Intro

After a few pointing runs with tcs3, we are not getting the performance we
want. The last few fits were just below 20" RMS.

Is this purely telescope mechnical problems? Wanted to use TPOINT to confirm
(or rule out software issues).

1. p_on - Can  I reproduce the pointing coefficients with a simulated pointing run?

Took the last pointing run coefficients and did a 'simulated' pointing run.
Our last run, produced:
  IH        +88.4050    33.04266
  ID        +16.8021    34.25800
  NP        -13.3887    13.00273
  CH        +57.1209    29.64567
  ME        +18.6128    14.94447
  MA        +24.5501     4.41281
  HCES     +169.0332    32.21872
  HCEC      -72.4414    17.33625
  DCES     -465.9502    26.80905
  DCEC     +151.4332    33.23230
  FO        +20.3480    19.23468
  TF        -15.7434    33.35904
  TX        -26.2959    10.19746
TPOINT should generate the same coefficients.
Produced this file: p_on.dat using the above terms.
Using tpoint it did this:

   indat p_on.dat
   USE IH ID NP CH ME MA HCES HCEC DCES DCEC FO TF TX
   fit
  
Did not get the same coefficients. If I set the the coefficients,and DO a fix

   [set coefficients]
   fix
   fit

I get a Sky RMS of 7.33
Here is a e9 plot p_on.pdf. You can see the error
are caused by the 4 points > 50 deg in declination.


2. p_off - Turn pointing off, and see what the software does?

Next I did a pointing run, with all the coefficients zeroed.
Produced this file: p_off.dat.

Ran TPOINT like so:

   indat p_off.dat
   lose
   fit

Here is the e9 plot: p_off.pdf.

Sky RMS & Popn SD is < 0.5". Not ideal but less that 0.5" shows that the
TCS is doing the sky to mount transformation correctly. So the problem
must with with my point map code.

3. Which Cofficients is wrong?

Did a few test (not show here), but TCS + TPOINT work together if set TX to 0.
Poiting file with TX=0: p_zero_tx.dat.

   indat p_zero_tx.dat
   lose
   USE IH ID NP CH ME MA HCES HCEC DCES DCEC FO TF 
   fit

   Sky RMS =   0.18
   Popn SD =   0.28

Fitting to the data, I get very close to the original coefficents. Fixing the 
coefficients to the  original values, I get this:

   indat p_zero_tx.dat
   lose
   USE IH ID NP CH ME MA HCES HCEC DCES DCEC FO TF 
   IH        +88.4050
      ...
   TF        -15.7434 
   fix
   fit

   Sky RMS =   0.62
   Popn SD =   0.62

All error below 1"...Things are working if I don't use TX!


3. Fit data from #1 with stars > 50deg declination removed

Took the data in #1, and removed the star with declination > 50 degress and did:

   indat p_on_removed.dat
   USE IH ID NP CH ME MA HCES HCEC DCES DCEC FO TF TX
   fit

   Sky RMS =   0.21
   Popn SD =   0.57

   [set coefficients]
   fix
   fit

   Sky RMS =   0.54
   Popn SD =   0.54

 All error within a arcsecond.

4. Is it my apply_tx() function or tpoint? 

I suspect it's my TX function():


   void apply_tx_3(
      double *dha, double *ddec,     // O: adjustment value in radians
      double ha, double dec,         // I: telescope postion 
      double tx, double lat          // I: tx and latitude
   )
   {
      double sh = sin( ha );
      double ch = cos( ha );
      double sd = sin( dec );
      double cd = cos( dec );
      double sl = sin( lat );
      double cl = cos( lat );

      *dha  += tx * cl * sh / ( ( sd * sl +  cd * ch * cl ) * cd );
      *ddec += tx * (cl * ch * sd - sl * cd) / ( ( sd * sl + cd * ch * cl ) * cd );
   }

Here is a grat of TX for declination bewreen -20 to +60. The slope after 50 degres increases.
Seems it should be more symmetric around zenith. Hmm...