The EEV2 has been used with UHRF on two occasions to date - in June 2001 in slit mode, and in September 2001 in slicer mode. In summary, the EEV2 performed well, with the expected increase in efficiency over the Tek. Slightly higher spectral resolutions are obtained with the slicer. Strong fringing is seen at red wavelengths so the EEV should be used with caution in the red. Care is needed to avoid significant RON contribution due to the smaller pixels, and a slower readout speed should be considered. Detailed reports follow.
The EEV2 is now fully commissioned
with the UHRF.
Observations were made by Sean Ryan. The UHRF was in R=300,000 mode
with the slit selected. The central wavelength was 4554A. The full
EEV2, speed NORMAL, was used with x4 binning in the spatial direction.
HR 4963, a secondary standard in Hamuy et al. 1992, was observed on 29 June 2001 through a wide (9 arcsec) slit. The weather was near-photometric, and the seeing was 2 arcsec. Accompanying narrow slit arc, biases, darks and flats were also examined.
Biases
Five bias frames were coadded (image of bias frame - limits are lo=579, hi=581 ADU). Strong structure was seen at the left of the frame in the X direction (X=1-25) and at the start of the overscan (X=512-516 - X in binned x4 pixels). The hot column is seen in binned column X=462, with a jump of 4.8 ADU around Y=1527. Other structure is at a very low level (plot of bias structure in X). In the Y direction, there is no strong features. The pixel-to-pixel variation is greater, and there is a slight ramp, but variations are much less than 1 ADU (plot of bias structure in Y).
Readout noise was measured as 4.0 e-, significantly higher than the expected 3.4 e-., after removal of a slight ramp in Y. This may be an effect of binning.
Darks
5 darks were taken, each 1800s.
A cosmic ray rate of ~700 CRs/1800s was measured using BCLEAN in FIGARO. This is higher than the expected figure of 400 CRs but still much better than MITLL3 (1500 CRs).
The bias level in the overscan drifted up and down by +/- 2 ADU.
The dark current level was consistent in the 5 frames. IN the X direction, the bad column at X=462 (binned) is very evident, with some leakage in X=461, 463. A slight hump of ~0.7 ADU is seen in X=80 - 135 (binned) (plot of dark structure in X - some residual CRs are present).
The average dark current was 0.56 e-/binned pixel/1800s, or 0.14 e-/pix/1800s (unbinned) in agreement with the lab value. (cf the Tek at 0.50 e-/pix/1800s).
A strong ramp in dark current was observed in Y=1 - 2000, varying from 1.08 (Y=100) to 0.33 (Y=2040) and a minimum of 0.24 (Y=4046) e-/binned pixel/1800s (plot of dark structure in Y)
Scattered light
The observation of HR 4963 (4.2 mag) was investigated for scattered light, using X=375 - 380 which avoid the spectral orders (image showing scattered light - lo=0 hi=100 ADU). The scattered light is curved similarly to the blaze response, peaking between Y = 720 - 2700 at 90 - 100 e- in 300s (plot of Y variation of scattered light). Note that this level is ~ 1000 x higher than the dark current.
Flat fields
Inspection of flats is difficult because the UHRF setup aligns the spectral features, leaving the spectral orders sloping across the frame. The main features that show up are a large number of dark dust rings, typically producing ~7% extinction, though the worst feature (Y=2896) had 24% extinction. (image of normalised flat field showing inter-order structure; image of flat field showing structure in order, X=200-500, Y=1000-1400).
Further tests are required to see if these rings are temporary. They corrected well during flatfielding.
Focus
The star spectrum appeared slightly out of focus, and was clearly split (indicating a doughnut-shaped input star) for Y<200 (plot comparing the star profile at Y=200 and Y=3900).
Efficiency
The spectrum of HR 4963 was binned to 0.01 A pixels. After removal of the blaze response, the continuum level was 207160 ADU/0.01A. After correction for the exposure (300s) and the gain (1.3 e/ADU), the countrate was 863 ph/s/0.01A over 16A centred at 4553A (plot of HR 4963, in ph/s/0.01A).
This star is a secondary standard remeasured by Hamuy et al. (1992) in 16A bands. At 4548A m=4.206 and at 4564A m=4.203, indicating m(4553A) = 4.205 mag.
Comparing to measured efficiencies for UHRF + Tek + slicer at 4232A (from the manual), and correcting for the better efficiency of the EEV, we expect 37.3 ph/s/0.01A. Correcting for the slicer efficiency (17%) and the (poorly known) slicer aperture throughput in 1.8 arcsec seeing (20-30%) we expect a countrate of 730 -1000 ph/s/0.01A for the slit observation, in agreement with the observation.
Comparing the observed countrate and the Hamuy
et al. measurement, we obtain an absolute efficiency for the UHRF in slit
mode (no slit losses) of 4.7% at 4553A. This is likely to break up
as follows:
| EEV2 | 83% |
| Coude train | 62% |
| Spectrograph | 9.1% |
| TOTAL | 4.7% |
A number of bright B stars and Sirius A were observed on 26 September 2001. The weather was near-photometric, and the seeing was 2 - 3.5 arcsec.
Resolution
The laser with the V cross-dispersing grating focused to a width of 3.4 pixels or 46 um (measured by ABLINE in FIGARO), equivalent to R = 940,000.
This is the best resolution obtained during commissioning of the UHRF - with the Tek typical resolutions were more like 52 um or R=830,000. We are now probably at the limit of the instrumental profile.
Fringing
During setup, a flatfield was taken at KI 7699A to check the level of fringing with UHRF. As for other instruments, a strong and complex fringing pattern was detected. While the co-adding of a large number of columns (as is usual in slicer mode) will help cancel out the pattern, more tests are needed with on-sky red observations to determine the level of correction possible using flat fields, and to measure any degradation of the S/N. At present the EEV2 is not recommended for the UHRF for lambda > 6000A.
Efficiency
While the weather was photometric, seeing was poor and variable. This has a significant (and poorly determined) effect on the countrate due to the small input aperture of the slicer. Avoiding the known variables, the countrate of the following stars was measured (in order of observation). Details are from SIMBAD.
(RAS never quite finished this. A 2005A proposal from Seth Redfield however
indicates that a V=4.9 target of theirs gave S/N~100 - SDR, 6 Oct 2004)
| Star | Type | B (mag) | Exp (s) | Seeing | Counts (ADU) | Count Rate
ph/s/0.01A |
| HD 181869 | B8V | 3.867 | 2400 | 2.0 | ||
| HD 218045 | B9III | 2.45 | 2400 | 3.5 | ||
| HD 10144 | B3Vpe | 0.30 | 1200 | 3.0 | ||
| HD 16978 | B9V | 4.06 | 3741 | 2.0 | ||
| HD 27376 | B9V | 3.45 | 2400 | 2.0 | ||
| Sirius A | A1V | -1.46 | 3600 | 2.0 |
Cosmic rays were removed using BCLEAN in FIGARO. On average, the CR
rate was equivalent to 760 CRs / 1800s, similar to the June 01 result and
higher than the lab. result.
