6 May 2004 Art Freiley MS 303-210 Jet Propulsion Laboratory 4800 Oak Grove Drive Pasadena, CA 91109-8099 Re: DSS 14 Tests for Mars Express Bistatic Radar (8 April 2004) Art: I wasn't able to complete my analysis of the DSS 14 noise data we collected before leaving on my trip. But it's mostly done now. I only need a bit more information on things like Tlna and Tfollowup to estimate TND and Top. You gave me a couple values earlier, but that's not enough to take care of all four channels - unless we make some simplifying assumptions. X-RCP Test (including general analysis procedure): I'll walk you through what I did for X-RCP. The other three were handled similarly, including S-LCP. We never attempted measurements on S-LCP; but the RSR was running, and noise diodes were visible at times. Fig. 1a provides an overview of the X-RCP test. We collected X-RCP data for 2506 seconds (22:00:29 - 22:42:15). The upper left panel shows a histogram of the RSR sample values. I see no problem with dynamic range; in fact, we probably could have increased the input levels a bit. The upper right panel shows power spectra. These are 1024-point (complex) FFT's, converted to power, and averaged over 3 minutes. Time increases from bottom to top. You can see from the different spacings that the noise power goes up and down; but there's nothing really exciting here. The lower left panel shows power in the individual sample values. Complex voltages are converted to power, then summed over one second and plotted on a dB scale versus time. Here you can clearly see that we were making changes. For the next step, I computed spectra, but averaged them over 10 seconds (sum of 244 power spectra) rather than the 3 minutes used in Fig. 1a. This gives 250 time steps of 10 seconds each. I averaged 128 frequency bins within each spectrum; this leaves an 8-point spectrum with each point (window) representing 3125 Hz. The results are tabulated in the first 250 rows of Excel spreadsheet NOISE2PRT_XR.xls. Columns E and L include the roll-off, so those are about half the values in Columns F-K. Fig. 1b shows the linear power in Columns E-L versus time (compare Fig. 1a). Columns F-K are indistinguishable in Fig. 1b; Columns E and L are the two lines that are at about half strength. In NOISE2PRT_XR.xls I have color-coded entries in the first 250 rows. Blue denotes values when we were looking at cold sky, green is when the noise diode was added to cold sky, orange is the ambient load (alone), and red is the ambient load plus noise diode. Rows that are black are ambiguous (two or more states occurred within a 10 second interval) or corrupted (such as by RFI during the S-RCP test). There is some subjectivity in assigning colors; but, where there was doubt, I tended to assign black - which means the values were excluded from further analysis. The X-RCP data can be divided into 14 distinct time intervals as follows (see also Fig. 1a, lower left panel). Time steps correspond to rows in NOISE2PRT_XR.xls; only 'colored' entries (not black) are listed. Note that X- RCP data collection continued through the X-LCP test. Interval Time Step Test X-RCP Configuration 1 001-016 X-RCP Ambient Load 2 019-031 X-RCP Ambient Load + Noise Diode 3 033-048 X-RCP Cold Sky + Noise Diode (w/cloud) 4 050-063 X-RCP Cold Sky (w/cloud) 5 066-068 X-RCP Ambient Load 6 069-072 X-RCP Ambient Load + Noise Diode 7 074-077 X-RCP Cold Sky + Noise Diode (w/cloud) 8 079-082 X-RCP Cold Sky (w/cloud) 9 100-113 Transition Ambient Load 10 137-145 Transition Cold Sky (overcast) 11 147-175 X-LCP Ambient Load 12 177-203 X-LCP Cold Sky (overcast) 13 205-229 X-LCP Ambient Load 14 232-250 X-LCP Cold Sky (overcast) Rows 253-266 in the spreadsheet give the average values for each of the 8 'windows' in each of the 14 time intervals. Columns 268-276 give some of the corresponding standard deviations; these were calculated to monitor data quality rather than to provide rigor