Authors: G.R. Meurer, G. Hartig, and D. Leviton
Purpose: One of the primary goals of the first Thermal Vacuum test of ACS (TV1, February to March, 1999) was to measure the the far ultraviolet (UV) throughput of the instrument's Solar Blind Channel. Measurements were made at two wavelengths 1236Å and 1469Å. The troughputs at these wavelengths were measured through all relevant filters and were monitored as a function of time throughout the test. Here we present the results of the throughput measurements and compare them with model expectations.
Experimental setup: The throughput was measured repeatedly during the first thermal vacuum campaign at Goddard Space Flight Center. During the campaign, ACS was illuminated by STUFF (STimulus for Ultraviolet Flat Fields; Leviton 1999). STUFF's calibration mode was used for all observations described here; the optical layout of this mode is shown in Fig. 1. Two lamps were available for calibration purposes, one producing strong line emission of Kr 1236 Å the other Xe 1469 Å emission. These will be referred to as the Krypton (Kr) and Xenon (Xe) lamps. Either lamp could be selected for observation by remotely moving a stage to align the lamp along the chief ray of the instrument. Other movable stages allowed (a) the light to be attenuated through a mesh filter (to insure safety of the SBC); (b) collimating pinhole pairs to be selected; and (c) a photomultiplier tube (PMT) radiometer to intercept the beam, thus to record the flux entering the instrument.
For monochromatic emission, and assuming negligible atmospheric absorption (measurements were made in vacuum) the count rate recorded by the PMT and the SBC are:
IPMT = FSTUFF QPMT ,
ISBC = FSTUFF TM1 TM2 Tfilter QMAMA ,
ISBC = FSTUFF TSBC QMAMA .
In the above, all quantities are implicitly evaluated at the relevant wavelength. FSTUFF is the flux emerging from the pinhole assembly; QPMT and QMAMA are the Detector Quantum Efficiencies (DQE) of the PMT and the MAMA; TM1, TM2, Tfilter are the throughputs of the ACS mirrors M1 and M2 and the relevant SBC filters. Combined, the three throughputs are TSBC. IPMT and ISBC are the quantities directly measured in the experiment. Their ratio is then
ISBC / IPMT = TSBC QMAMA / QPMT .
Quantities on the right side of the equation can be modeled using component level measurements. Throughput versus wavelength of SBC's MAMA were measured at the detector and reported by Ford et al (1998). (Note: somewhat higher DQE values are reported for the STF7 MAMA by Argabright, 1996). Laboratory filter throughputs are available on the WWW through the ACS filters page. These as well as the reflectance of the M1 and M2 mirrors are combined to yield TSBC QMAMA by the SYNPHOT package (IRAF/STSDAS) as implemented on the STScI computers. The PMT used to provide the reference flux measurements was the PMT "borrowed" from RAS/CAL. It is a CsTe PMT and was calibrated in 1996 as part of the STIS preflight calibration procedure. The calibration as a function of wavelength is reported by Ebbets (1996), and the calibration curve is reproduced in Fig. 2 (this figure was produced with the IDL program pmt_qe.pro). Spline fits through the throughput and DQE curves versus wavelength were performed by the program thruplot.pro (relevant data files are in the same directory) in order to determine the relevant quantities at 1236 Å and 1469 Å. Model components and predictions are reported in Table 1.
|Item||Throughput or DQE||Notes|
|QMAMA||0.1904||0.1303||DQE. Ford et al. (1998)|
|QPMT||0.1329||0.0848||DQE Ebbets (1996)|
|TSBC(F115LP) QMAMA||0.105||0.085||SYNPHOT model throughput.|
|TSBC(F125LP) QMAMA||0.050||0.084||SYNPHOT model throughput.|
|TSBC(F140LP) QMAMA||0.079||SYNPHOT model throughput.|
|TSBC(F122M) QMAMA||0.016||0.001||SYNPHOT model throughput.|
|TSBC(PR110L) QMAMA||0.063||0.079||SYNPHOT model throughput.|
|TSBC(PR130L) QMAMA||0.051||0.082||SYNPHOT model throughput.|
Our calculations assume monochromatic emission. However, we know from visual inspection (i.e. looking) that the lamps produce optical emission, hence they are not strictly monochromatic. Instead they produce emission longwards of the line of interest. Furthermore, the sensitivity of the CsTe PMT extends longwards of the CsI MAMA. In order to allow the flux emerging from STUFF to be more monochromatic, bandpass filters were placed in front of the lamps, partially covering the lamp exit windows. Then, for a pinhole observation, by judiciously moving the lamp stage one can select whether filtered or unfiltered light emerging from the lamp is selected. The filter used for the Kr lamp was a flight spare Ly-alpha filter (central wavelength, width = 1190, ~120 Å), while the Xe lamp had a filter having central wavelength, width = 1450, ~100 Å in front of it.
