links:
 the 4 bit histograms for all of the tests (.ps) (.pdf):
the results of each run followed by a simulation for that run (.ps) (.pdf)
the results for the runs whose output rms was close to the optimum value (.ps) (.pdf)

Sections:

running the tests
looking at the data
Summary

---

  Intro:  (top)

   See radar datataking with pdev  for a description of how the datataking works.

• 4 bit data was taken with different input rms voltages, decimations, and shifts.
• Histograms were made of each dataset
• rms results for all runs as well as a comparison of simulated runs were made.
• For those runs with 4bit rms close to optimum, the results were displayed as well had plots of the average bandpass.
  

---

Running the tests:  (top)

• 31 different runs were made.
• for each run, the rms from the scope was recorded
• I measured a 1.1 db drop scope to mock spectrometer.
• I varied the input rms voltage, decimation, and shift.
• The idl routine istat=pdevrmstmd() was used for a quick look at the 4bit rms and the average spectrum.
  

---

Looking at the results:  (top)

• idl was used to look at the data
• @phil, @pdevinit, @masinit
• pdevtmdchk() would input the file, compute the rms and various histograms.
  
• I was testing various decimation, shifts, and input voltages so the resulting rms was not alway the optimum
• I also used this data to verify that my 4bit time domain simulator worked.

The first set of plots shows the 4 bit histograms for all of the tests (.ps) (.pdf):

• Each page shows histograms for a single decimation
• The colors on a page differentiate the various input Voltages, and shifts
• The optimum 4bit rms should be 2.98

The 2nd set plots the results of each run followed by a simulation for that run (.ps) (.pdf)

• Page 1: the results of each run
• Top: input rms voltage measured at scope
• 2nd : measured 4bit rms
• 3rd. Upshift used
• bottom:decimation used.
• Page 2: simulation of the runs:
• Top: measured and simulated rms
• black is measured
• red is simulated (with no prefiltering)
• green: simulation with boxcar smoothing of data before filtering
  
• This give a little better match since the input signal was 20-25 MHz wide in a 160 MHz sampling bandwidth.
• middle: Measured rms/ simulated rms (no prefiltering)
• the first decimation is low because of the missing power in the input signal
• the other runs were within 10 % of the measured value.
• Bottom: measured rms/simulated rms (prefilter by 3)
• This does a little better job of matching the input at dec=5.

The last plots show the results for the runs whose output rms was close to the optimum value (.ps) (.pdf)

• All of these had 4bit rms's within 10% of the optimum.
• Top: input voltage for each decimation.
• 2nd: the measured rms (black). The green line is the simulated value using the measured input parameters.
• 3rd: the upshift used for the measurements
• bottom: The input voltage that we should have used to get the optimum 4bit rms for this decimation and shift.
• Black is the measured values, green is the simulation.

---

SUMMARY:   (top)

• radar 4bit time domain IF sampling can be done with the mock spectrometers
• 12, 16 bits are used through the sampling, mixing, filtering. At the last stage it is clipped to 4 bits
• The simulator gets within about 10% of the measured values.
• I/O rates:
• decimate = 5  (32 MHz bandwidth)
• 1 pol: 32M Bytes/secs
• 2 pol: 64 M Bytes/sec
  
• The table below shows the  decimations, down shifts, and input voltages that will give the optimum rms for 4 bit sampling:
•  

• bandwidth MHz	decimation	shift	input Voltage	Recs/Second
  				1pol	2pol
  32	5	13	.059	30.52	61
  20	8	12	.078	19.07	38.15
  10	16	12	.055	9.54	19.07
  5.16	31	11	.080	4.92	9.84
  3.2	50	11	.062	3.05	6.10
  1.6	100	10	.083	1.53	3.05
  .32	500	9	.075	.305	.610
  .156250	1024	9	.054	.149	.298

  
  

processing: x101/131031/mocktmd.pro, tmdplothist.pro,goodsettings.pro