Collimation at 4.0-m

Collimation Procedure for the 4.0-m Telescope at Prime Focus

There are 3 adjustments possible:

  • 1. Tilting M1: this is done by moving in or out the 3 defining hard points of the mirror cell. The screws are located in the roof of the Cassegrain cage and are equipped with dial indicators. The screws are actionned with a special tool. The scale of the dials is usually set so that it reads 0 when air is OFF. The dials are located on a radius of 54 inches while the hard points are on a radius of 68 inches. Therefore a 1/1000" turn of one screw, while the other 2 are fixed, will create a tilt of 2.34 arcsec of the mirror.
  • 2. Centering the PF pedestal (ie. Prime Focus Corrector + Mosaic): there is a centering mechanism on both the northern and southern sides, that works in a push-pull mode. A ±5mm range is available for centering. Dial indicators are installed and precise centering can be done. Holding screws (22 of them) must be loosen first.
  • 3. Tilting the PF pedestal: there are 10 push-pull screws all around the pedestal but no indicators are provided. Holding screws must be loosen first.

Only option 1 is readily doable any time at the telescope. Options 2 and 3 require more delicate adjustment inside the PF cage (see CH2150.540-C030 for details) as it has to be done at zenith with someone standing in the cage.

A 'pure' focal plane tilt (revealed by a defocus slope across the field) can be corrected by tilting MOSAIC at its interface with the PFC with push-pull screws. Beware at the installation of Mosaic: these captured screws must normally be loose to avoid a non-intentional focal plane tilt (it happened once)!

The PFC is designed to deliver a 48 arcmin non-vignetted field with FWHM under 0.5arcsec at the edge of the field (see Tom Ingerson's paper). The PFC optics include a field corrector for coma (paraboloidal M1) and spherical aberration (Ritchey-Chretien M1) and an Atmospheric Dispersion Corrector (2 pairs of rotating cemented prisms).

In terms of alignment requirements, the PFC optics have little optical power (1.076) thus are relatively insensitive to tilt, and behave like the M2 of a Cassegrain telescope where lateral centering is important to avoid misalignment coma. If decentering coma is present, the PFC optics must be translated laterally. In practice, for small corrections, we do tilt M1 instead. Zeroing the coma on-axis is not a sufficient condition. It is also important to check the field astigmatism off-axis (see simulations below). On-axis astigmatism doesn't reflect a misalignment problem but rather a problem with the cell of the primary mirror ("mirror pinched") and must be corrected via the active optics lookup table. If spherical aberration and/or field curvature are present, the longitudinal position of the PFC is not optimum and must be corrected by adjusting the back focal distance as Mosaic does not focus independently of the PFC (this was done once in Nov99 after the installation of MosaicII).


How to correct coma by tilting M1?

This can be done quickly by looking at a through-focus sequence for an on-axis star (9x100um steps is ideal, usually coma shows up nicely 200um away from best focus for a reasonnably bright star). Determine the cardinal direction of the coma (from the coma head to the tail): this side of M1 must be raised. The cardinal directions of the Mosaic field in the image display are: east is up, north is right. Alternatively, you can look for the minimum coma point in the entire field and consequently lower that part of the mirror. Iterate the adjusting tilts until the on-axis through-focus sequence is round and uniform. When wavefront errors are available from the Hartmann screen data, 1/1000 inches tilt is equivalent to an OPD of 230nm at the edge of the pupil, thus would correct about 230nm of coma measured on-axis. This is about the smallest correction that one can do with confidence and repeatability. Always keep a record of the tilt values of M1 (ie. values of the 3 hard points -south, northeast, northwest- dial gauges with air ON).



The following simulations are made at 650nm at zenith (ADC neutral). Only coma and astigmatism patterns do change; the trefoil pattern is not modified by misalignments. Reference field aberration maps for an aligned telescope are to be found in the PF imaging section of the main optics web page.

