EQM35: Low-Cost Belt Mod

[jif001]

 

For some time now I’ve been having trouble with the RA tracking on my EQM35. I use it for a wide-field setup (Sharpstar 61-EDPHII, 268mm) with an image scale of 2.9”/px, and the RA error was mostly tolerable but often not so. Recently the RA motor stuttered at all speeds other than maximum slew, and to identify the problem I connected a NEMA 17 in its place. It worked perfectly, eliminating the possibility of a connection or communication issue outside the non-maintainable motor itself. It looks like the stock RA motor is shot.  

So, I set about establishing whether the NEMA 17 could replace it. Initially the answer was a flat no given the need to connect to the existing gear train and the fact that it’s a bigger motor, so I pondered the possibility of a DIY belt mod instead, which had the advantage of eliminating some of the tolerance issues that gears present: the RA drive has four gears between the motor and the worm, with three points where gears mesh, each one introducing some kind of periodic error and backlash. To cut an already long story short, not only did the belt mod work but it was a spectacular success: my guiding has gone from a typical 2” RMS (if I was lucky, and I often wasn’t) to 1” RMS or lower during two nights of lengthy testing. On the first night of testing, it was matching my EQ6r Pro in roughly the same area of the sky (one on the Rosette, the other on the Cone). On the second night it was not so good, but still significantly better than what was previously normal for this mount.          

It was straightforward to establish the gear ratio on the RA train between the motor and the worm. There are four gears: the driving gear on the motor shaft (12 teeth); a gear driven by it (48 teeth); a coaxial gear rigidly attached to that (35 teeth); and finally, the driven gear attached to the worm (35 teeth) giving an overall gear ratio of 4:1.

My first effort used 20-tooth and 80-tooth pulleys that I had lying around, connected by a 200mm belt, all held together by a Heath Robinson lash-up (a 3d-printed bracket attached to the weight bar). It wasn’t very elegant, but it worked and proved the concept, so I set about refining the bracket and bought new smaller pulleys and a belt – this time 12 teeth / 48 teeth and a 158mm belt. The photograph shows how it looks now. Still some minor refinements needed, but my EQM35 is performing better than it ever has.

You may notice that the motor and pulleys are operating on the opposite side of the mount from the side the stock motor would be on. Viewed from this position the stock motor would be in the same orientation (drive gear facing you) on the other side of the mount, but conveniently the stub for the manual slow-motion control (that big old dangly spring that you probably don’t even possess) is well placed on this side and is longer than the stub on the other side, allowing more room for the large pulley.

Another convenience is the position of the bolt holes for mounting the old motor, which have been repurposed to mount a new 3d-printed box to house the cable sockets (second picture).

I should add that last month I had the mount in bits and replaced the fibre washers with needle bearings, as in this video -> (69) EQ3 Hyper Tune - YouTube (for an EQ3, but essentially the same mount). I didn’t get much chance before the motor problems to evaluate whether this alone made a significant difference, and the limited imaging time I did get left me unconvinced: it seemed as bad as before. I mention it because it’s possible that the improvement I’m seeing is a combination of the bearing mod and the belt mod. There’s no doubt though that the improvement following the belt mod has been dramatic – much better RA stability and improved guiding.

I know there are commercially available controllers and belt mods for this mount, but they are not far short of £300. Leaving aside the cost of a 3d printer (which has paid for itself handsomely in the two years since I bought it, and continues to do so with every astro-widget I make) this has probably cost me no more than 10% of that.

I’ve improved the RA, but what about the dec? Normally it’s pretty good and RA has been the source of all my issues, except one. Whenever I dither with this mount there’s a good chance my guiding will go off-scale due to dec backlash, and I usually have to stop-and-restart the guiding to recover it or lose one sub-exposure to drift – not great if you want to do unattended imaging. So, that’s the next target for improvement, but it’s a bit trickier to do a belt mod here. The gear ratio on the Dec axis is 11:1 and I won’t be able to get that ratio with two pulleys unless one is the size of a dinner plate. Perhaps the solution is a similar arrangement to the one I now have on RA but with altered dec slewing and guiding rates set in EQMOD, so there may be some trial-and-error involved. Those commercial upgrades no doubt have pulse rates to match their small pulleys, but I can’t alter the stock motor controller. Any suggestions here would be welcomed. For now, I’ve adjusted the worm carrier bolts to reduce the dec backlash but need to wait for the next clear night to test it. If that's improved it I'll leave it alone!

