CLASS Data analysis

TOD ANALYSIS

Data analysis is carried out in the Zeus computer server, with results (image files) archived in: /storage/drone/field_tests/TOCO/20220429/Analysis/Results_CLASS/

Currently, all data of the April 29th (2022) flights can be processed automatically based on detector crossing distance and time separation between crossings. Crossings are discarded if the 10 Hz frequency of the VPM is not found in the FFT or if it is below a certain threshold, which is related to whether or not the detector is on or properly biased.

The following plots show some of the results of the CLASS detector data analysis of the 29th of April 2022 flights:

Plot 1: Full TOD. The section corresponding to the current crossing (where the angular distance of the drone to the detector is below a certain threshold) is marked.
Plot 2: Extracted section of the Full TOD corresponding to ~12/20 seconds (150/90 GHz) around the time of the current crossing.
Plot 3: FFT of the extracted section. The location of the 47 Hz chopper frequency is marked in the plot.
Plot 4: The extracted section is filtered with a 3 Hz lowpass filter to retrieve the base signal of the crossing and with a 46-47 Hz bandpass filter to extract any component corresponding to the chopped signal.

Figure 1 - Flight 1 (150GHz), alt-scan, no chopping. Notice no power in the FFT at the 47 Hz mark and no chopping signal (orange, last plot).

Figure 2 - Flight 7 (150GHz), grid scan, chopping @47hz. Notice a peak in the FFT at the 47 Hz mark and the chopping is clearly visible only in the sections corresponding to the crossings of the drone through the detector (orange section, last plot).

Close up of FFT

Close-up of the FFT of Flight 7 (150 GHz): The chopping signal is shown at 46.6 Hz, which is consistent with the chopping frequency measured in the lab (as generated by the Raspberry Pi)

Close up of filtered signal

Close up of peak of detector r17c26 of Flight 7 (150 GHz): by applying a bandpass filter to the signal we can retrieve the section where the chopper signal was present.

TOD, FFT and filtered signals for Flight 8 (90 GHz)

Figure 4 - Flight 8 (90 GHz), grid scan, chopping @47hz, crossing for detector r01c00. The FFT shows no noticeable power at 47 Hz, while the bandpass filter does not show any chopper activity during the crossing.

OFFSET ANALYSIS

The following plots show the analysis of the offset measured between the peak of the TOD signal and the value of detector distance (usually a local minimum). The analysis disregards vertical motion of the drone and assumes that the 2D offset can be estimated with reasonable precision by measuring the the horizontal (azimuth) offset. For this, two methods are applied based on the type of scan:

Altscan: Measures the time delta between the detector minimum distance and peak power of the TOD waveform, the offset is then computed based on the scan speed of the telescope (1.5°/s) ->

Offset = (Ctime_det_mindist - Ctime_TODmaxP) [s] * (Scan speed) [deg/s]

Grid: Given that the telescope is staring, the offset is computed from the azimuth of the drone measured at the ctime of the minimum detector distance minus the azimuth of the drone at the ctime of the peak power of the TOD waveform ->

Offset = (DroneAz @Ctime_det_mindist [deg] - DroneAZ @Ctime_TODmaxP [deg])

Some results are shown below. Points outside of the y-axis limits are considered either outliers and/or a misidentification of a crossing.

Offsets plotted for Flight 4 and 7 (boresight -45°). Computed negative and positive offsets are expected for altscans (F4), while offsets for raster scans (F7) should remain relatively constant. The final plot shows an example of the measurement of one of the offsets for Flight 7.

The offsets can be explained by the physical position of the first mirror (VPM) on the telescope, which varies from the assumed center of telescope (center of the mount pillar, 8.5 meters high) during scans. This variation should not affect celestial objects, but does affect the drone given its 500-meter distance.

