Damn DEC Backlash!!!

[Horwig]

This is really a followup to a thread I started last year:

 

Firstly, a big thank-you to everybody that helped me out then, the mount started behaving for the first time.

HOWEVER...

I tried to image on a few of night last winter and got pants guiding, so I sulked and left the problem for another day.

That other day arrived this week.

The mount is stripped out again, and a problem was found. The motor shafts are locked to the timing pulleys with grub screws against flats on the shafts, the DEC grub screw was loose. Will replace with a split spring pin, hopefully problem solved.

Trouble is, I can't leave well alone. The new sprung worm blocks work well enough, with PHD2 sorting out the (nice predictable) backlash, but it would be nice to mechanically minimise if at all possible. Upshot is I've gone back to measure DEC backlash, and I can't get it under about 30 arc seconds whatever I do. (I measure it by using a dial gauge on a long metal rod clamped to the DEC axis, Sitech is set at 15 arc seconds per second DEC speed, there's 2 seconds dead time, seems to agree with 30 arc seconds). The DEC worm wheel has been re-checked to be square and centered to the worm.

I have two obvious adjustments, the setscrew to set clearance, and the spring plunger to set worm to wheel pressure. I've checked the worm block hinge, it's free to move, and has absolutely minimal play, the backstop setscrew is set at 50 micron clearance, and I've experimented with spring plunger tightness over its full range. Nothing I do seems to change the 30 arc second dead zone, minimising the backstop clearance shows no improvement, upping the spring force to max makes it worse if anything. Grabbing the DEC axis trunion reveals no slop, but I still get the 2 seconds backlash delay.  The only change I've done that improves matters was to change from thick lithium grease to a thin aerosol white grease, but the change was minimal.

I've tried attaching a rod to the end of the worm as before, there's no dead zone there, it must be just the worm/wheel interface. Could it be that they just don't mesh?

HELP!

 

 

Huw

[Oddsocks]

What type of bearings are used on the worm shaft and worm block carriers Huw, plain ball or tapered roller with pre-load?

In the original thread you had some engineers blue paste to check that worm-gear mesh was on the valley sides of the gear and not the valley floor, did you ever confirm with the blue paste that the worm-gear mesh was ok and not bottoming?

My money would be on combined end and side float on the worm carrier and worm shaft bearings as a minuscule amount of both end and side float on both bearing sets working together with a worm gear translates to a large combined end play, or backlash under load so with a worm-gear lift, twist and end float in any of the worm carrier or worm shaft bearings will contribute to backlash.

Having begun using a new mini-lathe earlier this year I found a surprising amount of end-play at the chuck with the standard plain ball bearings in the head stock, even with a large pre-load applied, and changing those to quality tapered rollers eliminated that.

Of course, the magnitude of the loads involved between a lathe and a telescope mount are somewhat different but the principle is sound and if you are currently using plain ball bearings on the worm block carriers and worm shafts I think that would be my first place for a change to quality tapered rollers with a good bit of pre-load applied and see if that made a difference.

Can’t think of anything else ATM but will come back to this if I do.

William.

Edited May 1, 2019 by Oddsocks
Typo

[tonyowens_uk]

Huw I looked over your previous thread on your R&D work on worm gearing so I think I've got a picture of your drive designs. Having refined some of the rough edges that were previous there I'd make a few suggestions for taking things further:

1. As you've discovered there are a great many variables involved in getting instrument gearing to run well. So focus on one problem at a time. Work on one drive at a time. Remove as many variables as you can from that drive before making changes so you can clearly see which are important and which are not. So in that vein get rid of heavy grease lube and use a semi-grease like engine assembly lube on the worm mesh.

2. You need to be able to measure the effects of your changes. As you know for imaging use (backlash less than 500 millisecs/7.5 arc-seconds) the tangential tooth clearance on worm gearing is only a few microns. That is not reliably measurable using most DTI's let alone lever arms, 'feel' etc. So either use the sky and PHD2 for measurements or buy/borrow a laser displacement sensor and do the work on the bench. If you can get a 1/4" bending beam torque wrench or a Tohnichi gauge in the correct torque range that will help you set bearing preload on the worm shaft in a scientific way.