For anyone wanting to give this a go, the parts list is:

·         NEMA 17 motor (4-wire with fitted connector) – a NEMA 14 might do it (I think that’s the stock motor) but haven't tried one so not sure.

·         48 tooth timing pulley – 6mm bore (for a 6mm wide belt)

·         12 tooth timing pulley – 5mm bore (for a 6mm wide belt)

·         158mm timing belt (6mm wide)

·         Custom mounting bracket and socket housing (if you have – or have access to – a 3d printer!)

·         2x M5 (5mm x 15mm) thumbscrews (standard M5 bolts would do if you're not fussed) + 2x M5 nuts

STLs available on request for the 3d-printed bracket and socket housing.

Total cost about £30 (less if you reuse the existing motor)

Edited June 6, 2023 by jif001
Title change

[fwm891]

Take a look at Davall gears web site they do mini timing belts and wheels (I’ve used them for an NEQ 6 belt mod) you can get your tooth counts in a smaller dimension.

[jif001]

Thanks - but they don't seem to provide any detailed product information or pricing. Is there another site that does?

 

Edited April 10, 2023 by jif001

[fwm891]

Try: https://www.bearingboys.co.uk/Timing-Belts-and-Pulleys-1039-c

3M timing belts and pulleys. Some down to 3mm pitch.

Also:

https://www.transdev.co.uk/rubber-belts/imperial-rubber-series/mxl-mini-pitch-series-transflex-classic/

It was the MXL series belts/pulleys I used for the NEQ 6 Pro mods ¼" width although you can now get them in other widths as well.

Edited April 10, 2023 by fwm891

[jif001]

Excellent, thanks 👍🏻

[malc-c]

When I was messing about converting an HEQ5 to belt drive in 2011 (😧)  I used a company called Belting Online for MXL pullies and a 1/4" wide timing belts.  These were the same spec of belts and pulleys that Frances and other had used converting there EQ6's which was the inspiration for doing the same with my HEQ5.

This is a neat design and I can see an opportunity for  the OP to provide these as kits for those who wish to do the same but lack the ability to have the parts 3D printed.  If I read the post correctly the drive uses a 4:1 ratio, so I'm guessing that it needs EQMOD to control it as SW seem to use weird ratios when geared which are near nigh impossible to maintain unless you have access to a CNC machine to make the small motor pulley form one piece of aluminium, which is where Rowan engineering came in and the rest is history.

Shame the DEC ratio is such that a complet kit could be offered... even so, I think the OP has an opportunity to run a side line business selling conversion kits, and even at £50 to cover the costs of 3D printing and his time.  Get in there quick before Rowan does  

[jif001]

Ha! Yes that did cross my mind, and I’m willing to provide kits if anyone is really interested! 

[jif001]

An update … 

Last night I managed about an hour of testing before it clouded over, and it confirmed the above conclusions. I left it running for as long as I could to get a decent guide log (it was quite windy, which wasn’t helping) and managed enough time to record a full worm cycle.

The guiding graph is attached, showing an RMS of 0.74”, which was consistent through the time I had it running (other than when it was wind-affected) and far better than I’ve had before. I dug out a pre-modification guide log and had a look at the frequency analysis plot and as you can see from the comparison, the high-frequency oscillations (left side) have been eliminated and some of the low frequency oscillations have lower amplitudes. Those high-frequency oscillations were a problem, with one of them having an amplitude of over 1” and a period of 10 seconds. PHD2 was constantly chasing that. The one I cannot influence is the largest peak, which is caused by the worm itself. 

Some of the low-frequency oscillations are probably a result of the wheels not being perfectly true. Those two grub screws holding the wheel in place will inevitably push the wheel slightly to one side, meaning its axis does not perfectly align with the worm axis, and avoiding any slight tilt in the larger wheel is also tricky. The effect of these can be seen at maximum slew rate when there is a visible wobble in the wheel. Some tinkering and fine tuning is needed to limit those effects, but the bottom line is the RMS figure and the significant improvement this change has made. 😊

 

Edited April 14, 2023 by jif001

[icefabio]

On 10/04/2023 at 00:41, jif001 said:

It was straightforward to establish the gear ratio on the RA train between the motor and the worm. There are four gears: the driving gear on the motor shaft (12 teeth); an idler gear driven by it (48 teeth); a coaxial gear rigidly attached to that (35 teeth); and finally, the driven gear attached to the worm (35 teeth) giving an overall gear ratio of 4:1.

