NIRC/PDS is a technique for using out-of-focus 3 micron images to phase the Keck I mirror segments. It may be used in place of PCS. There has been anecdotal evidence that NIRC/PDS may actually provide better NIRC speckle observations than recently run PCS solutions, however this has not been confirmed by careful testing.

Running PDS

1. Start the NIRC software. Unlike normal observing, you don't need to worry about a runname; PDS will provide one when it is initialized.

2. Type "user pds". This adds the PDS scripts to the PATH.

3. "initPDS" (Determines the UT date, creates a data directory with that date, and a reduction directory on /s/scratch1 with that date. Also initializes various parameters by running "set_path" and "set_keywords." Observer prompted for PDS phasing wavelength which is traditionally done at 3.3 microns.)

4. "darksky" (Takes calibration images: bias and "sky" frames which are used to determine a bad pixel map. Also sets pat4xa, coadd 250, and nod 0 40).
   An alternative is to run "copycal", that simply copies existing calibration images over to the new directory. It also sets the coadds, timing pattern, and nod.

   ===== AFTER SUNSET =====

5. Go to an HST standard and run "malign" in K band as usual (except that you will be using pat4xa). (Starlist "0000_irhst.standards" has some good stars to use.)

6. "take_snap" Write down the snap number (you will restore it later).

7. "reload_focus"

8. Rotator should be set to STATIONARY -180 degrees.

9. Go to a PDS star (starlist "000_PDS.lis") and center it.
   PDS stars are K or M stars with Vmag ~3.0.
   Center using "snapiv" and "cent # #" with coad 1 . Remember to set "coad 250" when centering is finished.

   ---

   Note! Steps 10-13 below could be skipped if you are in a hurry, and if the star has been accurately centered in the previous step.

10. "take_two p" (automatically centers the image and gives the opportunity to check for excessive vignetting)

11. "load tcm" (creates and loads a 400 nm three-color mode)

12. "offsets" [This one takes a looong time (~8 minutes)!]

13. Ask the OA to restore the snap after malign (step 6 above).

    ---

14. "pah" (loads the 3.3 PAH filter typically used for PDS).
    If phasing at a longer wavelength, insert appropriate filter.

15. "newrun" (creates a new PDS_RUNNAME and resets the ITERATION counter)

16. "pds_loop" and repeat until convergence. ("pds_loop" takes the data, runs the analysis, and sends the moves. Convergence occurs when the RMS flattens out or gets below 10-20 nm. This should take less than about 6 iterations.) A "snap" is recorded at each iteration. (Note: One pds_loop takes slightly more than 5 minutes. Also note that coadds, timing pattern, and nod are set in this script.)

17. Ask the OA to manually reload the final snap, so that it will be automatically loaded next time ACS is started.

    ===== OPTIONAL =====

18. "load random" (creates and loads a random mode. To save time this would not be run during science nights or when other more pressing engineering awaits.)

19. "newrun"

20. "pds_loop" until convergence.

21. Ask the OA to manually reload the final snap.