Intro

• a 1.6 MHz chirp 100 useconds long was used. The chirp was repeated every 625 usecs. (duty cycle of  16%).
• The transmitted signal was then sampled (from the coupler in the waveguide).
• The radar interface (RI) was used. Its setup was:
• The signal was mixed to baseband with the 30 MHz mixers.
•  The I and Q signal were each  baseband filtered with a 2. MHz filter.
  
• the data taking started on a 1 second tick
• data was .2 usec  sampled (continuously) keeping  4bits in I and Q.

14may12: test run

• Page 1:spectrum, voltage
  
• top: 30 second average spectrum of chirp
• 160 usecs about each 100 usec chirp was used
• bottom:over plot the i,q voltages for the measured chirp
•  .1 seconds (160 ipps) were over plotted. black in I, red is Q
• Page 2: generate a chirp for pulse compression:
• Top: inphase, bottom: quadrature voltage samples
• The black traces are the measured data.
• The red trace is a generated chirp:
• 100 usecs long, chirping from -.8 to +.8 MHz.
• I only plotted half the chirp (to show more detail).
• Page 3: pulse compression using the generated chirp
• Top the voltage samples after pulse compression. (each point is .2 usecs).
• I did this for 1000 ipps and then averaged the I and Q separately.
• Bottom: pulse power averaging 1000 ipps.
• The fwhm is about  .6 usecs (1/1.6MHz)
  
• Page 4: phase variation at the compressed peak.
• The  phase from the I,Q samples was computed for the peak voltage (after compression).
• the .625Usec samples were smoothed to 15.6 milliseconds (1/60hz).
  
• The phase varies by 2 or 3 degrees over the 30 seconds of data.

Summary  14may12:

• Sampled transmitter data didn't show any obvious problems.
• The spectrum of the chirp only showed a few db dip. In previous runs it had been larger. This may have been a problem with the leakage signal
• The compressed pulse did not show a signal a 1usec delay (the dish to platform distance). This had been seen previously.
• with the dome at 19.69 degrees the stw may have been smaller. We'll see what happens with 16may12 run.

16may12: Run with lro satellite.

• Data acquisition at AO started at 12:59:59 and went for 50 minutes.
• lrosar_16may2012_all_binary.dat
• Data from the experiment can be found here
• all 50 minutes is stored in a single file (14 Gbytes).
  
• The .2 usecs 4bit sampled data (with the ri)
• The headers have been stripped off (leaving just binary data)
• Look at the Readme file for the file format.
• txPwrMeter.log - log of transmitter power.
• column 1 mjd of sample (every 10 seconds)
• column 2 : power meter sample in dbm
• attenuation factor is 82.8 dbm
• The meter reads average power. The duty cycle was 100/625=.16
  
• Kw=10^(-x.xxx + 82.8 - 30)*.1)/duty Cycle

Processing the sampled chirp:

• 1 second blocks of data (1600 ipps) were input and then averaged to give 1ipp sec
• 800 samples around the chirped signal were kept (out of the 3125 samples in an ipp).
• We ended up with 2298 seconds (averaged ipps).
  

Plotting the data:

• Plotting the voltage chirp vs time (.ps) (.pdf)
• The voltage chirps were averaged to 10 5 minute ipps before plotting:
• The 10 5 minute averages were plotted with offsets for display. The first 5 minutes starts at the bottom (black color).
• Page 1:
• Top: Inphase voltage vs time
• 2nd: quadrature voltage vs time
• You can see that the shape is changing with time.
  
• 3rd: Inphase voltage. blowup 35 usecs around where the chirp goes through 0hz.
• bottom: Quadrature voltage. blowup 35 usecs around where the chirps goes through 0 hz.
• Vertical lines are plotted at the peaks (in I and Q) for the first 5 minute average.
• You can see that the peak is not remaining in the same time slot for the 5 minutes.
  
• It moves by about 1.2 usecs over the 50 minutes. +/- freq both move away from DC.
  
