Re: Pride Goeth Before a Fall

[Chris Albertson <chris@topdog.pas1.logicon.com>]

Tom Droege wrote:

> 
> What to do?  Well, I picked the screw for certain properties.  It is made
> by cold forming which makes a very smooth and cheap thread.  Possibly
> though, it also allows the rod to go through the die at an angle which
> would introduce a periodic error.  I note that the cork screw star trails
> were very regular.  They have a 60 second period.  Lots of time to do
> something in software to correct it if we know where the screw rotation is
> in time.  All that is needed to measure that is to add a limit switch with
> a once around flag on the screw.  Now one will know where it is.  There is
> already a DAC and a VCO to allow small changes in the stepping rate.  There
> should be enough range, I will check that there is.  I think I will leave
> this project for one of you to figure out how to code.

I took apart my Meade SCT once.  This is exactly what they do.
There is a screw sticking out of the worm gear.  The screw head
hits a micro switch every revolution.  Looking at their setup I'd
guess it cost them all of about $2.00 to implement.  It is very
simple.

The PC software could update the tracking speed every five seconds
This would be 12 times per rev.  This could reduce the error to
one pixel.  Going to a two second update period would not be bad
if the STAMP could handle it. (Does the motor skip a beat when the
rate is adjusted? I think you are using a count down rather then
a VCO so it may) 

One problem with this is that the PC will need to exactly know when
the switch closes.  I'd say to within 0.1 seconds.  We can't
poll that fast using the current interface.  I'd like the PC to 
know the current phase angle of the lead screw to within about one
degree.  I see two ways to do this:

1) Modify the way the STAMP talks to the PC so that the STAMP can
   send unsolistited asynchronous messages to the PC.  The STAMP
   would have to poll the switch at a fairly high rate.
2) Run the micro switch leads directly to the control PC, bypassing
   the stamp board.  I'd go to the printer port.  

Another big problem is finding a way to measure the periodic error.
We need to know:
 1) The amplitude of the error
 2) The phase 
 3) The shape of the periodic function.  Or do we assume a sin wave?
Taking images only gives us #1.  Any ideas?

This has to be measurable in the field.  If a system is ever disassembled
there has to be some way to redetermine the periodic error function.
Normally I think this is done by an autoguider of by manual guiding.



-- 
--Chris Albertson

  chris@topdog.logicon.com                Voice:  626-351-0089  X127
  Logicon RDA, Pasadena California          Fax:  626-351-0699