Crossings with Uncorrected Offsets

CLASS RESULTS (PRE-CORRECTION)

DRONE SOURCE BEAMS AS SEEN BY CLASS

SOURCE AS SEEN BY CLASS - 90 GHz

SOURCE AS SEEN BY CLASS - 150 GHz

CLASS SOLIDWORKS MODEL FOR DIFFERENT BORESIGHTS

Location of 90/150 GHz cameras confirmed on 04-aug-22 [conversation with Deniz Nunez from CLASS]

Boresight orientation confirmed on 05-aug-22 [CLASS bi-weekly meeting] (clic downfacing arrow to see the rotated boresight)

Crossings with Corrected Offsets

The coordinates of the drone as seen by each of the CLASS telescopes (mount 1 and 2) corrected with the actual position of the VPM during the scans allowed the offsets to become close to zero for all scans. This can be seen in the some of the new offset plots below:

ILLUMINATED DETECTORS

The following plots show some of the detectors hit during all flights, marked with an X when the drone illuminates a detector. The thicker the lines, the more hits each detector gets through the flight.

FLAGS ANALYSIS

The following plot shows the flag information of the RTK GPS during flights. For all flights, detected crossings occur only when flags are 16, 50 and 50 for flags 0, 1 and 2, respectively. The only time that flags change mid-flight is for F10, but this span of time did not match any crossings.

On Flight 10, although flag 2 is briefly different from flag 1, the first crossings actually occur when Flag 1 and 2 are equal to 50 (marked in the plot as "first crossing").

CLASS RESULTS (POST-CORRECTION)

Temperature and polarization of the source as seen by CLASS, with the main beam convolved out

Trajectory of the drone as seen in the detector array for Flight 3 (FLY313: 150 GHz, no chop)

CLASS RESULTS ANALYSIS

The apparent differences in width for some crossings seen in YunYang's CLASS data plot can be attributed to turn-around of the telescope during altaz scans:

The widened crossings seem to be a result of the telescope scanning the source while it was on its way to its initial scan position at ~500 meters (see Crossing at ~...4050 Ctime)

The CLASS result of removing crossings before and after the start of the scan is shown below:

SOURCE OFF NOISE ANALYSIS

The following plots show the data noise analysis for Flights 1 and 2 (150 and 90 GHz) with the source off, carried out on the 26th of april. A section that should not contain a crossing of the drone was analyzed in search of strong interference.

Left: Full TOD, Center: cropped TOD, Right: Relatively noise free FFT

Left: Full TOD, Center: cropped TOD, Right: Noisy FFT for the 90 GHz camera data

PAIRED DETECTOR ANALYSIS

Crossing for Flight 07 (-7.8 dBm) for two paired detectors normalized to the VPM signal power. Expected response (right) and flux-jump response (left). Orientation of 0° refers to a completely horizontal detector and 90° to a completely vertical detector (with respect to the ground).

( see other F7 flux jump examples)

*VPM signal power differs by ~15% between the two detectors.

*VPM signal power differs by a factor of 3 between the two detectors.

*VPM signal power differs by ~30% between the two detectors.

CLASS' F7 flux jump behavior explanation

The expected behavior of the flux jumps would be that higher power is measured in the detector that is aligned with the polarization of the source (vertical). Thus, the 0° aligned detector (with respect to the horizontal) should measure lower power than its counterpart. However, it would seem that the opposite effect is ocurring in the analyzed timestreams of Flight 07 (whose RF source output power was close to or saturating the detectors). In conversation with CLASS, this response is feasible considering that the VPM modulates radiation into each detector with a varying proportion that depends on its distance from its mirror (see the top right curves in the following picture). Then, either detector could measure higher power depending on the VPM's current grid-to-mirror distance:

DEMODULATION (47 Hz)

Attemps to demodulate the chopping signal have been carried out by obtaning obtaining the difference between two consecutive samples (blue and yellow curves) of the TOD, separated by a period of the chopping signal frequency (i.e, a separation of samples=0.215 s / (1/200 Hz) ). The resulting difference (green curve) is shown in the plots below (Differential Sgnals) and should correspond to the actual polarized signal emitted by the drone.