in the analysis. Rows 279-298 Column E give the sum of the power in the 8 windows for each of the time intervals. Note that six of the rows (e.g., AMB09+ND) link to activities occurring in the X-LCP test; since the noise diode was not seen in X-RCP, these fields have no values. Note also, however, from lines 263 and 265 that something is slightly different on X-RCP during the latter stages of the X-LCP test. In rows 279-298 (Columns G-H) I have entered your values for Tphys, Tlna, Tfollowup, and Tamb, where Tamb = Tphys + Tlna + Tfollowup (1) Since you gave me two values for the X1 ambient load Tphys (11.5C and 11.8C) and two values for the X2 ambient load Tphys (12.5C and 12.9C), I interpolated to get the five values shown here assuming X1 and X2 would be the same. By looking at various power ratios between X-RCP configurations, you can get estimates for the noise diode temperature TND and Top. These are shown in rows 279-298 (Columns J-K). For example, the noise diode level can be estimated from the change in power between Interval 1 and Interval 2: TND02 = (E280/E279-1)*H282 (2) Top in Interval 4 can be estimated by comparing the cold sky power in Interval 4 with the ambient load level in Interval 1: Top = H282*E282/E279 (3) The noise diode can also be estimated from the change in power between Intervals 3 and 4: TND03 = (E281/E282-1)*K280 (4) The four noise diode estimates calculated (there are additional possibilities) give TND = 21.80+/-1.05 (5) which doesn't seem too bad except that you found TND = 19.5 and 18.0 from the noise instrumentation during the X-RCP tests. I got five estimates for Top (23.97, 27.05, 19.60, 19.85, and 19.64). Because of the cloud, I'm inclined to go with the latter three, which are very tightly grouped but come mainly from the X-LCP test period. Top = 19.70+/-0.13 (6) Those are lower than your Top (21.4 and 24.8) by several degrees whereas my estimates from the X-RCP portion of the test (presumably the same inputs as yours) are several degrees higher than your Top. To summarize: Intervals Simpson Top Freiley Top Simpson TND Freiley TND 1-4 23.97 21.4 22.58 19.5 5-8 27.05 24.8 21.03 18.0 9-10 19.60 11-12 19.85 13-14 19.64 X-LCP Test For X-LCP 14 intervals can also be defined (Figs. 3a and 3b); but they are slightly different from the X-RCP intervals. The recording period was 2504 seconds, so spreadsheet NOISE2PRT_XL.xls contains 250 rows of 'window' data. The results can be summarized as follows: Intervals Simpson Top Freiley Top Simpson TND Freiley TND 1-2 22.38 3-4 23.39 5-6 18.36 7-10 18.48 16.9 17.47 15.5 11-14 18.49 17.0 16.81 15.0 Again, I would take the values from intervals 9-14 more seriously because the large cloud apparently was no longer sitting over the antenna. But my numbers differ from yours by 1.5-2.0 degrees. S-RCP Test There are also 14 intervals for S-RCP (Figs. 2a and 2b); but some of the data appear to be unusable because of the RFI. The recording period was 2344 seconds, so spreadsheet NOISE2PRT_SR.xls has 234 rows of 'window' data. We have no values for Tlna and Tfollowup. The results can be summarized as follows: Intervals Simpson Top Freiley Top Simpson TND Freiley TND 1-2 20.39 3-6 7-10 19.87 20.56 11-14 20.09 16.15 S-LCP Test The S-LCP results are very limited; there are only five intervals -- all with ambient load, two with the addition of the noise diode (Figs. 4a and 4b). It is not clear why noise diode energy should be visible in S-LCP when only the S-RCP diode was being enabled. We have no values for Tphys, Tlna, and Tfollowup. With these cautions, the results can be summarized as follows: Intervals Simpson Top Freiley Top Simpson TND Freiley TND 1-2 11.86 3-3 11.35 Concluding Remarks There may be some differences in procedure. There may also be some differences in the data being analyzed. It would be worth comparing these results in more detail. And I have done nothing with linearity. Regards, Dick Simpson