Data: The data consists of pinhole "splat" images with a paired dark image. The matched 1.0mm pinhole pair was used for almost all measurements reported here. Bracketing each SBC observation, the PMT count rates were "acquired" with STUFF. This means that the PMT stage was driven into the ACS light path, a PMT reading was taken and then the PMT was removed from the light path. Thus each observation has a before and after radiometer signal which we average and report here. For the dark exposures, a block filter is placed in front of the SBC and the fold mirror is put in the HRC position. For the PMT dark measurements, nominally a block filter is placed in front of the STUFF lamp, thus the PMT is also reading a dark (non-illuminated) reading. However, frequently this was not done. For those cases we adopt the mean PMT dark current of 1.7 c/s (observed range 0.0 to 3.0 c/s) and an uncertainty of 1.7 c/s.
The individual throughput measurements using the Kr 1236Å and Xe 1469Å lamps are given in Tables 2a and 2b respectively. The first two columns list the ACS database entry IDs of the splat and dark images. MJD is the Modified Julian Date. In Table 2a the Kr 1236Å lamp current Ilamp is given. The lamp current was always set to 0.20 Amps for Xe 1469Å measurements. The index of the mesh filter in front of the lamp assembly is given in the STUFF atten. column. The smaller the index, the lower the attenuation. The exposure time for the splat image is texp. In most cases the dark image exposure was the same length. The total MAMA count rate after subtraction of the dark rate is reported as the global rate. The ratio of the measured SBC and PMT dark subtracted count rates is given in the SBC/PMT column. The errors reported in the global rate and SBC/PMT columns are calculated from the measured number of counts from the SBC (and PMT) assuming Poissonian statistics.
The quantities reported in these tables were extracted from the images and the ACS_LOG database using the IDL program uvthru_calc.pro. N.B. We found that occasionally the STUFF items in the image headers were incorrect, whereas the ACS_LOG database does not appear to be corrupted. Image reduction involved three steps. First the rows (599:605) effected by the dead annode were replaced from the mean of rows 595:597 and 606:608. This was done to both the splat and dark images. Then the total counts were found. Finally dark was subtracted from the measured counts. The bad row replacement increases ISBC by 1.0% to 1.8% compared to the raw count rate. This can be seen in the old version of Table 2 where the data reduction did not have the bad row replacement.
Lamp behavior. Both calibration lamps showed evidence of degradation during the TV test. The Xe lamp showed a 65% decline in count rate (at fixed STUFF attenuation). This can be seen in the global rate column for entries 3475:4939 in Table 2b. The decline in SBC counts was mirrored with a declining PMT count rate and hence ISBC/IPMT was not effected.
A more serious problem is evident with the Kr lamp. For it the ISBC/IPMT ratio varied with lamp current. This is shown in Fig. 3 (plot made with IDL program PLOT_KR_VS_I in PLOT_RESULTS.PRO). All the data plotted were obtained on Sunday March 14 through the F115LP filter, and with ACS running on MEB side 2. The most likely cause for the anomaly is contamination by another Kr line at 1165Å. In validation testing, this line was shown to have a fractional strength of ~0.001 relative to the 1236Å line, at an operating current of 0.2A (as measured with a CsTe PMT). We suspect that this line became brighter during the TV testing, and that the relative strength of the line varied with current. Fig. 3 suggests that the line only becomes excited for I > ~ 0.1 Amp. Unfortunately we could not confirm this hypothesis; before we were able to obtain post TV calibrated lamp spectra, both lamps failed.
The data in Fig. 3 were obtained near the end of the TV campaign while the TV chamber was being warmed up. The log notes that the PMT tube was 23-24°C rather than its more normal 20°C. The PMT temperature probably does not effect ISBC/IPMT. However, the warm chamber and the late date of the observations suggest that the Kr lamp degradation was more severe. This could explain why the mean ISBC/IPMT = 0.639 at 0.15 Amp at this epoch, 11% lower than the average during earlier portions of the test.
We did some tests to see if other observational parameters could effect ISBC/IPMT. We find that count rate doesn't matter (compare entries 5897 and 5903), position of lamp relative to pinhole does not matter (compare entries 5897 and 5905), presence or absence of lamp filter does not matter (compare 5897 and 5908), and pinhole size does not matter (compare 5897 and 5910).
Throughput monitor. Throughput measurements were measured repeatedly during the TV campaign, using both lamps. Figure 4 shows the variation of Kr 1236Å throughput (plot made with IDL program PLOT_MONITOR_KR in PLOT_RESULTS.PRO). All data here were taken with a lamp current of 0.15 Amp. Only two of the Kr 1236Å data points taken on MJD 51251 are shown (entries 5899 and 5909) so as not to overly weight this date. The full Kr dataset from this day is shown in Fig. 3 and discussed above. Figure 5 shows the Xe 1469Å throughput results (plot made with IDL program PLOT_MONITOR_XE in PLOT_RESULTS.PRO).