[...]

The gear ratio on the Dec axis is 11:1 and I won’t be able to get that ratio with two pulleys unless one is the size of a dinner plate. Perhaps the solution is a similar arrangement to the one I now have on RA but with altered dec slewing and guiding rates set in EQMOD, so there may be some trial-and-error involved. Those commercial upgrades no doubt have pulse rates to match their small pulleys, but I can’t alter the stock motor controller. Any suggestions here would be welcomed.

[...]

Hi @jif001,

great job!!

what calculation should be done to find the ratios of the RA and DEC axis?

How did you get to 1:4 and 1:11 ratio?

Many tks

Fabio

 

[jif001]

The RA gear train is as follows:

• 12 teeth on the drive gear (attached to the motor) - call it gear A
• 48 teeth on the first driven gear - gear B
• 35 teeth on the co-axial gear rigidly attached to that - gear C
• 35 teeth on the last driven gear (attached to the worm) - gear D

Gear A drives gear B. Gear C rotates with gear B because they are attached to each other co-axially. Gear C drives gear D. So:

• 1 rotation of gear D requires 1 rotation of gear C (both with 35 teeth).
• 1 rotation of gear C requires 1 rotation of gear B, since B and C are co-axial
• 1 rotation of gear B (48 teeth) requires 4 rotations of gear A (12 teeth)

From all of that, 1 rotation of the worm (gear D) requires 4 rotations of gear A, hence 1:4

For dec the gear train is:

• 12 teeth on gear A (attached to the motor)
• 66 teeth on gear B
• 35 teeth on gear C (co-axial with, and attached to, gear B.)
• 70 teeth on gear D (attached to the worm)

As for RA, Gear A drives gear B. Gear C rotates with gear B because they are attached to each other co-axially. Gear C drives gear D.  

• 1 rotation of gear D requires 2 rotations of gear C (70:35)
• 2 rotations of gear C requires 2 rotations of gear B (co-axial)
• 2 rotations of gear B requires 11 rotations of gear A (66:12)

So, 1 rotation of the worm (gear D) requires 11 rotations of gear A, hence 1:11

 

Edited May 11, 2023 by jif001

[icefabio]

@jif001 many thanks for the explanation. Now I understand where I made a mistake!!  

[jif001]

Another update.

Dec is now belt-modded following a Eureka! moment. This is how it went ...

The Dec axis gear ratio on an EQM35 is 11:1. The smallest pulley I have is a 12-tooth and can’t get one much smaller, but to get an 11:1 ratio with that I would need 12 x 11 = 132 teeth on the larger pulley and, as stated earlier, it would be the size of a dinner plate and would clash with other parts of the mount (if I could even source one).

Then I had an idea - the stock stepper motor is a 1.8 degree-per-step type ... could I use a 0.9 degree type? The drive is based on pulses from the motor controller, which has no idea what type of motor is connected, so a 0.9 degree motor will turn at half the rate of a 1.8 degree motor (200 ‘steps’ sent to a 1.8 degree motor would result in 1.8 x 200 = 360 degrees of rotation - i.e. one full turn - whereas the same steps sent to a 0.9 degree motor would result in 0.9 x 200 = 180 degrees of rotation - half a turn in the same time). That would mean I would need a gear ratio of only 5.5:1, so in effect I would be halving the motor speed but doubling the rate again through the lower gear ratio with a smaller pulley, overall resulting in the same rate as the original 11:1 gear train. That means I can use 12-tooth and 66-tooth pulleys (5.5:1), or a 10 and a 55 if I could find them. Both are much more reasonable propositions.

Finding pulleys of the right size has been difficult. I eventually found some in China via Ebay and Amazon, and right now I'm a week into the one-month delivery time! I have a 12 tooth pulley so I 3d-printed a 66 and after some trial and error managed to get it to grip the worm shaft without slipping. In tests over two nights I got guiding with an RMS below 1" and for this mount and this image scale it's perfectly fine. Attached are some pics, including the new Dec setup and about 90 minutes of H-alpha data from the Sadr region of Cygnus to demonstrate the performance. 

As expected it didn't (it can't) completely cure the Dec backlash. The worm and worm gear are what they are and a belt won't have any effect on them. All I can do there is try to optimise the meshing. Nevertheless, I can at least be satisfied that the additional backlash that was in the rest of the Dec gear train (3 meshed gears) has been eliminated.