3. The gold standard in instrument adjustments is use of flexures. Try not to use bearings where possible. The conical tipped preloaded setscrews may work for a while (subject to differential thermal expansion and wear) but flexures are the way to go. I can help with standard designs for doing different adjustments and as you have a mill and are not aesthetically obsessive you should be able to make anything needed!

4. Remove the payload and balance the mount as well as possible before performing mesh setting.

5. Your RA and in particular Dec drives will have fairly narrow torque limits within which they exhibit good tracking and backlash control behaviour. If there is no 'design' as such in your drives and you dont have these limits, I'd suggest you choose conservative values like 5 Nm for RA and 2 Nm for Dec and balance and operate your mount to within those limits. It should be possible to jury-rig means of assessing your out of balance in each axis. 

6. Remove as much friction from the main bearings as possible. If you are using a pair of simple ball bearing pillow blocks on each axis ensure they are properly aligned and preloaded enough to remove clearance. but not enough to raise starting torque to more than 1 Nm or so.

7. Your wormshaft bearings ideally should be angular contact bearings and must be square to each other and axially preloaded. Again use starting torque as the criterion for correct preload. But here a much lower figure in the range 0.-02 - 0.1 Nm should be used, depending on the type of bearings and the detailed design of your mount.

8. Dont use Rep-rap commercial-grade timing pulleys and belts if you can avoid it. Use quality kit. Try Transmission Developments in Poole or Misumi for quality kit. For your mount Gates GT3 type in 9 or 15mm width would be about right. Only one pulley should be flanged not both. The belt needs to be tensioned significantly to work properly but don't overload or obsess about this.

9. All instrument gearing for imaging needs to be lapped then burnished. Never use aggressive grinding media for this, especially with a soft metal wormwheel unless you want to turn it into a permanent grinding wheel! The abrasive gets embedded into the soft surface and can be impossible to remove. Aluminium gears should be hardcoat anodised to BS 2536:1995 to 20 um thickness. Burnishing should be done with oil or light grease, minimal torque load, low speed and lots of direction reversals. 

10. As you've discovered the mesh preload used should be as low as possible especially in thin wormwheels to minimise nonlinearity from wheel deflection out of its own plane. Astro imaging is desperately sensitive to the slightest imperfections and flexure and friction enforce very conservative torque limits.

11. Keep drive adjustments independent of each other. Your mesh clearance adjuster for example imposes a parasitic lateral displacement of worm to wheel when it makes a radial clearance adjustment. This changes the mesh line. If the worm radial runout is significant there will be a visible nonlinearity from that effect for example

12. You use a gearbox on your servomotor. Look hard at that for quality standard. Most such things don't come up to the mark for transmission linearity and backlash. Backlash can be measured directly down to a degree or so using hand tools and a protractor on the bench. Anything more can be measured by using the Guiding Assistant feature in PHD2, with at least 40 minutes of 1 second guide camera exposures. Doing an FFT on this data will provide a very revealing analysis about the mechanics in your drive, including the gearbox.

Have a think about these points if you want to improve your tracking capabilities. I can give you a hand with some of this if you want to contact me off-list.

 

Tony Owens

Edited May 1, 2019 by tonyowens_uk

[Horwig]

Thanks for taking the time to give such full answers both, lots to consider.

Firstly, the mount is stripped back, the OTA removed, and the fork removed from the polar axis for ease of measurement. So all tests are done off load.

Initial results this morning. The worm bearings are plain ball, but preloaded. Measurement just now shows end float of about 20 microns when changing directions on the motor, on a 21 cm dia. wheel I think that equates to about 40 seconds of arc. Could be the principal source of error.

Also the pillow blocks show appreciable slop, resulting in about 100 microns displacement at the worm when push/pulling the trunion hard by hand, this also needs addressing, but is only going to give problems under heavy load.

Engineers blue was applied to the drive, nice even pattern, no bottoming out exhibited.

Bearing research starts now.

Much appreciated both of you.

 

Huw

[Horwig]

An update, the preload was not as tight as I'd thought, tightened it, backlash is now closer to 16 arc seconds, getting there I think.