The mean ± dispersion of the Kr data is ISBC/IPMT = 0.705 ± 0.030 (only the data shown in Figure 4). For the Xe data ISBC/IPMT = 0.964 ± 0.011. In both cases the scatter in the ISBC/IPMT measurements is larger than the Poissonian uncertainties, which are shown as vertical errorbars (usually smaller than the symbol) in Figures 3 - 5. Much of this is due to short time scale (minute) variations in lamp output. The PMT measurements taken immediately before and after each SBC observation only agree to about 1%, a factor of a 3-5 higher than the Poisonian statistics expectations. This is enough to explain all of the scatter in the Xe data. The scatter in the Kr data is larger: 4.3% of the mean. It probably results from varying contamination of the Kr spectrum by the 1165Å feature.
The Xe throughput shows no evidence of degradation over the 22 days of measurement. The Kr lamp data in Fig. 4 hints at some throughput data but only because of the last two data points. As suggested above this is more likely to be due to degradation in the lamps spectral properties, rather than in the instrument throughput.
Comparison with model predictions. The dotted lines in Figs. 3 - 5 show the model/component level expectations, as discussed above. These are ISBC/IPMT = 0.790, 1.002 for monochromatic emission at 1236Å and 1469Å respectively. The average measured ratio with the Kr lamp at 0.15 Amp is 11% below this expectation. However, Fig. 3 shows that for 0.1 Amp or less of current, ISBC/IPMT = 0.837 for the Kr lamp, 5% higher than the model expectations. This value is more likely to represent monochromatic 1236Å emission. If the 1165Å line were to dominate the Kr spectrum, then we expect ISBC/IPMT = ~0.26 (the wavelength requires extrapolation from component level measurements, hence the model is more uncertain). For the Xe lamp the mean measured count ratio is only 4% below the model predictions.
Comparison with CEI specifications. The Contract End Item paragraphs 4.4-017 and 4.4-018 specify minimum assembly level throughputs of 0.100 and 0.047 at 1216Å and 1500Å, respectively, for the SBC. Our Kr and Xe lamp measurements are sufficiently close to these wavelengths that they can test whether the instrument meets the specification. The assembly level specification includes the M1 and M2 mirrors, the MAMA, but not the filter. Hence with the filter and with the calibrated PMT the CEI specifications are equivalent to ISBC/IPMT > TspecTFilter/QPMT = 0.550, 0.468 at 1236Å and 1469Å. These levels are plotted in Figs. 3-5.
The instrument easily meets its 1500Å specifications (Fig. 5). Assuming no error in QPMT then TM1TM2QMAMA = 0.099 at 1469Å. For the Kr lamp, ISBC/IPMT is better than specifications except at a lamp current of 0.2 Amp (Fig. 3). It is these high current levels that are most effected by spectral degradation. Considering only currents of 0.1 Amp or less, and assuming no error in QPMT then TM1TM2QMAMA = 0.152 at ~1236Å.
Results for all filters. Table 3 shows the synphot model predictions and actual measured count ratios for all relevant filters. For the Kr lamp data, only the measurements with a lamp current of 0.15 Amp are presented. For the F115LP measurements, already discussed thoroughly, only the mean measured ISBC/IPMT is given. Only one observation is available for each of the other filters.
|Filter||wavelength||ISBC/IPMT||measured / model|
The measured/model ratio ranges from 0.47 to 0.97 for the Kr lamp data. This is due to the 1165Å contamination. Agreement is worse for the filters which have much lower 1165Å throughput than 1236Å throughput. Conceivably these data can be used to model the throughput at both 1165Å and 1236Å, if the filter transmission curves are well constrained.
The situation is much nicer for the Xe lamp data. At 1469Å the measured/model ratio ranges from 0.91 to 0.96 (mean ± dispersion = 0.932 ± 0.020). This indicates that the average 7% difference from the model is not in random differences in the filter transmission curves.
Conclusions: During the first TV campaign throughput measurements using the STUFF apparatus were made at two wavelengths by comparing count rates of the Solar Blind Camera and a calibrated PhotoMultiplier Tube. The measurements of the Kr lamp indicate that the 1236Å emission is contaminated by another Kr line at 1165Å. By lowering the Kr lamp current this contamination is minimized. The Xe lamp data showed no evidence of contamination of the 1469Å emission. In both cases the measured count ratios are within 10% of model predictions and easily meet Contract End Item specifications. The measured throughput of the combined M1 and M2 mirrors and the SBC is 0.152 at ~1236Å and 0.099 at 1469Å. There was no throughput degradation over 22 days of monitoring to better than 8% at 1236Å and to better than 2% at 1469Å.
Acknowledgements: Francesca Boffi kindly calculated the model ACS throughputs at 1236Å and 1469Å.