Improve mount periodic error
From McWiki
Old Worm Tuning
 On advice from an online article, checked the worm blocks for squareness by measuring, with digital caliper, the distance between the front and rear edges and turning the worm blocks slightly with a wrench to bring those two distances equal, thus guaranteeing that the two blocks are parallel. (Square would be even better, but no easy way to check that.) The spectrum shows error at 1-worm cycle (2.65 arc-seconds) (real periodic error), and a larger 4.09 arc-second peak at 76 seconds. +3.4/-4.8 overall (RMS 4.132). So 8 arc-seconds peak-to-peak, which I would like to improve. Worse, the large 76-second component means it won't respond to traditional periodic error correction, which works only for multiples of worm cycles. |
Replaced G11 RA Worm Gear
Aug 14. I haven't tried it yet (hopefully tonight) but yesterday evening I installed a replacement RA worm gear on the G11 - the new "high precision" worm from Losmandy. It's brass, rather than the original stainless steel, and is machined to higher tolerances.
My expectation is that this will smooth out the smaller and highly periodic ripples in tracking. It won't, by itself, improve the 76-second error, though, because that is caused by pressure in the worm bearings, which are not replaced.
I did try to improve that too, however. In reassembling, I carefully squared the first worm block to the mount using a small machinist's square, and I carefully squared the second to the first by measuring multiple corner-to-corner distances with a caliper. This should have the two blocks nicely parallel and nicely orthogonal to the RA axis, which should give the bearings their best chance to run freely with little binding.
I set a fairly loose worm-to-gear connection with a bit of backlash, to assist with smoothness (backlash doesn't matter in RA since drive is always in the same direction.)
Next clear sky chance (possibly tonight) I'll do a long Pempro PE measurement to see what has been accomplished.
First new RA Gear test
Aug 16, testing new gear install. Set up about 8:30, did drift alignment etc., then gathered PE data. First thing I note with the new gear: it's noisy. Probably means I didn't adjust backlash well - maybe a bit loose. Will wait for PE data; if I have to open it up anyway that'll be a good time to look at backlash.
Wow. PE is dramatically improved. Standby for new graph. Way, way better.
 On second thought. No it isn't, I think I was displaying the wrong scope axis in PemPro. PE seems more periodic, which is good, and the 76-second error is slightly reduced (3.6 sec down from 4.1), which is good. But the actual error is worse +/- 5.5 (11 peak-to-peak). RMS improved - down to 3.744. Today I'll do some more tuning and backlash adjustment & try again. To do:
|
Tuning new RA Gear After First Test
Trying to get the improvement that new RA gear should have delivered, opened the mount up again today. Things I found & adjusted:
- The Oldham coupler was too far onto the shaft, with the result that the nylon coupler thing was not properly seated in the slot. Tightened that up, aligned, etc.
- With a balanced load on the scope,
- Adjusted backlash to minimum free play (0.46 gap with feeler)
- Re-checked squareness of worm blocks in all directions
- Lightly greased the worm with white lithium grease
- Ran through entire RA revolution in both directions several times
Not sure what more to do than the above. Next chance, I'll do more tests again. The new o-vision one-piece worm block is starting to tempt me. I'm not prepared to spend most of my available dark time testing PE - I'll buy the new worm if I don't see hope of improvements via adjustments soon.
Test of Tuned HP Gear
 Good improvement in the evenings test (Aug 17).
While there is any 76-second error, PEC will not work, because the 76s error is not periodic with worm cycle, so any PEC training cannot be correct for more than one worm revolution. I could probably get the 76-sec reduced further by obsessing over worm block alignment. However, I'm going to try an experiment first to consider leaving it as good enough here. I figure I can guide out any error with frequency bigger than several seconds; say, 10. There is very little error below that in the frequency chart. So I think this may be a curve that can be flattened out by autoguiding. Before spending money on a major worm upgrade such as the ovision, I certainly want to confirm this. So next clear night's test plan is to gather PE data while autoguiding, to determine if the result is stable enough. This will mean
|
PEC with Guiding
Last night I tried guiding for the first time. I had some trouble getting my intended guide camera (Lu135) to work, but eventually switched to an old DSI. (See autoguiding for more info on the problem.)
 It did produce quite impressive results. PE is now +1.5 / -1.6 ", very good. A bit surprising, the 76 second error is still there, although small (0.68 ", substantially less than one pixel). I knew it was still there in the unguided data, the surprise is that it wasn't guided out entirely. I guess it was sudden enough and large enough to have caught the guider by surprise. |
 So, I may still need to eliminate that 76s error entirely when I'm ready to go to long focal length imaging but for short focal length, that's just fine, and I'll start practicing some images. Colour next. Oh, and I posted a question in the Maxim group, hoping someone has hints about getting the Lu135 working. They do; so I haven't given up on guiding with the Lumenera yet. |
Replacement Worm Block
Having improved tracking as much as I seem able to, and wanting more to leave some headroom for longer focal length imaging later, I decided to order an oVision worm block, so I can install and calibrate it while the weather is still warm. order placed today, apparently takes a few weeks to be manufactured & shipped.
Factory Data
I received the custom test data on the oVision worm block from the factory.