Hopefully a proper metal pulley will make a slight improvement, and I have some work to do to fine tune it all. If there is anything new to report I'll add it here.

Edited June 17, 2023 by jif001

[jif001]

Parts list for the whole job:

RA Axis

NEMA 17 1.8 degree motor. You could re-use the existing motor though (I replaced mine because it had failed) 

12 tooth GT2 timing pulley – 5mm bore (for a 6mm wide belt)

48 tooth GT2 timing pulley – 6mm bore (for a 6mm wide belt)

158mm GT2 timing belt (6mm wide)

Custom mounting bracket and socket housing

2x M5 (5mm x 15mm) thumbscrews (standard M5 bolts would do if you're not fussed) + 2x M5 nuts + the existing bolts used to attach the motor (add some washers)

Dec Axis

NEMA 17 0.9 degree motor (a NEMA 14 might do if you can get a 0.9 degree version, but  the stock motor needs to be replaced here. Keep it as a spare in case your RA motor ever fails)

12 tooth GT2 timing pulley with 5mm bore (alternatively, 10 tooth - but not yet tried)

66 GT2 tooth timing pulley with 6mm bore (or 55 tooth if using 10 above)

200mm GT2 belt 6mm wide (or 196mm)

custom motor and socket housing 

1x M5 nut + the existing bolts used to attach the motor (add some washers). 

 

I’m happy to supply STLs on request for the 3d-printed brackets and socket housings if you have – or have access to – a 3d printer.

 

 

 

Edited June 6, 2023 by jif001

[malc-c]

What electronics are you using to drive the system?  -   I presume if its still using the Synscan box then you have set custom ratios up in EQMOD to compensate for the ratios hard coded into the mounts  motorboard firmware?

[jif001]

I’m using the standard EQM35 motor controller with no changes made to EQMOD settings.

The aim of the exercise was to achieve the same ratios as the hard coded ones because I recognised at the start that there was nothing I could do to change them. On the RA axis I did it with 48 & 12 tooth pulleys (ratio 4:1, as in the original gear train). On the Dec axis I used  66 & 12 tooth pulleys, but that ratio is 5.5:1 and would run at twice the required speed because the original ratio is 11:1, so I halved the speed with a 0.9 degree motor in place of the 1.8 degree motor. I chose 66 & 12 (i.e. 5.5:1) specifically because I could halve the motor speed. I’ve also got a 55 and a 10 on the way from China (and a 66, to go with the 12, just in case…)

The upshot is that the pulse rates going to both axes drive them at the correct speed, with no software tweaks needed. 

Edited June 6, 2023 by jif001

[malc-c]

Right I follow what you were doing.   EQMOD has the ability to enter offsets etc which was useful when I was messing about doing the original HEQ5 belt mod, as I had to use off the shelf parts which resulted in a 4:1 ratio.  

[jif001]

I did think of that as an option for the Dec axis when I was unable to match the 11:1 ratio, but the motor idea meant I didn’t need to the end. 👍🏻

[icefabio]

On 06/06/2023 at 03:27, jif001 said:

Dec is now belt-modded following a Eureka! moment. This is how it went ...

@jif001 great job and great news!!

You are the only one who succeeded in this hack!  Everyone stopped at just changing the AR axis.

Thanks for sharing this simple and genious solution.

F

PS I recently bought EQStarPro otherwise I would have made this change too!

[jif001]

44 minutes ago, icefabio said:

@jif001 great job and great news!!

You are the only one who succeeded in this hack!  Everyone stopped at just changing the AR axis.

Thanks for sharing this simple and genious solution.

F

PS I recently bought EQStarPro otherwise I would have made this change too!

Thanks for the feedback 👍🏻 I’d be interested to read any other threads on this subject if you can point me at them … there might be useful tweaks there!

[icefabio]

2 hours ago, jif001 said:

Thanks for the feedback 👍🏻 I’d be interested to read any other threads on this subject if you can point me at them … there might be useful tweaks there!

