Drift problem continues

Finally got a chance to test the full system,  after the refractor only system worked so well. I’m not out of the woods yet.First, after all the weight has been put back on the mount, the polar alignment has shifted again. This time it wants 5 ticks in Az and 9 ticks in Alt. So, it appears that the original PA wasn’t wrong – the PA changes with weighting down the mount. Don’t know why this is, Bisque seems quiet about this.

Second,  I played with the idea of randomizing the  order of points when doing the TPoint model. AAG mapper does the points in order, and they are quantized in Alt values. I exported the points into Excel and a) randomized the order, and b) varied the Alt so the points have more variety.

In 50 point runs, random points give the same polar alignment results compared to ordered points. However, you get longer slews on average, and much more large errors on the orthographic view. As a result, you get a bigger circle RMS value for the model. For example, ordered 50 point runs gave a supermodel with 20 arcseconds radius while randomized runs gave a radius of 28 arcseconds.  When I ran the big model (374 points) I got a radius of 20 arcseconds instead of the earlier (November) 9.9 radius.

The points have a general NS line rather than being symmetric. Another odd feature, the dD versus n graph shows a linear trend upwards. PWallace notes the abnormalities, but doesn’t have a theory as to cause.

Also, the error buckets are in 20 as bins rather than 5, another reflection of the increased errors from randomizing the points. Note that the effect of randomization would be more pronounced in the bigger model, since the typical slews would be very short in the ordered case. Random slews jump all over the sky, and result in bigger errors. I suspect the randomized model is a better characterization of the scope since one often makes large slews to targets.

test images showed slight Dec drift in 10 minute exposures with ProTrack running. In the East and South ProTrack can almost eliminate the drift now, but not quite. The drift seems to be more pronounced in the West for some reason.


FocusMax fixed

Finally, success with the FocusMax problem. Steve made some changes in constructor/destructor logic and cleaned up a memory leak.  The new version installed and ran fine for 320 iterations using ACP, Maxim, etc using the FocusFilterTest plan.

Drift Problem Solved?

I seem to have found the source of my drift problem, although it is perhaps slightly early to tell for sure. Unfortunately, the answer raises a number of other, potentially scary, questions.

The Story

I removed the Edge 11 and mounted the Sky90 refractor. Got everything balanced, etc, ran my tests. Still drifting, same as before.

At this point I still had the 350 point TPoint model in TSX, created with the original complete Edge + refractor setup. It was still helpful in pointing, but I certainly didn’t expect it to be useful in ProTrack. I decided to rerun a model since the 350 point model certainly no longer applied to the simple refractor setup, and I was guessing that might be the next thing Daniel would ask for. I ran a 50 point model to start, and WTF? the polar alignment is WAY off, 11 tics in azimuth and 5 tics in altitude (11 arcminutes and 5 arcminutes, I think each tic is 1 arcminute). The earlier responders were right about the alignment; I had ignored that issue since I “knew” it was OK, I had just aligned 2 months earlier.

I ran several more 50 point models while I got the alignment adjusted, then ran a 128 point model for now. Alignment is “Good enough” in both axis; I will do a more complete model when I get back to the desired configuration. The drift is completely gone in all directions (East, West, and South targets). 10 minute exposures are great even without PEC or ProTrack. Hallelujah!


So, this raises questions. The prior polar alignment was done November 12, with 7 or 8 50 point runs followed by the 350 point run all run the same night. Somehow, in two months the alignment shifted that much.

1) I have occasionally bumped into the OTA or counterweight shaft while working in the small area in the dome. Would this be likely to cause an 11 arcminute shift?

2) Would typical maintenance tasks such as greasing the gears, adjusting camera positions, or rebalancing be likely to cause this?

3) Can the polar alignment be impacted by the weight of the payload/counterweights? The original system was about 50-60 pounds of OTA and cameras; I needed both counterweights and the extension shaft to get it balanced. The refractor-only setup has maybe 10 pounds of payload, only needs 1 counterweight up near the mount, no shaft extension. Perhaps the heavier system can make the pier “bend” somehow?

4) Scary – can a mount settle or shift this much over 2 months? Might it still be shifting?

ImageThe concrete base is 18 inches square on top, about 28 inches square at the bottom of the hole. The hole is 42 inches deep. The base and pier are isolated from the building. The pier itself is 8 feet long, 8 inch diameter steel pipe filled with sand. I am located in Arizona, with desert “caliche” sand/clay in a flat yard. The cement was poured in January 2012, good polar alignment was done 9 months later onNovember 12, then it is bad two months later in January 2013. Is this normal for settling? Might I hope it is done, or should I expect continuing shifting?

4) Other than running a new TPoint model, is there an easy way to verify polar alignment precisely? I expect to want to verify alignment every week or two, at least until I am confident it is stable.



My next task is to remount the Edge and verify it is working correctly. I will run new TPoint models there to see if alignment shifts. Assuming things are under control there, I will remount the whole system and recheck.