The aid of an engineering brain required please

[Horwig]

Would anybody care to give some thought to my conundrum, it's driving me up the wall.

I have been building my fork mounted scope for longer than I care to remember. I recently sorted the scope itself out with a new focuser and found the source of the annoying astigmatism, now it's the turn of the mount to give me problems, and I just can't get to the bottom of it.

My problem is backlash in the DEC axis, nothing out of the ordinary there I hear you say, problem is, when I reverse DEC direction at guide speed, the scope continues to move in the original direction before reversing, not a huge distance, about 30 arc seconds, but it's enough to make any kind of guiding in DEC virtually impossible.

I've tested the DEC motor and drive belt to the worm, they are fine, so it must be flex somewhere, but where? There is no similar problem in RA, that is right as rain.

The fork is heavily constructed from thick gauge ali, in fact I've gone for over engineering wherever possible, and yes, all bolts have been checked for tightness.

Here is the original thread of the build:

I know it's a needle in a haystack, but any input would be welcome.

Huw

[Louis D]

Switch to a harmonic drive and live with ripple instead of backlash?

[Gina]

Without having the hardware in front of me I can't think of anything.  Sorry.

[sloz1664]

Were the axes balanced when you did your tests, have you have implied "it's over engineered" That could have the effects you are experiencing if they are out of balance. I would have looked at eliminating the worm  on both axes and going direct belt drive.

Steve

[Oddsocks]

Without being able to see the mount first hand I would hazard a guess it would be down to moment of inertia.

The moment of inertia on the DEC axis is many magnitudes larger than that of the RA axis so that one possibility is belt stretch or mount flex.

It might be that swapping to kevlar or steel cored timing belts will be sufficient to stiffen up the DEC drive, assuming you are using polyester cored belts at present?

Are the belts really the right size for the gears? The flat face of belts teeth should not 'bottom' on the motor's gear while still leaving room for individual teeth to slide from front-to-back. Edge faces of the belt teeth need to be firmly pressed up against the "walls' of each motor gear tooth as it meshes with the gear, else the belt can slip a little in the direction of rotation when the direction of rotation changes.

Aluminium gears are notorious for belt slip as aluminium wears off the gear with use and leaves a slippery coating of powdered aluminium on the surface of the gear and the belt. If the belt is not a snug mesh with the gear then when the motor changes direction the belt can slip a little within the confines of each individual belt-gear tooth mesh.

Another possibility is, if using second hand drives, that there is some wear in the motor reduction gearbox of the DEC drive, if new drives, possibly there is some mis-adjustment of the gearbox and it needs shimming.

Check how the motor pinion gear and the worm gear are attached to the shafts. Is it pinned with an expanding dowel, solid dowel, keyed or simply held with a grub screw? Pinion gear fixings that use a simple grub screw against a machined flat on the motor output shaft can allow the pinion to move a little from side to side as direction of rotation changes, a solid dowel or expanding dowel drilled right through the gears hub and the motor output shaft would be a stronger fixing less likely to slippage.

If the drives are identical try swapping the drives over and see if the problem moves with them.

If the motor control system uses sophisticated DC braking or pulse reverse braking before driving continuously in the reverse direction (or some other method of stopping the armature quickly), is the control system applying this to the RA and the DEC motors?

Check for end float in the DEC worm bearings, there should be some method for adding preload to the worm so that the worm can not move axially as the direction of forces change when swapping from forward to reverse drive, you may need to add or remove shim washers between the end of the DEC worm thread and the face of the worm bearings, or the bearing mounts, so that the worm can not move axially. Similarly, check for DEC worm mesh adjustment. As the direction of forces change when you swap rotation direction the DEC worm must not be able to lift away from the DEC gear otherwise the mount will just keep travelling in the same direction for a while as the slack is taken up.

Thats about all I can think of 'in the dark'

William.

[Horwig]

Thanks for a full and detailed answer Oddsocks.

Firstly, belts and motor, I initially suspected the motor control system, do clamped a one meter long rod to the motor pinion and watched what that did on direction reverse. apart from the dead zone of backlash it was normal, similarly for the worm shaft. But thinking now, that was done with the worm assembly removed, so off load, will try on load tomorrow.

