Bandpass structure 3 MHz about center.
An increase in the spectrum was
seen at the center of the spectrum. To test this the red noise source
was used to
The setup was:
The plots show the spectra with and
without the 6db pad (.ps) (.pdf):
- The output of the red noise source was split and sent to the
- The mixer chassis output was sent to Box 103 and box 104.
- The hi/lo lo's were 175MHz and 325MHz.
- We averaged 300 seconds of data. The band width was 170/8=22MHz
with 8192 channels (hi res mode with decimate by 8).
- The measurement was repeated with a 6db pad on the output of the
noise source. This would tell us if the increase in spectrum was from
noise or gain.
- The digitizers rms was 50 counts and 25 counts for the 2
measurements. The pshift parameter was 0x1555 and the hr_shift was 7
- I left the pshift paramter fixed for both measurements. The 6db
spectra probably needed more power. You can see a little scalloping in
units of 2^n caused by underflows in the butterflys.
- Page 1: The 300 second average spectra. Top plot is without the
pad, bottom plot is with the 6db pad in. The vertical scales are linear
(and measure the same range). The multiple plots are the spectra
polA,polB, hiband,lowband, of the two boxes.
- Page 2: Box103 Hi band polA.
- Top: black is full power. Blue has 6db pad. The two spectra
have been normalized to the median value in +/- 5 MHz. The vertical
scale is linear. You can see the curvature +/- 3 MHz.
- Center: horizontal blowup +/- 2MHz. The 2 spectra track each
other pretty well (at least the left half).
- Bottom: Ratio of spectraNoPad/spectra6dbPad using a db
- Looks like the pad was 5.7 db.
- The ratio is mostly flat. You do not see the curvature from
the individual spectra. This says that the curvature is gain related
and not noise being added.
- The dip at 0 to +100Khz needs to be looked into.
- The increase in spectra +/- 3MHz about the center is gain and not