Periodic error:
Error scatter chart:
The error is as large as 4 arc-seconds in a small number of spots (which is still small), and less than 2 arc seconds in most spots. More important, it is completely periodic - this graph is of 3 worm periods. That means that PEC will remove it with high accuracy. And that means that the 76-second error must be gone! (I don't have factory data on the frequency chart, but will gather that myself.)
This is very promising. I may not even need PEC turned on - depends on how rapid the transition is to the error blips. And with PEC turned on, I expect to see a very smooth plot that guiding can easily cancel completely.
Status
Installed the new worm Saturday, September 20, 2008.
- Generally went in ok
- Note the shafts aren't well lined up - relying quite a bit on slip in oldham coupler. Should I find a way to adjust this for better alignment?
- Moved old brass HP worm to Dec (thinking it would help smooth guider corrections)
Sunday, Sept 21, did some PEC runs.
Fill out:
- At first it looked good. (4-5 periods)
- Tried uploading PEC data to mount - strange results, didn't help
- Longer test (9 periods) went quite bad (+6.8/-8.6). Strange bumps in frequency.
Why on earth is there still a 76-second error component? It should be gone entirely, since it was dependent on the bearings in the old Losmandy blocks.
And what on earth is that error component 1/2-way between the 76s error and the worm error?
- I think I may have the worm too tight - perhaps as it proceeds around the RA gear it's hitting tight spots. Try loosening it a bit (more backlash).
- Confirmed from internet groups and a real test tonight (Sep 23). Gear too tight causes these symptoms. Measured carefully - had less than 1mm movement on counterweight shaft, supposed to have 1-2mm. Gap in block was < .7 both sides. Set motor side to .7, other side to .85. Play in CW shaft now 1.5mm.
- I think the misalignment of the motor and worm shafts, causing a large movement in the Oldham coupler, may be responsible for either or both of the 76 error and the mystery error. Pressure in that dimension is probably similar in effect to pressure from misligned bearings in the Losmandy setup.
Further Adjustment
October 12, did further adjustment:
- As recommended on 'net, ensured the outer end of the ovision block was exactly flush with the edge of the mounting plate (it wasn't). This centers the worm over the RA gear.
- Lightly greased both sides of the Oldham coupler to smooth the alignment motion.
- Bend motor mounting shafts straight to align the motor shaft and worm shaft.
- It was quite out of alignment (theory: bent by pressure from Pelican case. Check that, remove foam if necessary
- Straightening this shaft alignment reduced the Oldham coupler motion to almost zero, from the substantial motion in 2 axes that was happening before. Stiction in this motion has to have been a factor in the rough tracking noted.
- Loosened the set screws on both sides of the Oldham coupler, ran several revolutions, then re-tightened. This to ensure it wasn't too tight.
- Ran mount through complete range of motion, stop to stop, several times.
Don't know when I'll get a dark evening to test the results of this. Standby.
Attempted Test (Failed)
This evening (Oct 12) it seemed possible to test. CSC said partly cloudy, and there was sky showing as the sun set. So I set everything up. It became more and more cloudy until it was completely solid overcast, and I packed it in. Never even got to the point where I could see polaris for alignment. Saw Vega through cloud for a few minutes.
Frustrating. Next chances will be later in the week.
Another Attempted Test (Failed)
Evening of october 14, it was clear. Full moon, but thought it would be OK for PE test.
Complete disaster. Couldn't ge Maxim to connect to scope (and it takes a many-minute timeout each time it fails). After a couple of hours, I figured out the cable was broken. Tried a replacement but it doesn't work - not sure it was pinned properly. Switched to manual for polar alignment, but clouded over with full moon meant no stable stars, couldn't align.
Gave up and packed up.
Next thing I must do:
- Make a couple of new cables;
- Ordered new, better, USB-to-serial converter (a double-serial unit, to reduce cables again. Won't need external USB hub with this.)
- Use Gemini DEBUG mode and a terminal emulator to confirm I can communicate. (I believe I can because the GPS worked).
Update October 18:
Made 2 new cables (one spare) from scratch, being very careful with connections. Went over Gemini thoroughly, tested with debug mode, reset speeds, etc.
Everything now working - Maxim can connect to the scope again.
All this in daylight, now need another clear night to test. It's supposed to be clear tonight, but may be busy with family things, so we'll see.
Finally, another test - success!
 |
 The frequency spectrum shows 2.6 arc-second error at 1 worm period, and 0.7 arc seconds at 76 seconds. Still not sure why there is any 76s, but clearly the adjustment of shafts and oldham helped. |
 Just for fun, I started a data run with PEC turned on, with this new data loaded. After one worm period it was looking quite impressive (+1.7/-1.6), but I was too cold and tired to gather more data. |
This is good unguided performance, and well within the range of what autoguiding can reduce to zero, so I'm going to call this satisfactory for now.
Too cold and too late to try any test imaging or guiding - it was midnight by the time I got the PE data collected, and there was frost on everything.
Time to start thinking about how to continue this through the winter, as it'll be too cold for setting up and taking down in the back yard. Gotta get an enclosed observatory. For now, maybe a month or two of observing out of the garage will give me some shelter and a little extra season.
