Sunday, June 2, 2013

Lessons in Precision

I explained to a ScopeRoller customer a couple of weeks back that the cost of precision is a sliding scale, and an asymptotically increasing scale at that.  "Fifth of an inch precision is free; hundredth of an inch precision costs a bit, thousandth of an inch precision costs a lot."  I had that lesson myself over the weekend -- but that cheaping out on precision has its own costs.

I mentioned that I was not happy with the ability of Big Bertha to hold collimation, and that I suspected that the spider holding the diagonal mirror was the problem.  As the telescope went up in altitude, I could see the laser beam from the collimator moving as well.  I decided to replace the .0325" thick aluminum legs, which were held in position by tension, with .040" thick steel legs.  These should be at least four times as stiff as the aluminum legs because of material and extra thickness.

In addition, the cylinder that connected the legs to the diagonal mirror holder was made of acetal, and left about an inch of the 1/4"-20 screw that holds the diagonal mirror holder unsupported.  My thought was to replace the acetal cylinder with a full length aluminum cylinder instead.  Because aluminum is about 20 times as stiff as acetal, weight for weight, I could go to a somewhat smaller cylinder and have far better stiffness.  I had hoped to find a piece of aluminum tubing 1/4" ID with about an 1/8" wall, but the local metal supply store had nothing like that, so I took a piece of scrap .811" aluminum rod, turned the ends, then bored a 1/4" hole through the middle with the lathe.

Here is the spider, waiting for the flat black paint to dry:

The trick here is that the legs needs to be 120o apart, so the attachment holes on the cylinder where the legs mount need to be 120 degrees apart.  I decided that yes, I could use some of my precision gadgets to make them 120o +- 0.5o -- but why bother?  I could just measure with a protractor, and that would be good enough.  Right?

Wrong.  It turned out that they were far enough off that the cylinder was not centered in the tube -- not even close.  So I ended up doing the extra work to do it right -- and it wasn't really much more work than doing it sloppy and wrong.  Here's a picture with another piece of aluminum in place, not the one that I drilled:

It sounds a bit like a Rube Goldberg contraption, but it isn't.  The tilting table on the right is at 90o; mounted on the tilting table is a rotary table made by Sherline.  It has a coarse measure that reads to the degree (and you can certainly interpolate quite a bit more finely) and a vernier measure (the red handwheel) that gives marks down to 0.1o.  Mounted on the rotary table is a 3" three jaw chuck, which held the cylinder I was trying to mark.  With this, I was able to determine that my old tapped holes were as much as 10o off of where they should be.

Use precision tools for precision work; use brute force tools for brute force work.  I could have tried to drill and tap the holes with the vertical mill, but because the cylinder was 3" long, and the chuck really does not hold something that long terribly well under any real load, it was not likely to be a success.  So I settled for marking where the holes should be with an end drill for angle and length down the cylinder, and then drilled and tapped the holes on the drill press.  Once you have marked the hole's center with an end drill, the twist drill in the drill press does a pretty impressive job of following that hole, even if you aren't exactly on location with the twist drill.

Anyway, here it is installed:

Yes, there is a slight bend to the arms; better to have some compression rather than tension, and by bending the arms slightly, I was able to get the diagonal centered to better than 1/16" -- perhaps even as close as 1/32".  (The curve may also smooth out diffraction spikes -- the four or six arms that you often see in observatory astrophotographs.)  This is about as much as I can expect or even need for this part of the problem. 

There is no longer the gross and obvious miscollimation of the diagonal as the telescope rises in altitude.  The sky did not clear here last night until I was exhausted and ready for bed, but perhaps tonight I will roll it out and see how it goes.

No comments: