Belt drive backlash

[alacant]

Does converting -e.g. an eq6 mount- from geared to belt driven reduce backlash?

Thanks

Edited February 26, 2023 by alacant

[teoria_del_big_bang]

If done correctly then it should reduce backlash.
How much it improves it I would say it depends where most of the backlash is coming from.
Assuming the mount is similar to the mechanics of an HEQ5, which is all I have experience with, then it should remove a lot of the backlash from the gears between the motor and  worm gear shaft (basically the 3 gears the timing belts and pulleys replace) but a lot of the backlash will be between the worm gear and the final gear on the DEC or RA shaft.

This can be reduced by very careful adjustment but takes time and patience as too much adjustment and the gears will bind once per rev of the worm and too little and there will be excessive backlash.
You also will have to just reduce the backlash best you can around this worm as machining errors means there will always be some cyclic backlash as through the 360 degrees the worm rotates it will become tighter at one point and will be loose (and hence have backlash) 180 degrees from this tight point. From what I read it is the luck of the draw if you have a good one with little cyclic backlash or a bad one, the tolerances they work to are just not held close enough to eliminate this completely.

That is my understanding anyway and what I found when I belt modified and tuned my HEQ5.

Steve

[vlaiv]

There are two sources of backlash in HEQ5/EQ6 drive mechanism.

First is motor gear part which is replaced by belt mod and second is worm / worm gear engagement that can be tuned out to some extent.

Belt mod removes that first source but does nothing to the second.

Ideally if you want to completely remove backlash - you'd need another mod. You would need to add spring load or magnetic load to worm/worm gear engagement.

With my HEQ5, I can't completely tune out worm/worm gear backlash because worm gear seems to be not quite circular but squished a bit. If I tune it at one point - it will either bind or have some play at 90 degrees to that. This happens both in RA and DEC - but I've found a way around it.

For RA it is not as important if you preload your mount properly (just a bit preload is needed - a bit of east heavy) - that backlash is far from drive gear and guide corrections won't disturb it. 

With DEC, I can't use above trick as it is not constantly moving, but there is another trick. As backlash depends on the part of the circle - I just setup DEC in such way (using clutches and rotating the ota in DEC) so that I'm at good part of the circle without issues for duration of the session (gets tricky when there is meridian flip as DEC turns for 180 degrees - but if error is equal - you should land on good section again).

[Peter Drew]

Still beats me that wormwheels and worms are eccentric in this age of high tech machining.  I don't recall significant issues with gear sets I used to make on very everyday equipment.  I was very careful with each step of the process though.     🙂

[vlaiv]

22 minutes ago, Peter Drew said:

Still beats me that wormwheels and worms are eccentric in this age of high tech machining.  I don't recall significant issues with gear sets I used to make on very everyday equipment.  I was very careful with each step of the process though.     🙂

Maybe worm wheel is made out of very soft material (it usually is because there is large difference in wear and tear between worm and worm wheel) and when it is pressed onto a shaft - it gets bent out of shape because shaft is not very round?

Edit:

maybe they are both round but mounted a bit ex-center for some reason? (Assembly being "low skilled" work?)

Edited February 26, 2023 by vlaiv

[Peter Drew]

Quite likely.  You do get what you pay for.     🙂

[oymd]

3 hours ago, vlaiv said:

There are two sources of backlash in HEQ5/EQ6 drive mechanism.

First is motor gear part which is replaced by belt mod and second is worm / worm gear engagement that can be tuned out to some extent.

Belt mod removes that first source but does nothing to the second.

Ideally if you want to completely remove backlash - you'd need another mod. You would need to add spring load or magnetic load to worm/worm gear engagement.

With my HEQ5, I can't completely tune out worm/worm gear backlash because worm gear seems to be not quite circular but squished a bit. If I tune it at one point - it will either bind or have some play at 90 degrees to that. This happens both in RA and DEC - but I've found a way around it.

For RA it is not as important if you preload your mount properly (just a bit preload is needed - a bit of east heavy) - that backlash is far from drive gear and guide corrections won't disturb it. 

With DEC, I can't use above trick as it is not constantly moving, but there is another trick. As backlash depends on the part of the circle - I just setup DEC in such way (using clutches and rotating the ota in DEC) so that I'm at good part of the circle without issues for duration of the session (gets tricky when there is meridian flip as DEC turns for 180 degrees - but if error is equal - you should land on good section again).

HI Vlaiv

What do you mean by East heavy? And by how much should I be out of balance?

[vlaiv]

1 hour ago, Peter Drew said:

Quite likely.  You do get what you pay for.     🙂

I'm not entirely sure that is right

I think that in some instances you pay much more than thing is worth.

Here is an example - EQ6-R costs as much as 50cc moped (give or take).