The worm endfloat was addressed yesterday, does not seem to be any play there, but again, I should test under load.

I had come to suspect the spring loaded worm to wheel pressure, just come back from the obsy where I had replaced the spring with a threaded bolt to tighten the mesh, that had no effect.

Next step will be to remove the protective cover off the DEC drive assembly so I can get a dti in there to see what's moving, my guts tell me it has to be in the drive somewhere.

Huw

[Gina]

My DIY fork mount uses all belt drive - no worms, but is just for astro camera and SLR lens for widefield so no ultra-strong over engineering required.  Having thought about it for a while, I too think inertia could be a problem, particularly with a heavily over-engineered construction.

[Peter Drew]

It's a bit like ringing the AA and saying your car won't start, so many reasons why. Difficult to diagnose from a distance. What, if any, means for clutching or damping are provided for the declination movement?

[Horwig]

Thanks for all your replies,

10 hours ago, Peter Drew said:

It's a bit like ringing the AA and saying your car won't start, so many reasons why. Difficult to diagnose from a distance. What, if any, means for clutching or damping are provided for the declination movement?

Yes, sorry, I know I'm asking for remote diagnosis, but I was struggling to focus on fault finding.

Interestingly, this morning I used a dti to monitor the mount whilst moving the axes, and guess what, it happens on both, but it's only really a problem on DEC. So it looks more as if it's a design error not flex.

Huw

[Horwig]

More time playing with the mount. Yes, there is a slight reversal of motion in the RA, but it is tiny, a couple of microns on the edge of the fork, so nothing at all to worry about.

The problem with the DEC appears to be linked to the meshing of the worm and wheel. The rotation of the worm axle seems regular and smooth, and  there is no endfloat I can measure.

Now I have replaced the spring adjuster with a solid one, I have been playing with mesh tightness. I can change the backlash and direction reversal, but not remove it. Also, compared to moving the RA at guide speed (1/3 siderial), the DEC is very 'jumpy' not regular at all.

Time to strip and clean everything down and look at them on the bench I think.

It might be the basic design of the worm block, it's hinged, and swung up to the wheel, but that's how the RA works as well.

This is the culprit during its build, it was modified since these photos were taken, the hinge, (the horizontal block coming out from the fork tine, and fixed to the worm carrier by two black socket headed bolts) has been widened to the whole width of the worm carrier. The spring is out of sight under the assembly, running parallel to the hinge.

Is this the best way of doing it with reasonable engineering skills?

Huw

[Doc]

Just out of curiosity has the worm and worm wheel been lapped together as if not then errors can occur.

[Peter Drew]

Is the drive belt tight enough?, the DEC belt on our 16" SCT was not and the gear "jumped" in action. Tightening the belt sufficiently cured the problem.

[Horwig]

15 minutes ago, Doc said:

Just out of curiosity has the worm and worm wheel been lapped together as if not then errors can occur.

The worm and wheel were bought new from Beacon Hill for this mount, I assume he does lap them, but true, that would cause what I'm seeing. Will clean and inspect them tomorrow.

12 minutes ago, Peter Drew said:

Is the drive belt tight enough?, the DEC belt on our 16" SCT was not and the gear "jumped" in action. Tightening the belt sufficiently cured the problem.

The drive belt is as tight as I can make it, I tried clamping a one meter rod to the end of the worm axle, think giant clock hand, that turned smoothly at guide speed.

What strikes me is that the worm might not be accurately perpendicular to the wheel, more measuring to be done.

Thanks for all the avenues for investigation.

Huw

[Stub Mandrel]

If it is actually carrying on moving in the wrong direction before reversing it must be a combination of backlash or flex in the drive mountings and momentum. You can't reduce the mass of such a large setup.

I suspect you are testing by driving DEC one way, then reversing it . The momentum to overcome is proportional to velocity. At guiding speeds where the motor may move a few steps in a second  the momentum will be much less. Try testing at such low speeds and you will hopefully see less overshoot.

