On-Sky Testing of of the NASA42 Camera

Image of M81 obtained with NASA42 Camera---Kathryn Neugent

This summarizes the results of of on-sky testing by Phil Massey and Kathryn Neguent after Ted Dunham adjusted the operating voltages of the NASA42 Camera chip in order to improve the linearity. Specifically we wanted to perform the following tests:
  1. Measure the gain and read-noise
  2. Test linearity on the sky
  3. Determine best way to flat field broad-band data
  4. Determine color terms and count rates in the broad-band filters
  5. Test the uniformity of the PSF around the chip
  6. Identify any remaining software user issues
In addition, we wanted to come up with some practial observing notes.

The final voltages were achieved by Ted on May 10th, and we conducted our tests on May 11th, a photometric night but one mared by poor and variable seeing.

General Comments

Observing with the NASA42 Camera system as a pleasure, particularly with the only 18 seconds deadtime between consecutive exposures (binned 2x2 and through a single amplifier). That, and the uniformity of the shutter even at for sub-second exposures makes it nearly trivial to obtain sufficient twilight flats even though multiple filters.

Quick facts

Test 1: Gain and read-noise

We measured the gain and read-noise for Amplifier A (the default single amp readout) using Janesick's method on pairs of dome flats and biases, restricting the sample to the middle 200x200 section. We obtained values of the gain of 3.71-3.78 e/ADU, and values for the read-noise of 4.27-4.35 e-. Using a full-blown analysis Ted finds 3.84e/ADU and a readnoise of 4.6 e-. So, reasonably solid values are 3.8 e/ADU and 4.5 e-.

Test 2: Linearity on Sky

We took a series of 26 exposures of a (random) star near zenith. The exposure times ranged from 0.2 to 50 seconds, with a resulting peak count rate of 280 to 64,000+ above bias (roughly 1632), i.e., taking it right up to the 65,535 range of the A-to-D converter.

Because the seeing was so bad (fwhm 4.5-7.5 pixels) we used measuring radii of 30 and 50 on the star. Our results are shown in the two plots. As far as the stellar photometry is concerned, there is no sign of any non-linearity over the full range of the A-to-D converter

Test 3: Flat Fielding the Data

We took a series of exposures of the dome flat through each broad-band filter, and also exposures of the "uniformally illuminated" twilight spot shortly after sunset. There was a clear gradient from left to right on the images, amounting to about 4%, with the twilight having more counts on the left. Which (if either?) is right?

We answered that definatively by simply moving a (random) star around the field. Using twilight skies as the flats gave consistent photometry to <1%; the photometry based on flattening by the dome flats varied by 4% depending upon location of the star in the field.

Test 4: Color Terms and Zero-points

We used a Landolt SA near zenith to determine color terms and zeropoints. Lower case letters denote the instrumental magnitude; upper-case denotes the standard magnitude.

u-U=5.19-0.09(U-B)

b-B=3.62+0.00(B-V)

v-V=3.34+0.01(B-V)

r-R=3.21-01(v-R)

The color terms are all negligible except for a modest term at U, due primarily to the filter. We did find a horrendously large color term for the I filter, to the extent that we don't believe the results. We will remeasure it.
Count rates in e/sec/image
U B V R
3.19 17.2 17.5 19.8

Test 5: PSF Uniformity

We took images in good seeing unbinned to see how uniform the PSF is. The results are still to come.

(6) Software Issues

Nothing serious