Bill Kunkel
Revised by Mark Phillips: March 2000

1. General Remarks

Operating the WFCCD to best advantage depends crucially on performing the alignment of the optics as an operation separate from focusing the telescope to obtain a “satisfactory” image.

Here alignment means performing those tasks that produce the most uniform image possible at the TEK5 detector of an “astrometric” aperture mask in the aperture wheel illuminated by the daylight skylight that “leaks” through the dome during the day. Specifically this excludes illumination by a flat-field lamp which, even at maximum current, represents a “red” light source and weights the longwave end of residual chromatic aberrations excessively. By alignment, this document restricts attention exclusively to the re-positioning of the field-flattener, and rotation of the dewar. Re-orientation of the grisms is a specialized activity that should not be attempted by persons other than Oscar Duhalde or Bill Kunkel.

The focus and alignment operation are, in a strict sense, separable, though in practice the quickest recipe to achieving a satisfactory setup is to do a preliminary focus, then to align, and last, to touch up the focus again. Although the focus operation may preferentially be done with a filter (say the B-band if the science centers there), the effects of alignment are achromatic.

2. Focusing

Install an astrometric aperture mask in an available aperture slot. Two such masks exist, one with 7 columns of 16 rows, and one with 7 columns of 11 rows. The latter is preferable. Two Hartmann masks should be installed in the filter wheel, with their straight edges parallel to the slit orientation. The masks should be placed to obstruct opposite sides of the collimated beam.

The focus position is read from a dial-gauge on the WFCCD side directly below the cover of the filter wheel mechanism. All focus values during the first year of operation have been between dial values of 0.202 and 0.216. A focus series usually consists of a sequence of images stepped in increments of -0.002 inches, going from higer to lower readings. There is hysteresis in the positioning of the camera lens that is somewhat ugly and problematic to control. The center of symmetry in focus quality shifts on the chip in (x,y) by several hundred pixels, depending on the direction of lens travel. By preserving the direction of lens travel, the shift is minimized. By convention, focus adjustments have gone from higher to lower values. If a new setting overshoots, it will be necessary to return to values 0.003 higher than the intended destination and start adjusting anew. So, for example, if a best focus is anticipated at 0.207, a reasonable series might begin at 0.210, stepping to 0.208, then 0.206, and ending with 0.204. Good exposure times with daylight leaking through the closed dome are ten seconds with a gain of 3 and a readout time of 78 seconds.

A pair of focus images taken through the Hartmann A and B masks, respectively, should be obtained at each focus value. An exposure time of 10 seconds should be adequate. Each pair of Hartmann images can be analyzed using the “hartmann” task of the WFCCD package in IRAF. (Type “help hartmann” in the WFCCD package for details on how to run this task.) This program outputs the pixel shifts between the two images to the terminal, and also provides a graphical display of the shifts which is quite helpful in determining the optimal focus value. After running through the entire focus range, it will become apparent that best focus for chip center is not the best compromise focus for the entire chip. The observer is left the decision as to what compromise in focus deviations provides the best tolerable error over the largest useful field.

Note that the hysteresis problem in the focus mechanism is further aggravated by mechanical “stiction” that provokes the lens to drifting in position after the last turn of the screw is applied, continuing lens motion in the original direction of motion. To combat this “stiction” it is useful to stop turning the adjustment screws 0.001 units above the destination, at which point one sets the clamp to the left side of the WFCCD (when facing the filter mechanism). During the clamping motion the lens drifts. If it has not reached the destination dial reading the clamp can be loosened and retightend, thereby provoking some additional drift. If after several loosenings and re-tightnings the destination has not been reached a bit more screw motion may be required. The steps in approaching 0.210 are

  1. Loosen the clamp handwheel at the left side of the filter cover.
  2. On the right side turn the push-pull screws constraining the focus torque arm until the dial reads above 0.212;
  3. then turn the screws moving the dial downward until a value of 0.211 is reached.
  4. Tighten the handwheel to the left of the filter mechanism, watching the drift. If the drift doesn’t reach the destination 0.210, loosen the handwheel and re-tighten. Apply more screw motion only if the drift does not reach the destination of 0.210.
  5. If the motion or drift have overshot, repeat the procedure from step 2 above.

3. Alignment of the Dewar Rotation

Dewar rotation is wisely done after getting the initial focus. Before attempting to rotate the dewar, the ten screws on the two “half moons” holding the dewar fast to the WFCCD base plate must be “backed off” by about half a turn, until the residual torque with which a screw can be moved is about the same for all ten screws.

After the screws have been backed off, an image of the hole pattern should be taken (without binning). If we think of the resulting image as a stack of rows with seven holes each, with the imexam task of IRAF measure the (x,y) position of the fourth hole in the top row and the bottom row, setting the cursor on each and striking the “a” key. The difference in x between these is the dewar rotation error. Translate the error to degrees with the formula:

  • ERROR = 57.3 * (x(top)-x(bottom))/(y(top)-y(bottom)) degrees.