With so much mass it is possible that overcoming the inertia of the system combined with 'stiction' and flex in the drive may contribute to measured backlash as much as any mis-alignment of the gears. My worry would be that in trying to overcome backlash the gears end up being adjusted too tightly, leading to excessive wear.

A screw-jack/jacking screw under the worm carrier would allow positive adjustment of the worm position.

Check for end float in the worm and play in all linkages.

Use the DTI to see if the worm carrier moves out of engagement when you reverse, and if this is smaller when moving at guiding speed.

[Oddsocks]

16 hours ago, Horwig said:

What strikes me is that the worm might not be accurately perpendicular to the wheel, more measuring to be done.

Looking at the wear pattern in the picture of the gear, and assuming it is not just an image artefact, it appears that the worm thread is bottoming in the gear thread and that the worm is not centred to the gear. The mesh wear marks in the photograph of the gear are not central to the width of the gear so that the worm and gear are not meshing at the deepest point.

If this is the case when you tighten up the worm-gear mesh all that is happening is the peaks of the worm teeth are pushing against the valley of the gear teeth while still leaving a large gap between the walls of the teeth for backlash to occur.

Time to get out the engineers blue and check where the mesh points between worm and gear really are....

 

[Stub Mandrel]

1 hour ago, Oddsocks said:

it appears that the worm thread is bottoming in the gear thread and that the worm is not centred to the gear

Well spotted :-)

[Horwig]

10 hours ago, Stub Mandrel said:

The momentum to overcome is proportional to velocity. At guiding speeds where the motor may move a few steps in a second  the momentum will be much less.

Trouble is, it's at guiding speed (1/3 siderial) I see the problem.

9 hours ago, Oddsocks said:

Looking at the wear pattern in the picture of the gear, and assuming it is not just an image artefact, it appears that the worm thread is bottoming in the gear thread and that the worm is not centred to the gear.

That picture was taken soon after the gear was assembled, I hope it's an artefact of the image.

 

Thank you both for your input, at times like this it's nice to have someone looking over your shoulder.

This evening I stripped everything out, washed the worm and wheel in petrol to remove grease, and re assembled everything dry without the covers, and guess what, the worm was not true to the wheel, and not by a little, lets put it this way, I didn't need a micrometer to see the problem, mk 1 eyeball was enough. Very embarrassed.

Quite a bit or re jigging to be done, hope it's what causing the problem, it sounds about right.

Huw

 

[Horwig]

I'm going to say this very quietly, I think it's sorted!

Sprung worm carrier re-worked to bring everything into alignment, then two hours spent with toothpaste on the gear, don't know if it will do any good, but it won't harm it I hope, but what a mess.

With the covers off I could hear the motor better, and it was occasionally making strange ticking sounds, and the backlash was variable, luckily I had a spare gearbox, so swapped them and the ticking went away,

Direction reversal problem seems to have vanished, and the backlash in both axes is now about the same figure, which Sitech removes easily, waiting for dark so I can try some guiding tests.

Thanks all for the suggestions, in the end it was a bit of everything.

Huw

[Oddsocks]

Were the mods successful Huw?

I see you have opened a new thread asking about opinions of the EQ8 as a possible replacement so I’m guessing the star test showed DEC problems have not been solved.

Having reviewed both this thread and the original build thread the only design weakness that stands out is long hinge arm for the worm carrier. For the hinge to allow the worm to swing in and out of mesh must mean there is some lateral movement in the hinge and a very small lateral movement in the hinge would translate to a large backlash at the wheel.

A possible solution would be to drill a pair of tapped holes in the mount frame, either side of worm carrier. After adjusting the worm mesh use a pair of bolts in the newly tapped holes to press lightly against the sides of the worm carrier and this will prevent any lateral movement of the worm carrier, in theory this would remove any backlash contributed by movement in the hinge.

The bolt face should be machined slightly domed where it presses against the worm carrirer and greased so that the carrier can still lift and fall in response to mesh variations.

If this were sucessful then the domed bolt could be replaced with a hollow nosed bolt fitted with a single ball to give reduced friction and wear.

I’ll add a sketch to the post later this morning when I get a minute or two spare.

