A Bench Star

[NoelC]

A couple of weeks ago, when it looked like the stars were never going to come out again, I set about building one to use indoors.
I spent a couple of days drawing stuff on the back of envelopes before coming across a reasonably priced 'z drive' on ebay.  With an arduino stuck in one end and a pinhole torch strapped to the top I figured I could get it to run at the same speed a mount will run:

It's inclined at an angle to suit the elevation of my belt modded HEQ5 Pro mount and set to run from one end to the other in about 18 minutes.  The torch gives a reasonably good guide star of around 4HFD at .01S exposure and the length of the track at a range of 4.5m allows a little over one HEQ5 worm revolution.  The period of the driving worm on the Z drive is around 14S.

This is a view of one of the PHD logs with the mount running unguided at solar rate (for convenience), for about 1.5 worm revs. Which confirms my experience: periodic error of the order of +-15 arc seconds.  Although relatively smooth, it gives the guiding a lot to work on, and typically PEC has created high frequency transients in guiding without being able to completely smooth out the curve.  The DEC trace describes a curve because at the declination the mount is working at, the telescope describes an arc with respect to the bench star which travels in a straight line (whereas real stars of course describe an arc). It also serves to show if the track is perpendicular.  What you cannot see is the backlash which is bad on RA and disastrous on Dec (not as simple to correct as you may suspect).

The bench star is running in a straight line (along a cord) rather than in a circle, but according to my fairly basic trigonometry it affects the tracking speed by less than 1% at each end with regard to the centre over this width (310mm).

I built it to use as a test bench for tuning my mounts. Reading one of the forums they said that the kit necessary to test a mount on the bench was beyond the reach of the amateur, so I thought I would give it a go.  The kit used commercially for calibrating mounts I suspect is a 10,000 increment rotary encoder; which in my case that would cost several times the cost of the mount.

[JohnH]

That's an amazing job. Way beyond my skills!

Well done.

John

[Mike]

Noel why not have the 'start' on a rotating disc, turned by a stepper motor with a micro-step motor controller?

[NoelC]

Mike
Did you mean star on a rotating disk?  Didn't consider it but thinking about it;
Surely the resolution would be an issue (200 steps a turn X 64 microsteps)?  I guess you could use a geared stepper (5X200X64 = 20 arcsecs/microstep).  The Z track is stepping at approximately 1 arcsec/microstep (currently 16 microsteps/step) in just under a second.  Also at that range wouldn't the diameter of the disc be huge unless you restricted the altitude and declination to near 0° (which makes your optical errors larger). 

Do you think a disk would work?  Have you seen something similar?
Noel
   

[Mike]

Yes a micro stepper with gears would give you as much a range of motion and control as you wish. I would imagine a star on a rotating disc would give a more realistic movement as it would simulate a star near the zenith. I guess it depends how far away your artificial star is from the scope and how sharp the star is.

[NoelC]

Mike
At the zenith I think stars travel in a straight line observationaly (at right angles to the polar axis). 
I think you may mean the centre of the disk is placed at polar centre.  Not sure how that would affect measurements of periodic error; but assume it's possible, which would allow measuring multiple periods ad infinitum, however it would require polar alignment (groan).

My target resolution is 1/10th arc second, as that is close to a single step of the mount.  In practice you could work with less than that, but would need to be better than the 20 arc seconds produced by a 5:1 or the guide star becomes very elongated and error un-manageable.  A gearing ratio more like 100:1 would be required.  I'm not aware of any steppers in my budget range (under £30) that will do that.  Are you?  Also when I've tried to get huge belt reductions, I've generally not got beyond about 20:1 nor been free of introduced periodic error. The size of the disk might also be an issue.

I did consider using a cheaper rotary encoder and gearing it up with timing belts, but the problems of consistency at high gearing and eccentricity of the cogs used seemed in-surmountable.

Noel