Rdev filter shapes with 11/ 17db roll off coef

The coefficients for the rdev spectrometer were updated on 07mar08 to give more roll off. Data was taken with:

decimate by 8 (20 MHz) bandwidth
16 bit output data
shiftenable=5
noise (a/d rms=43 counts) , and noise (43 counts) + sine waves (31 + 35MHz) with a/d rms=150 counts.
32K length spectra were computed and then averaged for 13 seconds.

Page 1 shows the spectra from just the noise. The top plot is a linear scale, the bottom plot is in db. The bottom plot shows a roll off 17db from the center to the edge of the band.
Page 2 has a 31 MHz and 35 MHz sine wave added to the noise. After mixing to baseband the signals end up at 1 and 5 MHz.

On 06mar08 filter coefficients were installed with an 11 db roll off (for the 20 MHz bandwidth). The setup was:

decimate by 8, 16 bit output, output shiftenable = 7 and 5
Dana's noise source with a -4db sine wave at 31MHz and 35 MHz. The red noise source and the 260 Amps in the receiver room.
32k length spectra averaged for 13 seconds.

The plots show the 11 db roll off filter shapes (.ps) (.pdf):

Page 1: dana's noise source with A/D rms of 43 noise counts + sine waves of 31 and 35 MHz (which are digitally mixed to 1,5 MHz). The A/D rms of the noise + sinewaves was 150 counts. The rms of the output data was 1274 counts (with an output shift enable of 7). the top plot is linear while the bottom plot has a db scale. The birdie frequencies show up at the correct locations.
Page 2: red noise source with 260 MHz amps in control room. The A/D rms was 213 counts on the noise (there were no sine waves injected). The filter roll off is about 12 db.
Page 3: Danas noise source with noise only. The A/D rms was 43 counts. The rms of the output data was 182 counts. The filter roll off is 11 to 12 db.
Page 4: danas noise source with A/D rms of 43 counts. Output shift enable was changed from 7 down to 5. The rms on the output data is 45.8. We still have the 12 db roll off. The rms noise within +/- 3 MHz of the center of the spectra was computed after bandpass normalizing with a 3rd order polynomial fit. The measured rms was .112 Tsys. The expected rms from the radiometer equation was: .111 Tsys.

processing: x101/080307/rdevtest.pro,x101/080306/rdevtest.pro