Given the large mass and moment of inertia that the heavy mount and OTA generates on the worm carrier the spring used to press the worm into mesh would need to be very strong to prevent the worm being pushed away from the gear during drive reversals. Increasing the spring tension would increase wear and drag considerably but a possible solution is to add backstop adjuster screw to the worm carrier so that the principle adjustment and mesh pressure is still taken by the spring but once the spring tension is set then a backstop screw is bought to lightly bear against the worm carrier to limit lift of the carrier during drive reversals. Obviously, a backstop only works if the gear is perfectly centred to the axle and the carrier does not lift and fall by very much during a normal rotation of the axis.

In the assembly of the dec axis I notice there are only two mounting bolts and spacers used to fix the hub axle to the frame, this may allow a little twisting of the gear as the direction of rotation changes, I would add a third spacer and bolt to stiffen up the mounting of the axle but I doubt this is where the backlash is coming from.

William.

[Peter Drew]

+1 for an adjustable backstop to at least minimise the problem. I've tried most meshing designs and a backstop seems to be the best solution. 

[Davey-T]

Bit of a major redesign but would the worm gear be better pivoting parallel to the worm wheel ? it could then be centred on the wheel and pivoting would not move it off centre.

Dave

 

[Horwig]

Thanks both. The request for info on the EQ8 was a sign of desperation this morning, guiding tests did not go well, here are some results, some were even worse:

The mount if fairly well polar aligned, so with DEC guiding disabled it would giving a nice slowly diverging line, but once guiding is re enabled it's either oscillating or diverging off the scale.

Backlash on both axes is now controllable, and the reverse direction problem has indeed gone, but watching a dti on a rod from the end of the OTA as the scope is moved at guiding speed in dec still shows jerkiness of a cyclic nature, a steel rule attached to the worm axle also shows the same pattern jerkiness.

This morning I removed the scope from the mount, and ran the same tests, everything was smooth and regular, so it's obviously load induced.

Oddsocks, you spotted the flange bearing with only two screws, that was only to test fit the drive components, it runs with all four bolts.

The spring on the worm block is valve spring from a Rover V8 engine, it's no wimp.

Yes indeed, I think it's time to beef up the worm adjuster, but if that doesn't work, this mount will be facing an uncertain future, I'm getting tired of fighting it.

On a different note, I'm using good old fashioned ST4 guiding, for some reason Ascom pulse has stopped working altogether on this system.

Huw

Edit: Sorry dave, my post took so long to write, yours came in whilst I was writing, yes, in theory that is a much better solution, that's how Mathis Instruments mount their worm blocks, what worries me about it is the possibility of play in the pivot causing motion in and out of the page (does that make sense)

[Davey-T]

32 minutes ago, Horwig said:

The spring on the worm block is valve spring from a Rover V8 engine, it's no wimp.

From messing around with various mounts over the years trying to improve them I discovered that the worm tensioning spring can be too strong and counter intuitively it worked better with a "weaker" spring but the number of turns and strength made a lot of difference so a weaker spring with more coils worked better.

Dave

[Davey-T]

1 hour ago, Horwig said:

what worries me about it is the possibility of play in the pivot causing motion in and out of the page (does that make sense)

Something like this with the pivot mounted in square tube with chunky bearings (excuse the dodgy drawing) I think the spring tension would need to be adjustable and possibly a pair of springs would work better.

Dave

[Stub Mandrel]

Looking at the graphs, as DEC approaches the line and the correcting movement stops, it appears to 'recoil' upwards slightly. This is seen as a reversal and is over-corrected swinging the mount past the line. Exactly the same seems to happen as it then approaches the line for the other direction.

I think the heavy but balanced OTA is wobbling like a slow pendulum, possibly with a period of several guide pulses.

I suggest increasing the exposure time to allow the OTS to settle, consider reducing the aggression and reducing the backlash compensation.

It may also help to unbalance the OT slightly to or use single-direction DEC guiding.

This should greatly reduce the tendency to have the big overshoots.

It may also be worth adding a tunable mass damper at 90 degrees to the OTA to defeat that resonance.