This is a very detailed one: https://www.webastro.net/forums/topic/181159-test-eqm-35-pro/

F

[jif001]

On 12/06/2023 at 12:02, icefabio said:

This is a very detailed one: https://www.webastro.net/forums/topic/181159-test-eqm-35-pro/

F

That was very interesting - thank you. It revealed the fact that choosing 'Auto Detect' in the EQMOD Mount Options was the wrong thing to do! The thread gives figures for microsteps per axis revolution for RA and Dec, and BOTH were incorrect in my EQMOD settings! I confirmed this by doing the calculations myself, and the results agree with those in the thread you quoted. Another lesson learned - don't trust the Skywatcher motor controller to report the correct parameters to EQMOD! I know this mount shipped with the wrong parameters in the firmware, but I assumed this was fixed by updating it....

 

EDIT: after testing the settings below I discovered they did not work. NINA 3PPA started by pointing the scope at the ground (or it would have, if I hadn't stopped it). Resetting to the original values fixed it. So, don't use the numbers below. I'll have to try to figure out WHY they didn't work!

 

For reference (and to make this whole thread a little more comprehensive) the correct numbers are as follows for the standard mount (note: numbers are for the EQM35 only, but the method applies to any similar mount):

                                                             RA:               Dec:
       A: motor degrees per step            1.8                1.8             

... this is a characteristic of the motor

       B: motor steps per motor rev       200               200              B = 360degrees/A
      C: Micro-steps per step                  64                 64             

... this is a characteristic of the motor controller and the motor (each 1.8 degree step is made up of 64 micro-steps)

       D: micro-steps per motor rev     12800           12800            D = C x B
       E: gear ratio                                   4:1               11:1             

... this is a characteristic of the gearing of the mount

       F: micro-steps per worm rev      51200         140800           F = E x D
       G: worm revs per axis rev              180               65               

... this is a characteristic of the mount (basically, the no. of teeth on the worm gear)

       H: micro-steps per axis rev       9216000      9152000         H = G x F
                    
 F and H (in red)  are the new values to be set in the Custom Mount dialog (open EQASCOM Toolbox / press the Driver Setup button / tick the Show Advanced Options box / select Custom from the drop-down in Mount Options / press the wrench button). There is a field labelled 'Tracking Offset' there too, but right now I don't know what to do with that so have left it at zero (Google isn't helping).

Using the belt mods described in this thread none of the RA numbers change but for Dec A becomes 0.9 degrees and B becomes 400 steps as a result. The gear ratio changes to 5.5:1 also. These changes have no effect on the above values of F and H because, basically, all I did there was divide by two in A and multiply by two in E.

I've yet to try these new numbers in the setup, so will report back here if I learn something worth knowing!

 

Edited June 13, 2023 by jif001

[malc-c]

With the mount connected and powered up, run the attached program.  Select the com port and then connect - It sends native commands (as if being a terminal) to the motor board which will provide you with the values set in the firmware

SyntaTester.zip

[jif001]

6 hours ago, malc-c said:

With the mount connected and powered up, run the attached program.  Select the com port and then connect - It sends native commands (as if being a terminal) to the motor board which will provide you with the values set in the firmware

SyntaTester.zip 721.35 kB · 1 download

Thanks very much for that. Do you know of any documentation explaining its use? 

I've been trawling various forums over the past few days trying to figure out how this mount works 'under the hood' and this looks like it will provide the answers. As a minimum, it throws light on what was previously incomprehensible terminology used in some posts I read in the EQMOD forum (e.g. the meaning of :a and :b and :I ...).

These are the numbers it reports:

... and these correspond with the 'Autodetect' numbers I've seen. They don't, though, match the numbers I have quoted in my previous post. The RA gearbox ratio in the image above is 47:12 (not quite 4:1) even though I have counted the physical teeth on each gear wheel and confirmed the ratio is exactly 4:1. The same numbers are used for Dec (so 47:12) even though, once again, I have determined by inspection a ratio of 11:1.

On the 'Data' tab (presumably standard numbers for numerous mounts including EQ6, EQ8, etc.) it reports this for the EQM35:

... and this agrees with my numbers (see the red numbers for RA above).  

It's an interesting conundrum. Maybe the firmware is not correct, but it works  for imaging so at least I know I can fall back on the defaults ('Autodetect Mount') when the sky is clear and tinker with the numbers in the day!

[malc-c]

4 hours ago, jif001 said:

Thanks very much for that. Do you know of any documentation explaining its use? 

 

I don't have anymore than the standalone exe.  I can't recall how I come by it.  I recall a link on a  forum which discussed reverse engineering the boards..  I I can find that I'll post the details 

 

[jif001]

It seems to be mentioned a lot in a Green Swamp Server forum. I think the author has a presence there.