I am still going to change the worm block bearings to angular contact, these are what I've been looking at :  https://www.bearingboys.co.uk/SKF--FAG/7201BEP-SKF-Single-Row-Angular-Contact-Ball-Bearing-22415-p

Also, the pillow block pack on the driven trunion is going to be replaced with a pair of taper roller bearings, these look possible:  https://www.bearingboys.co.uk/Imperial-Taper-Roller-Bearings/HM-801346-X2310QVQ523-SKF-Imperial-Taper-Roller-Bearing-82351-p

Any thoughts from anybody?

Huw

[tonyowens_uk]

1 hour ago, Horwig said:

An update, the preload was not as tight as I'd thought, tightened it, backlash is now closer to 16 arc seconds, getting there I think.

I am still going to change the worm block bearings to angular contact, these are what I've been looking at :  https://www.bearingboys.co.uk/SKF--FAG/7201BEP-SKF-Single-Row-Angular-Contact-Ball-Bearing-22415-p

Also, the pillow block pack on the driven trunion is going to be replaced with a pair of taper roller bearings, these look possible:  https://www.bearingboys.co.uk/Imperial-Taper-Roller-Bearings/HM-801346-X2310QVQ523-SKF-Imperial-Taper-Roller-Bearing-82351-p

Any thoughts from anybody?

Huw

1. The angular bearings are fine Huw. Loads of preloading capacity - they will be nice and stiff and still run with low breakaway friction. They are unsealed I believe so lube with fluorocarbon thickened grease e.g. DuPont Krytox then keep them clean and forget about them.
2. Those are are man-sized roller bearings Huw. I assume you are happy to deal with the overconstraint issue that will arise if you run the driven Dec trunnion shaft in those and keep the other idler trunnion shaft just 'propped' in a ballbearing pillow block! (either the fork or the OTA will flex a bit as the OTA is moved through its range of motion, which can be undesirable if significant). You could either align the constrained trunnion accurately with the 'prop' one, or use a flex-plate interposed between the drive trunnion and the OTA to decouple the tip/tilt nonlinearities. Or you could just do it and take your chances!
3. I assume your existing trunnions are imperial shaft sizes. Otherwise you would find that 40mm ID taper bearings are cheaper and stiffer than those HM-801346-X2310 units. You can avoid the problem of sourcing precision grade taper bearings (usually P5 or P6) to deal with the 20 micron run-out tolerance on their inner races if you can make the trunnion shaft hollow and optically align it with the opposing 'prop' trunnion!
4. As for the basic notion that the driven trunnion shaft should be run in a bearing design that constrains it like a machine tool spindle - basically you are correct! And taper roller bearings in popular sizes (as used for auto wheel bearings) are the cheapest, stiffest solution. Unfortiunately not the lightest and most compact.

Tony

[tonyowens_uk]

2 hours ago, Horwig said:

Thanks for taking the time to give such full answers both, lots to consider.

Firstly, the mount is stripped back, the OTA removed, and the fork removed from the polar axis for ease of measurement. So all tests are done off load.

Initial results this morning. The worm bearings are plain ball, but preloaded. Measurement just now shows end float of about 20 microns when changing directions on the motor, on a 21 cm dia. wheel I think that equates to about 40 seconds of arc. Could be the principal source of error.

Also the pillow blocks show appreciable slop, resulting in about 100 microns displacement at the worm when push/pulling the trunion hard by hand, this also needs addressing, but is only going to give problems under heavy load.

Engineers blue was applied to the drive, nice even pattern, no bottoming out exhibited.

Bearing research starts now.

Much appreciated both of you.

 

Huw

Sounds like you have nailed a big part of your issue!

[Horwig]

I'm slowly getting a workflow together for jobs to be done.

The bearings for the worm linked to above were larger diameter than the present ones, so would need a new worm block to house them, so I've found some angular contact bearings that match the size of the present ones: https://simplybearings.co.uk/shop/Bearings-Angular-Contact-Only/c3_4592/index.html

If the worm block shows any sign of flex, I will the go to larger bearings and fabricate a new stiffer block.

I've some left over 4 inch ali tube section with 3/8 wall thickness which should make a housing for the tapered bearings, and you are right Tony, I should address the single pillow block on the far side, I'll probably just replace it with the pair I'm taking off the driven trunion, should be better than what's there now.

