4.2 Science-grade Arrays
4.2.6 Skydips
to substrate modes, which is expected to be an∼10% effect. It also does not account for inefficiency due to the fact that the actual slots have finite length and the impedance varies near the ends, which is expected to be another∼10% effect that will be largest for the lower frequency bands.
Lyot Accounts for the loss due to the truncation of the beam at the Lyot stop. It is estimated using Zemax simulations of the MUSIC optics [138].
Filters Accounts for the loss and reflections of the dielectric, metal-mesh, and infrared shader filters. The loss of each filter is estimated using the methods and references outlined in Sayers et al. [138]. The reflections at the interface of each filter is estimated via simulation.1
The product of these efficiencies is presented in theTotalcolumn of Table 4.7. We expect that the MUSIC in- strument will have single-polarization optical efficiencies between 7% and 12%, depending on the observing band.
In general, the optical efficiencies inferred from hot/cold are in good agreement with the expected optical efficiencies. There does appear to be a slight band dependent discrepancy. In Band 0 and Band 1 the measured efficiency is larger than expected, whereas in Band 2 and Band 3 the measured efficiency is smaller than expected. The quoted uncertainties are simply the dispersion observed across the detectors of a given band.
There is also systematic uncertainty in the measured values that arise from our uncertainty in the assumed values of the recombination coefficient and the thickness of the Al section. This systematic uncertainty is not included in the quoted uncertainties. However, it should effect all bands approximately equally, and therefore would not explain the band dependent discrepancy. The true efficiencies would be 30% greater than the quoted values if zero Al is etched away during fabrication instead of the assumed 15 nm. The true efficiencies would be 50% less than the quoted values if 30 nm of Al is etched away instead of the assumed 15 nm and if the recombination coefficient is 7.1µm3sec−1instead of the assumed 9.4µm3sec−1. Note that the
+30%
−50%systematic errors correspond to the absolute boundaries of what we believe are reasonable values for the recombination coefficient and thickness. Without independent measurements of the recombination coefficient and thickness of the MKIDs employed in MUSIC, it is difficult to determine the overall normalization of the band dependent optical efficiency to better than approximately 50% accuracy.
opening in the hex plate of the telescope was covered with a large piece of EccosorbR and a final IQ sweep was collected to act as the room-temperature load reference measurement. The ambient temperature during the reference measurement wasTamb=285.5 K. The dataset is analyzed according to the procedure outlined in Section 3.2.5.
Analysis of the dark resonator skydip data yields a measurement of the direct absorption spillover frac- tion. We find that the median spillover fraction for the dark resonators is fspill,dir=0.28. However, there is significant variation among the dark resonators, with spillover fractions ranging from 0.20 to 0.36. This is much larger than the∼0.01 measurement uncertainty on the individual measurements, and is most likely due to a dependence on focal plane position. In other words, the direct absorption spillover fraction is most likely larger for resonators positioned near the edge of the focal plane than for those positioned near the center.
However, we cannot confirm this hypothesis, because it is difficult to extract a beam location for the dark resonators. When fitting the skydip data for the antenna coupled resonators, we place a prior on the direct absorption spillover fraction that accounts for this large variation. We also measureτdir=0.18±0.05 for the dark resonators.
The best fit antenna spillover fractions fspill,antare presented in Figure 4.10 as a function of the azimuthal offset of the detector from the center of the focal plane. We observe a clear increase in the antenna spillover fractions for detectors situated towards the edge of the focal plane. This increase is largest in the lower frequency bands. The black dashed line denotes the expected spillover fraction due to the expected inef- ficiencies of the MUSIC optics. It assumes 1% absorption for each of the five mirrors in the optics chain and Ruze scattering from the primary mirror. In the central region of the focal plane we measure an excess antenna spillover fraction of [0.08, 0.04, 0.02, 0.10] for the four observing bands.
We use the Atmospheric Transmission at Microwaves (ATM) model by Pardo et al. [129] to infer the atmospheric transmission spectrumATM(ν)from the value ofτ225measured by the CSO tipping radiometer.
The optical depth in band is then predicted as
τant=−ln R
FTS(ν)ATM(ν)dν RFTS(ν)dν
, (4.3)
whereFTS(ν)denotes the bandpass measured via FTS. We compare this prediction to the resulting best-fit estimate ofτantfrom the skydips. We find excellent agreement in Bands 0, 1, and 2. This gives us confidence that for the three lowest frequency bands we can useτ225and the ATM model to infer the sky loading for any given observation. In Band 3, the predictedτantis approximately 30% larger than the value measured via skydip. Recall that the bandpasses for the MUSIC detectors are shifted by 3% to lower frequency due to an increase in the ratioε/d of the Si3N4dielectric relative to the engineering-grade tiles used to calibrate the filter geometry. Because of this shift, Band 3 overlaps with the waterline at 325 GHz, which makes the predictedτantvery sensitive to both the ATM model and the bandpass measured via FTS. Therefore, we do not believe that we can accurately predict the loading and responsivity for the Band 3 resonators at this time.
Figure 4.10: Spillover fraction fspill,antas a function of the azimuthal offset of the detector from the center of the focal plane. Each panel corresponds to a different observing band. The black dashed line denotes the expected spillover fraction based on the expected inefficiencies of the MUSIC optics (see text).
Figure 4.11: Optical depthτant as a function of the azimuthal offset of the detector from the center of the focal plane. Each panel corresponds to a different observing band. The black dashed line denotes the ex- pected optical depth based on the averageτ225=0.1036 reported by the CSO over the course of the skydip measurement, theATMmodel for the atmospheric transmission, and the detector bandpasses measured via FTS.
Table 4.8: Median background loading, calculated using detectors of the specified band on the A2 and B2 arrays. All temperatures are referred to the cryostat window. Texc is determined from the hot/cold data,Tspill
is determined from the skydip data, and Tsky is determined from the FTS bandpass measurements and the ATM model assumingCPW=1.3 mm. These three temperatures sum to give the total loadingTload. Poptis the equivalent optical power incident on the MKID.
Band Texc[K] Tspill[K] Tsky[K] Tload[K] Popt[pW]
0 30 45 10 85 3.3
1 20 30 20 70 5.5
2 25 30 25 80 3.9
3 40 55 70 165 4.9