Compare by any metric and you will find that 50cc moped is better value for the money - cost of material, cost of manufacturing, advanced materials, precision manufacturing - you name it.

Not to mention things like EQ5 vs EQ5 Pro goto.

2 stepper motors + some plastics and printed circuit board or two (total less than 50 quid) - sold for 400 quid more.

[Peter Drew]

Very true but you pay for the function.  A 50cc moped would carry a telescope but the tracking would be terrible, conversely if you fitted wheels to the EQ6-R you'd never get anywhere at 1  rev per day.        🙂

[vlaiv]

5 minutes ago, oymd said:

HI Vlaiv

What do you mean by East heavy? And by how much should I be out of balance?

Mount moves from east to west as it tracks - and it is supposed to be in balance.

If there is slight imbalance - east side being heavier then the west (and that will depend on where the scope is point - east side of pier or west side of pier if extra weight is on "scope" side or "counterweight side") - it will act against motors moving - it will "eat up" backlash in worm gear.

If mount "overshoots" - because of too aggressive correction or because it's leading in that moment of time (compared to ideal tracking), and mount is perfectly balanced - there will be backlash in the system and pulse to bring it back wont actually bring it back - it will just "eat up" some of backlash.

But if mount is always "leaning" against worm gear because it is not perfectly balanced - any change in RA speed - either forward correction (faster tracking) or backward correction (slower tracking) will instantly take effect. This is not true if setup is west heavy.

in above image in top example - object is "leaning" against the side that is pushing it, so it does not matter if system is slowed down or sped up - it will still be in contact with that side.

In lower part of the diagram - object is "leaning" against opposite side. Here if we slow down - it will stay at that side - but if we speed up - it will start trailing - which is not good.

Just a small imbalance is needed - but you also need to lower guide speed. Too high guide speed will cause too much jerk which is not good. I use x0.25 sidereal.

 

[pipnina]

4 hours ago, Peter Drew said:

Still beats me that wormwheels and worms are eccentric in this age of high tech machining.  I don't recall significant issues with gear sets I used to make on very everyday equipment.  I was very careful with each step of the process though.     🙂

I should imagine it depends a lot on the method used to manufacture. One would expect a lathe to produce very good, concentric and round parts. But if it's cutting too fast, too much pressure on the part, and the part is not supported at both ends, you may end up with the part being produced with an undesirable shape. I think gears are also not easy to dimensionally inspect. Mind I don't have any experience with that aspect of machining myself...

In theory, a larger gear being driven by the worm would be less affected by backlash between the gear and worm, as 1mm of movement at 200mm diameter is half that of 1mm of movement at 100mm diameter driven gear. Could be that, given less-than-perfect materials, the real solution is to make the whole thing bigger? Alas brass is rather expensive, but other materials must be suitable too, like hardened mild steel, if properly lubricated and protected...

I keep thinking about if some way might exist to re-circularise the worm, some kind of lapping process perhaps? Lapping paste instead of lubricant on the worm and driven gear, get the backlash just shy of binding and take her for a whole shaft rotation or two. Then adjust again. Maybe after a few goes around it will be much closer to a snug fit?

I might not be brave enough to try it on my own kit, mind you haha.

[wimvb]

16 hours ago, alacant said:

Does converting -e.g. an eq6 mount- from geared to belt driven reduce backlash?

Thanks

The belt can introduce its own tracking issues. If you analyse a guide curve (phd log viewer, RA frequency analysis is the best tool for this), you may notice several periods (peaks in the frequency curve). The following applies to the AZ-EQ6,  but should be similar for any Skywatcher mount. The peak at about 480 s in the frequency curve is due to the worm assembly. A 120 s peak can be associated with one full turn of the stepping motor, and a 10 s peak with the stepping motor advancing one cog of the belt wheel. If the belt is too loose, this peak gets more pronounced. Otoh, you need guiding exposure not exceeding 1 s to resolve this peak properly. You can minimise all peaks by careful backlash adjustment and careful tensioning of the belt. Phd can handle a moderate amount of (DEC) backlash and worm eccentricity, and even the 120 s period, but the very short 10 s period is almost impossible to guide out.

[alacant]

Thanks everyone. Some useful thoughts.

Disclaimer:  I've been forced into this. I hate looking at graphs and am fed up with pulling mounts apart because the graph "isn't good enough".

But anyway... Here is my analysis of a supposedly misbehaving eq6 mount. Gears only.

Hysteresis:

Now with PPEC at 122s:

We had the dreaded 10s spike on this example which was fixed by moving the driven gear relative to the idler gear a few teeth. That left the 122s PE..

Hesitant conclusion

Using PPEC for RA is I feel a good idea. The belt kits I doubt are going to give more, if any, benefit than the algorithm. Oh, and (note to boss): the cost.

Cheers and thanks again.

Edited February 27, 2023 by alacant