Huw

[Horwig]

Thanks again Tony for putting up with my ramblings.

Things are getting serious. New angular contact bearings are in, they make a difference, but have not solved the problem. I'd been noticing that when I dis-assembled and then rebuilt the drive, backlash would be different, and I couldn't work out what was changing, till I rechecked the worm to wheel centering, it was off, a lot of head scratching later, I hope I've found what's causing it, the slop in the stacked pillow blocks is so great that the wheel can twist against the worm, so before I get anything more done I'm going to have to replace the DEC trunion bearings with a pair of tapered rollers.

Interestingly, with things as they stand, roughly half the backlash comes from the worm, the rest from the gearbox/belt drive, and most of that I suspect is probably from the gearbox.

 

Huw

[Horwig]

Just a quick update on where we're at with this.

The new trunion bearing is in, now at least I have repeatable results for DEC backlash, but it's not that brilliant, about 20 arc seconds or there about, with the gearhead on the motor accounting for something like half of that. After a lot of research I found some Japanese zero backlash gearboxes from KamoSeiko, and found one on ebay to play with, actually calling them gearboxes is wrong, they use ball bearings:

Here it is on a Pittman motor compared to the Pittman gearmotor that's in at the moment. I just about have space to accommodate the extra length.

New timing pulleys are on order, but there's a 15 working day delay, damn.

Anybody out there come across the Kamo ball reducers? This shows potential, but is there anything I should know that would make it unsuitable?

 

Huw

[Horwig]

First of all, a big THANK YOU to Tony Owen for his input on this rebuild.

I'll post the latest news in case it is of benefit to somebody else building a mount.

The DEC axis rebuild went slightly out of control, I ended up replacing the RA axis as well!

Here's the new RA axis in it's bearing block.

Here's the worm and wheel re assembled, note the red engine rebuild lube used to lubricate the worm, very clever stuff, the colour shows how well it distributes itself.

And here's the fork re-assembled, note the new bearing block on the left hand tine.

The hinges on both worm blocks have been replaced with flexors, as can be seen on the DEC worm block to the left of the mount awaiting re assembly.

First results are promising, PHD2 is showing guiding of about 0.5" repeatably, and it's applying DEC backlash of about 175 ms.

Last night I had a good PEC record run, and did a quick and dirty 300 second test frame before clouds came in.

That's a single luminence frame, just stretched and a little noise reduction added. Guiding looks good, but is it my eyes, or do I have triangular stars?

 

Huw

Edited August 29, 2019 by Horwig

[johnturley]

Interesting that the design of your fork mount appears similar to that of my 14in Newtonian built by Rob Miller of Astro Systems in 1984.

John 

[Horwig]

3 minutes ago, johnturley said:

Interesting that the design of your fork mount appears similar to that of my 14in Newtonian built by Rob Miller of Astro Systems in 1984.

John 

It started life as a Beacon Hill mount many years ago, looking like this

[Davey-T]

Great to see some reward for all your efforts Huw, looks like perseverance is getting results now.

Stars as you say look strange, like little hearts enlarged on my screen, could be optical or guiding issue I guess.

Dave

[Horwig]

17 minutes ago, Davey-T said:

Great to see some reward for all your efforts Huw, looks like perseverance is getting results now.

Stars as you say look strange, like little hearts enlarged on my screen, could be optical or guiding issue I guess.

Dave

Thanks Dave, it feels as if I've been at this forever.

I did have triangular stars when I first built this scope (I called them 'mickey mouse ears' I seem to recall), they were caused by mirror edge clips too tight, then I had astigmatism caused by a safety strap araldited to the centre of the secondary. These also could be caused by pinching of the secondary I suppose, but there are no clips, just three blobs of silicone.

Guiding was good, don't see many major spikes there.

One problem solved causes another one to shine through, hey ho, the life of an engineer...

Huw

[Davey-T]

Always something to give you brain ache, my  wedge mounted Meade SCT that has been working fine ever since the darker nights returned suddenly started producing oval stars in RA direction a couple of nights ago, visible even on 1 second exposures and identical on 300 seconds so not guiding fault, no idea why as nothing has been altered