Project TBASERDN040126 January 26, 2004 Baseline tests Q-band rx. Weather is calm, intermittant snow, mostly cloudy. ~20F at start (16:30 ET). Tsys approx. 100K. Setup with Spectrometer, 50MHz, 8 samplers. 2 beams, 2 pols, 43.0 GHz. Other four samplers looking at IF Rack noise source. L1 -> ODM1 -> OR1 -> CM1 -> CF1 -> ACS J? (bank 1, rcvr 1) R1 -> ODM3 -> OR3 -> CM5 -> CF5 -> ACS J? (bank 1, rcvr 2) NS -> ODM2 -> OR2 -> CM3 -> CF3 -> ACS J? (bank 2, rcvr 1) NS -> ODM4 -> OR4 -> CM7 -> CF7 -> ACS J? (bank 2, rcvr 2) L2 -> ODM5 -> OR5 -> CM9 -> CF9 -> ACS J? (bank 3, rcvr 1) R2 -> ODM7 -> OR7 -> CM13 -> CF13 -> ACS J? (bank 3, rcvr 2) NS -> ODM6 -> OR6 -> CM11 -> CF11 -> ACS J? (bank 4, rcvr 1) NS -> ODM8 -> OR8 -> CM15 -> CF15 -> ACS J? (bank 4, rcvr 2) ACS in 4 bank, 2 sampler/bank mode, 10sec integrations. GBT stationary at access. Scans 3-10, T=60s. Scans 11-16, T=300s. Changing setup to 800MHz BW. ACS in 4 bank, 2 samplers/bank mode. L1 -> ODM1 -> OR1 -> CM1 -> SF1 -> ACS J? (bank 1, rcvr 1) R1 -> ODM3 -> OR3 -> CM5 -> SF2 -> ACS J? (bank 1, rcvr 2) NS -> ODM2 -> OR2 -> CM2 -> SF2 -> ACS J? (bank 2, rcvr 1) NS -> ODM4 -> OR4 -> CM6 -> SF6 -> ACS J? (bank 2, rcvr 2) L2 -> ODM5 -> OR5 -> CM11 -> SF11 -> ACS J? (bank 3, rcvr 1) R2 -> ODM7 -> OR7 -> CM15 -> SF15 -> ACS J? (bank 3, rcvr 2) NS -> ODM6 -> OR6 -> CM12 -> SF12 -> ACS J? (bank 4, rcvr 1) NS -> ODM8 -> OR8 -> CM16 -> SF16 -> ACS J? (bank 4, rcvr 2) T=120s. Scans 17-19 Spectrometer CPU Earth transporter down. Rebooted and reconfigured. Scans 20-23 As above. Scans 24-29 Terminated 3/4 OR outputs on converter Rack A. Scans 30-33 Crossed S11 and S12 in IF Rack. Rebalanced. ANALYSIS: 50MHz scans: The L1 and R1 baselines are quite flat, with a just discernable wave in the 5 minute scans. The mean total power does jump around by a percent or so. The NS channels all exhibit flat baselines. Only the ODM/OR 4 channel from the noise source shows TP changes more than a few tenths of a percent. L2 and R2 are much less spectrally stable. File qL2L1s15_16.ps shows the baseline for scans 15 and 16 for L2 (red) and L1 (green). File qL2s16recs.ps shows baselines taking record 1 of scan 16 as the reference spectrum, and each 10 second record in sequence as the signal spectrum. Note the baseline pertubation that progresses during the scan time. The vertical scale of all these files is percent of Ta/Tsys. 800MHz scans: These are much worse in appearance, showing ripples with roughly 20-40 MHz period and amplitudes up to 3%. File qL1R1s24_28.ps shows the baselines for both L1 and R1 for scans 24-25 (red), 25-26 (green), 26-27 (blue), and 27-28 (cyan). (All plots are scaled by 100 so the vertical axis is percent of Ta/Tsys. Also, no mean was subtracted nor offset applied so the mean of each trace indicates how much the average total power changed between the two scans.) File qL2R2s24_27.ps shows the baselines for both L2 and R2 for scans 24-25 (red), 25-26 (green), and 26-27 (blue). Since the ups and downs in mean total power generally tracks between the front-end channels, it is likely the major source of TP variations is the antenna (sky) temperature. The noise source channels were generally very flat, at least on the same scale. See q4nss24_27.ps. Scans 30-33 showed that the dominant ripple moved with the front-end channel when the IF Rack transfer switches are crossed. See qnsL2s30_31.ps for an example. It is interesting that the noise source channel (bank 3, rcvr 1 in this case) does show some ripple at the lower frequencies. Termination of 3 of 4 Optical Receivers showed no improvement in the baselines, so the ~60MHz ripple found earlier correlated with the Equipment Room temperature seems not to be a major factor. File qL1s27_28records.ps shows how the L1 baseline progresses for each 10 second integration in scans 27 and 28. For each record in these two scans, progressing from the bottom, scan 27 record 1 is taken as the reference. Each trace is offset by 0.5%. File qL2s27records.ps shows how the baseline progresses thru scan 27 for channel L2. The traces are offset by 5% in this plot. File qL2s24_29records.ps shows a longer sequence (in this case taking every other record). Note that after about 7.5 minutes, the spectral shape returns to that in record 1 of scan 24. This periodicity in the baselines was also seen in data taken last week. Summary: The 800MHz baselines are clearly not acceptable, even in the best channels L1 and R1. I suspect the baseline ripple originates in the front-end (or at least before the IF Rack) since parallel channels carrying the IF Rack noise source behaved much better. During subsequent tests, we should take data with the front-end beam switches crossed to see if the poorer performance in beam 2 follows the switch position or not.