  • See on this coma field aberration map the effect of a 1mm decenter of the PFC+Mosaic (with respect to M1). The field astigmatism pattern changes very slightly.
  • A 1mm decenter of the PFC+Mosaic (with respect to M1) corrected by the appropriate tilt of M1 (25 arcsec) returns to the ideal case of 0 coma on-axis but the astigmatism pattern worsens a bit.
  • Sensitivity of the 3 types of misalignments: 1660nm (=0.23" in d80) of on-axis coma is produced by 1mm decenter of the PFC, or 25arcsec tilt of M1 or 0.35deg tilt of the PFC.
  • When on-axis coma is present, if one really wants to know whether it comes from a tilt of M1 or a decenter of the PFC, it is possible to look at the differences between the field astigmatism patterns. For a same amount of coma, the case of a M1 tilt does generate more astigmatism and with a particular pattern, where the vector axis always points away from the axis thus rotates around the field (beware, when you examine images, this pattern becomes apparent only for a minimum amount of tilt).
  • Besides generating misalignement coma, a large tilt of the PFC, 0.2deg in this case, will produce very particular astigmatism patterns: radial inward if the tilt is in the Y direction (ie. north-south, which is the prism axis in neutral position), and circular if the tilt is in the X direction. A subsequent appropriate tilt of M1 to zero the coma on-axis would almost not decrease the astigmatism pattern.
  • Conclusion:


How to quickly correct astigmatism with the active optics of M1?

If astigmatism is present on-axis in a through-focus sequence, it is possible to correct for it with a TWEAK of the primary mirror (read first the TWEAK section of the 4m active optics on-line manual!). This is a 'quick-and-dirty' method but it can improve the images when the lookup table is not working well. On-axis astigmatism is caused by a non ideal support of M1 in the cell: that problem typically varies with telescope position, thus a TWEAK is valid only for a specific telescope position ±5deg (ie. about ±20min of exposure time). So if you follow the same object, you should actualize the TWEAK about every 40min.

Run the IRAF task mscexam on each out-of-focus image of a single star and look at the ellipticity angle (called 'pa') with the ',' option for example. To make sure that you are in presence of astigmatism, pa should flip by 90deg across best focus and be fairly stable on each side of the best focus. If the pa exhibits large variation and/or does not change roughly by 90deg across focus, don't even try that method!

Following is a list of measured pa for the IN-FOCUS image (called 'pa in') and the corresponding angle to enter in the tweak table of the tcp (called 'pa tweak'):

  • if pa in > 0, pa tweak = 360 -2*(pa in)
  • if pa in < 0, pa tweak = -2*(pa in)
  • Hint for double-checking: a 'pa in' of 90deg is equivalent to a vertical astigmatic image in the mscexam xgterm window (not in the main image display), while a 'pa in' of 0deg is equivalent to an horizontal astigmatic image.


  • pa in 90 -> pa tweak 180
  • pa in 67.5 -> pa tweak 225
  • pa in 45 -> pa tweak 270
  • pa in 22.5 -> pa tweak 315
  • pa in 0 -> pa tweak 0
  • pa in -45 -> pa tweak 90
  • pa in -67.5 -> pa tweak 135

What amplitude shall you enter in the tcp tweak table? This is even more empirical: in reasonnable seeing of 0.8", a 1um astigmatism will be visible at 100um off the best focus, but hardly visible at 50um... Once again, although this technique works, it is experimental and must be used with care until a better lookup table can be built. This method can actually also be used to update the astigmatism lookup table using the seeing images recorded every night (ie. not waiting for a dedicated engineering night).


Note: SCALE OF THE WAVEFRONT ERRORS (optical path difference at the edge of the pupil): at the 4m, 0.1" image degradation in d80 is produced by 0.70u of coma, 0.30um of astigmatism, 0.26um of trefoil, 0.24um of quadrafoil.



Maxime Boccas, 28March01