The test results are presented in Fig. 1 (bottom). Each point was derived from a sausage cube, i.e. a matrix with 50 Z-steps sampled at . Since the lorentzian line FWHM was about 15 steps in Z-units, the line was very well sampled. The line fits were lorentzian in form. Unless otherwise indicated, all spectra were obtained with the telescope pointing at zenith.
As Fig. 1 (top) shows, the temperature inside TAURUS was slowly rising over the weekend. This is very well coupled to the dome temperature shown as dashed (24.6.00) and dotted (25.6.00) curves. By comparing the top and bottom figures, it is clear that the overall thermal time constant of the TTF is measured in hours. In the appendix, I have worked out the time constant for each component of the TTF, but of course these are all in thermal contact, so the time constant is driven by the largest element.
Sequence A. The initial sequence in A did not go well. The line profile was highly asymmetric, the line FWHM seemed to vary in time, and the expected temperature drift was not that well behaved. It was then discovered that the time constants were not correct at the CS100 (512,256,512). Once these were changed, the loop was opened and closed with (256,256,256) and this produced the final point which falls below the curve, but the line profile was still asymmetric. It was then realized that maybe Ysshould be returned to 0, as it used to be before the time constant value was modified in 1998.
Sequence B. With the reset value of Ys=0, the line profile was now highly symmetric, and the temperature drift very well behaved. The drift appears to couple to the instrument temperature which in turn was driven by the ambient dome temperature. The last two points arose from slewing the telescope by -45 away from zenith and back again. Note that the last point does not sit on the fitted line.
Sequence C. Tests were continued the following day. Once again, the temperature drift was well behaved, although rather slower than the previous day. The TAURUS temperature could only be measured to 0.1C accuracy, so the temperature slope for Sunday has a large uncertainty.
Between 1h and 3h UT (25.6.00), the TTF was tested for slew effects. These are quite dramatic as shown in Fig. 1. In order, the points were taken at different slew (hour) angles from zenith ( , ): -10, -20, -30, -40, -50, +10, +20, +30, +40. The last point was taken after returning to zenith.
There are several key things to note. Surprisingly, a tilt of the TTF in one direction on the sky is not symmetric with a tilt in the opposite direction from zenith. In other words, the direction of the Z-drift was expected to be the same in either tilt direction. It is not clear why this should be so (but see Appendix A). Furthermore, TTF does not recover its response exactly immediately after returning to zenith (see also the last measurements in Sequence B).