Intro

• sband narrow
• pfs 20Mhz bw, 2bit sampling, both pols
• 40 seconds of radar data followed by 15secs cal off and then 15 seconds calon
• the high correlated cal was used. The calvalues are: [7.04,6.75 K]
• the data file /proj/radar/2002PZ39/2020/feb11/p05_cal/data20200212015506.000' was used to look at the data.
• The program pfs_unpack was used to unpack chan 1 of the data file.

Voltage histogram and Tsys

The first set of plots show the voltage histograms and Tsys computation (.ps) (.pdf)

• Page 1: voltage histograms of the 2bit data
• black,green: cal off histogram
• red, blue: cal on histogram
• The digital levels are   -3,-1,1,3
• The optimum sigma/level is 1.004 for 2 bit sampling
• The measured values were about .9 so the levels were a little low.
• Page 2: smoothed total power data to 1 millisecond
• top calon,caloff in a/d^2 units (voltage^2)
  
• bottom: calOn,calOff in degK
• the scaling was calDegK*(calOff/(calOn-calOff))
• Tsys came out as 90K. This was the OC Tsys which is normally around 25K
• the other pols came around 300K (according to dylan). It's normal temp has been around 66K

Total power histograms of sampled and simulated data.

• The sampled data
• simulated multibit data
• simulated data clipped to 2 bits.
• To scale the simulated data to the same levels as the sampled data
• compute rms of the sampled voltage data
• force the simulated voltage data to have the same rms
• clip to 2 bits
• generate power, smooth/decimate and the plot and fit a gaussian to the histogram
• recompute the voltage scaling so the width of the clipped power gaussian matches the width of the sampled power gaussian (for the cal off).
• The redo the computations.

Plots of the power histograms and gaussian fits:

•  The histogram used .1 width bins.
  
• The histogram fit was:
• z=(x-A1)/A2
  
• hfit=A0 * exp(-(z^2)/2) + A3
  

Fitting results:

• the gaussian fits were pretty good to the sampled data, so there wasn't terrible compression.
• I computed the  simulated data scaling so that the center of the measured power histogram was the same as the 2bit simulated power center.
  
• The measured data saw the width get narrower as the cal was on. Since the average value increased, the width should also have been greater.
• The simulated 2bit data showed the width increasing as the cal came on.

processing: x101/200212/pfscal.pro  pfshist.pro  pfshistsim.pro

14feb20 Checking compression levels in the if2 setup

• staying  10db below the 1db compression show no compression.

The sbn receiver was setup and the power levels were adjusted to give the standard histogram for the pfs. The resulting values were:

The left amp was well below any compression. the right amp could have had up to .1db of compression.

We then did a tsys using the output of the 260Mhz amps (and the square log detectors). This can be compared with the tsys values measured with
the standard tsysall measurement done later in the day:

• The polA measurement was close.
  
• the polB measurement was farther off than expected.
• The voltage output was set so tsys was 1 volt.
  
• We were able to read to a few millivolts, so a 6.75 kelvin increase (10%) would be well within the measuring error.
• Since polB is messed up anyway, I'd wait until it gets fixed, and then recheck the levels.
  

SUMMARY:

• Tsys measured from the pfs 2bit 20mhz cal was a factor of 3 higher than normal
• The was seen in polA and polB
• looking at the voltage/power histograms there was not an obvious large compression
• I doubt of the 2 level correction would give you a factor of 3 to 4 power correction.
• we tested the compression in the standard sband radar pfs 20mhz 2bit setup.
• we saw no compression in polA at the output of the 260 Mhz amplifiers. (polB is questionable)
• I guess we could use the RI to 2bit sample a cal test to see if it sees a large increase in Tsys.