Most of the commissioning data were taken during bright moonlight,
hence our night sky brightnesses are not very useful for estimating
dark sky conditions.
The spectrophotometric standard star BD+28° 4211 (V = 10.55) was
observed through a 6 arcsec wide slit in subarcsecond seeing.
The figure shows the image as it is displayed on the data-taking screen, with the bluest order (15) at the left and the reddest (order 6) at right. In each order, blue is at the top and red at the bottom. Clicking on an order (except for order 14, for which we are missing data) will bring up the extracted spectrum, to give an idea of the count rate at various wavelengths in each order. Units on these plots are in DN/pixel; multiply by 1.3/60 to get into units of electrons/pixel/sec.
Also, below are some tabulated numbers which give the count rates at the peaks of the orders.
(Note that order 14 is missing because the automated reduction routine had problems with the bad columns at the bottom of the CCD. In the spectrum, the wavelengths affected by the bad columns are repeated in order 13 at the top, so while S/N at these wavelengths suffers, there is still full wavelength coverage.)
Below is the spectrum in electrons/second/pixel. Note that the dispersion in low-dispersion mode ranges almost linearly from 1 Å/pixel at 4000 Å to 10 Å/pixel at 1 micron.
Below is the sky spectrum summed over 23 pixels (= 3.5 arcsec), which was the extraction window for the above BD+28 4211 spectrum. Each pixel is 0.154 arcsec. Again the y-axis is electrons/sec/pixel.
In doing signal-to-noise calculations, note that you generally won't observe faint targets with a 6-arcsec slit width. Hence the sky values below would be divided by 23 pixels, multiplied by the number of pixels along the slit you expect to use for the target spectrum, and multiply by the ratio of slit width to sixe arcsec. As an example, using a 1.5 arcsec slit width and a 3 arcsec extraction window you would expect sky signals of (3/3.5) * (1.5/6) = 0.214 times the values given below.
Using the BVRI facility filters for ESI, the Mark A photometric standard field was observed. ESI in imaging mode appears to be 10%-20% more efficient than LRIS. The following table compares the performance of ESI and LRIS: