

Have finished the controller PCB and attached to a new 3D printed faceplate.

[Sunfish]

Wow. Looking very good. Great job. I can’t see my onstep working that well on a fork but clearly your upgrade is a great success.

[Sunfish]

I think I can use the TMC2209 drivers on my MKSGENL v1 which may give some advantages in changing mode between slewing and tracking. It would be interesting if a change in guiding pulse rate could also be affected and if this would be an advantage.

[ChrisV]

I don't know about using the 2209 on that board. But I know as I went from the a4988 to tmc2100, 2130 to 2209 control improved greatly

[ChrisV]

If anyone is interested I have a google drive with the instructions for stripping down the Celestron CGEM mount, general maintenance & upgrading it to ONSTEP (https://drive.google.com/drive/folders/1jpYW-6Y1l1IwhpLdF2MUBTwg1_3hzFl3?usp=sha ring)