• This is not a sampling problem since that would have +/- freq move in the same direction.
• This could be the bandwidth of the chirp generator changing with time. We checked the generator the next day and the 10MHz reference was plugged into it.
  
• Dana pointed out that this could also be a phase change in the signal or an lo (probably the signal).
• Page 2: Plot the start and end of the chirp
• 1,2: start of chirp for I and Q
• These look pretty stable for the 50 minutes
• 3,4: end of chirp for I and Q.
• These look like they may have moved by on .2 usec sample.
• Plotting the phase drift vs time (.ps) (.pdf)
• Page 1: How the average phase of the chirps changes with time.
  
• Top: Voltage chirp averaged over 50 minutes (voltage i,q)
• Bottom : The average phase (over the entire chirp) vs time for the 50 minutes,.
• The phase was computed for each sample of the chirp using the 1 second averaged chirps.
• This was done for each of the 3000 1 second averages.
• for each channel (in the chirp) phase jumps were corrected
• The average phase (over the chirp) was then computed for each 1 sec averaged sample
• The jumps in the chirp occur at 66 to 76 second intervals.
• Page 2:. How the phase of the chirp changes across the chirp and with time.
• The 1 second chirp phases were averaged to 1 minute resolution
• The phase as the first minute was subtracted from the other 49 minutes. this was to remove the phase across the chirp.
• The 50 1 minutes phase averages were then overplotted (with no offsets)
• The plot shows the change in phase of each channel with time relative to the start of the run.
• The top strip is the start of the experiment.
  
• Plotting the average spectra of the sampled chirp (.ps) (.pdf)
• The ipps were average to 5 minutes before computing the spectra.
• Top: Average spectra spaced every 5 minutes. Offsets have been added for display. Time starts at the bottom (black).
• There is a glitch at DC in the spectra at the bottom. This goes away at the end of the run.
• Bottom: blowup of average spectra around DC.
• The power increases and then goes close to 0.
• The dashed horizontal lines show the 0 power baseline.
• After looking at the data from the start of the run i thought that the signal was AC coupled somewhere (it isn't)
• After plotting all the data you can see that it is not AC coupled (unless it is changing with time!).
• Dana pointed out that this could come from our fixed 20 MHz leaking into the chirped 20 MHz.
• The chirped signal is +/-.8 MHz about 20MHz. So DC maps to 20MHz.
  
• there is a switch that sends our 20 MHz (from our dds ) or the chirped 20 MHz to the transmitter.
• If there is some leakage through this switch,then when the chirp passes through 20 MHz  you would have AO and LRO signals beating together. If the phase of these two changed with time you could have the dc component go from 0 to some larger than normal value.

Summary/Todo: 16may12:

• Find out where the 66-76  second glitches are coming from. Dana and ganesh thought that ron may have seen this problem before.
• Check the synth switch down in the control room to see what kind of leakage there is.
• Measure the phase drift in our system during the day while moving the azimuth and za.
• Take some data directly from the lro chirp generator (probably want to run it at 30 MHz  to use our baseband mixers.
• See if we still see the chirp changing over time.

06aug12 test run

    We did a test where lro generator was sampled by the ri (the sband transmitter was off). The setup was:

• The lro generator 20 Mhz output was upconverted to 30 Mhz and then downconverted and sampled by the ri.
• Signal path:
• lro output -> balanced mixer to 30Mhz -> 30Mhz ifAmp -> 30Mhz complex mixer -> 2MHz baseband filter on I and  Q -> opamps -> scope -> RI.
• The base  band signal was set to about +/- 2.5 Volts peak to peak (RI is +/- 2.5 volt input).
  
• RI setup:
• .2 usec sampling, 4 bits,  start on 10 sec tick, 1 buf/ipp, 10000 usec/ipp, 50000 samples/ipp.
• data was written via gio to /share/aeron5/lrotest_06aug2012.010
• The lro stepped through the following sets of pri's spending about 5 secs at each:
• 525,1600,850,1000,1150,1300,1450,1750,1900,1600 usecs.