Intro:

The setup:

• ch 430 txmitter/receiver with 420-442 filter installed on receive.
  
• cycle thru topside (60 seconds), power, and mracf
• topside: 20 millisecond ipp,rf=500 usecs
  
• mracf, power 10 seconds each, 10 millisecond ipps.
  
• 
  
• Rdev setup transmitter:
  
• 160 MHz clock, decimate by 5: 32 MHz Bw, 16 bits output, data  rate 32 MHz i,q.
  
•  txSamples: 20000
• HghtDelay: 30000 samples
  
• hghtSamples: 65000 samples starting (30000)/32e6=.9375 milliseconds =140.6 Km range.
• Set rms in counts at digitizer to about 80 counts.
• Rdev setup sine wave.
• Same setup as Tx (20 millisecond ipp).
  
• sine wave= 30MHz+ 16010 hz, amp=-2dbm.
• 16010 hz was chosen so that:
• We have about 10 cycles across the tx window.
•  the sine wave does not repeat at the ipp rate. This lets us see any skipped ipps.
  
• dig rms about 450.
• Take about 90 seconds of data (4500 ipps).
  
• Note:
  
• on 19oct09 the 160MHz clock for rdev was not locked to the station clock. This caused a freq offset of about 3 hz.
• On 20oct09 the test was repeated with the 160 MHz clock locked to the maser. The sine wave data shown is from 20oct09.
  
• Data files:
• 430tX: '/share/tdevs0/sasdr_gold.20091019.b0a.00700.pdev'
• SineWave: /share/tdevs0/sasdr.20091020.b0a.00000.pdev
• 
  

The Sine wave test:

• Page 1:Spectra of 1st ipp
• Top: spectra of 20000 tx samples.
• 2nd: spectra of 65000 data samples.
• 3rd: blowup of tx spectra showing sine wave
  
• 4th: blowup of data spectra showing sine wave.
• Page 2:Measure phase differences of the sampled sine wave for 4500 ipps.
• Compute the phase of the sine wave for all data and tx samples: phDataSmp[nsmp,nipps],phTxSmp[nsmp,nipps].
  
• Top: phDataSmp[0,ipp] - phTxSmp[0,ipp].  The expected phase difference is:
  
• Time DataSmp[0,ipp]-TxSmp[0,ipp]=(30000)/32e6= .9375 ms
• Period SineWave=1./16010=62.4610 usecs
• Excess phase= .9375/.0624610 mod 1. = .009375 cycles.
• Measured Phase dif=.0828 cycles
• phase Error=4.5862  useconds.
• I used dataOffsetCnts to compute the start of the data window.
• 2nd:phTxSmp[0,ipp+1]-phTxSmp[0,ipp]
  
• The measured phase difference .2 equals the expected phase difference.
• 3rd:phDataSmp[0,ipp+1]-phDataSmp[0,ipp]:
  
• The measured phase difference (.2) equals the expected phase difference.
• Page 3: Sampled and computed sine wave for 1 ipp:
• The sampled data (black) was plotted for 1200 usecs of the 3rd ipp. The time gap between the tx and data samples was included.
• A 16010 hz sine wave was computed (red) using the starting phase of the measured waveform.
• Top: The measured (black) and computed (red) sine wave.
• The green vertical bar is the end of the tx window. the blue vertical bar is the start of the data window.
• 2nd: Blowup start of Tx window.
• The computed and measured sine wave are in agreement.
• 3rd: Blowup end of Tx window. The measured and computed sine waves are still in phase.
• 4th: The start of the Data window
• The sampled data occures 4.6 usecs early (147 samples).

The topside data:

• average the high and low freqhopped data separately
  
• compute spectra every pulselen/2=250 useconds. This gave 7 spectra covering the height range 234 to 459 km in 37 km steps.

• Page 1: 20 second average of the first height
• The vertical scale is db about Tsys.
• The fall off in the spectrum is due to an analog filter in the crr2 receiver.
  
• Black is the up shifted spectra (+ 64khz)
• Red is the down shifted spectra (-64khz)
• Birdies that show no freq shift are not coming from the transmitted signal (eg. rfi..)
• Page 2: The 7 averaged heights after combining the frequency hop.
• Top: 420-440 MHz:
• Each height is offset for display purposes.
• 2nd: blowup around the ion line.
• 3rd: blowup around the down shifted plasma line
• 4th: blowup around the up shifted plasma line.
  

Summary:

• There are no skipped ipps for the 90 seconds of data (unless we skipped exactly 5 ipps).
  
• The ipp to ipp timing is correct for the tx to tx and data to data samples. The measurement is good to about 50 ns.
• Within 1 ipp the data window is starting about 4.6 usecs early.
• It is important that the rdev clock is locked to the external reference (station clock).
  
• The topside data looks ok.

<- page up
home_~phil