Now go to the instrument and note the lowest dial on the instrument, the one touching an arm sticking away from the dewar wall. Note the value. As with the camera lens, the dewar seating also allows for considerable hysteresis in alignment steps. For this reason one should always approach a desired alignment setting in the same direction during a trial series. By always coming from the same direction, even though the dewar seats irregularly, approaching from the same direction assures that the seating error will always be the same, and so should cancel to a high degree. This means that, should the rotation effort jump past the destination value sought, one should wisely back off by a full dial turn and try nudging toward the destination an additional time. After one finishes the “nudging” of the dewar, do NOT tighten the ten screws; they should remain loose, and a follow-up exposure made (with the dome lights off!). As an aside, it goes without saying that a series of adjustment should be done by the same person, since handling differences leave different hysteresis residuals.

One can usually achieve a residual rotation error of one arc-minute (half a pixel in a y-difference of 1,500) in three iterations.

After the last rotation iteration the ten screws under the “half moons” should be re-tightened to roughly comparable torque. They need not be muscled down, though.

4. Alignment of the Field Lens

Field lens alignment should be attempted only after a fairly satisfactory (though not final) focus has been achieved. As with the focus adjustment and the dewar rotation, an exposure of twenty seconds through an astrometric aperture mask provides the diagnostic CCD frames. However, now the two HARTMANN masks must be installed in the pair of filter positions to each side of the grating positions (unless no grating is installed). In one position the obscuring half should face toward the filter rotation axis, while in the opposing position the obscuring half should face in the opposite sense. Either the East side or the West side of the collimated beam can then be obstructed. The former transmits the “East” beam, the latter the “West” beam. Do NOT try this test obstructing the North or South portion of the collimator beam, as then the error dispacements are in the same sense as the automatic stepping in a focus sequence.

A typical measurement exposure consists of a focus loop of three exposures. Overlap of images makes the 7 x 11 hole mask unsuitable, so the top left of the 7 x 16 hole mask is simulated. The holes are a bit tedious to disentangle, and should be grouped as shown shematically

       TOP LEFT of FRAME   *       |- *          *          *    . . .
                           *       |- *          *          *    . . .
                                G1 |   
                           *       |  * -|       *          *
                                   |     |       
                           *       |  * -|       *          *
                           *       |- *  | G2    *          *
                           *          *  |       *          *
                   (W)                   |
                           *          *  |       *          *
                           *          * -|       *          *
                           *       |- *          *          *
                           *       |- *          *          *
                           *    G4 |  *          *          *
                           *       |  *          *          *

Groups 1, 2, and 4 are identified; the lowest member of each is the first exposure of the loop, the upper member the last. There are eleven groups per column.

The table below shows the last digits in the X-readings from an Imexam series. The Center, N, and S entries are from column 4; while E and W are from columns 1 and 7, respectively. Center, E and W are from row 6, while N and S are from rows 1 and 11, respectively.

Action         Center      N         S         W         E

Mover TEK5      33.10    35.17     32.09     21.98     74.20
to S limit      32.19    35.21     28.78     20.78     73.17
                33.09    35.16     32.08     21.97     74.19
                 -.90     +.04     -3.30     -1.20     -1.03

These readings show the greatest dissimilarity between the “N” and “S” displacements. So the trial correction applied was to move the entire dewar plate Northward by 0.050 inches, keeping the E-W dial pair at the same settings. After the adjustment the next exposure produced the readings shown

Action         Center      N         S         W         E

Move N from     38.39    39.18     38.73     98.10     79.53 
0.55 to 0.00    37.45    37.91     36.03     96.58     78.30
                38.39    39.17     38.72     98.10     79.55
                 -.94    -1.26     -2.70     -1.52     -1.24

Quite apparently further northward motion is required, and after that a touch of westward motion and a nudge upward in focus from 0.206 to 0.2095 produced a final set of readings shown below:

 Center       N         S         W         E
                24.45    25.11     24.67     83.91     65.54
                24.76    24.31     23.88     83.73     65.37
                24.44    25.12     24.66     83.90     65.64
                 +.31     -.80      -.78      -.17      -.17

Note that because the hole pattern is rectangular, and the N, S holes are further from the chip center than the E,W holes, that the field curvature toward negative values increases as the square of the dis- tance from the center of field symmetry. Adopting displacements smallet than |0.40| as acceptable, the focus setting tabulated in the last example is as good as one can currently obtain.

When a measurement pattern as the last set of readings is obtained the alignment process can be considered as concluded, and should require no further “touch-up” until the dewar is re-mounted. For the obvious reasons the alignment procedure should be done last.

Table of Contents