p2445 multi freq pulsar observations

Intro:

01apr09: Unstable power levels reported in 430 data.

• I used files 8-20. Each file has 178 seconds of data.
• Wapp1 was configured as :
  
• 50 MHz Bw, 3 level, add polarizations, 192 lags centered at 432 MHz, and 32 usec sampling.
  
• The 420-442 filter was in.
• The source J1713+0747 was being tracked.
• 4.5 milliSec period, dm=16, S400=36 milliJy.

• For each file:
• Input the file averaging the spectra to 1 second.
• Compute the total power for each second.
• Also compute an average spectra for each file.
  
• Create a dynamic spectra covering the 13 files with 1 second resolution:
  
• Normalize the 178 spectra of each file to their median value.
• Compute the rms noise in the flatten image over channels 50:140 (those that had signal). When doing the computation, throw out any outliers.
  
• Scale the image so the full scale lookup table covers +/-2 sigma.
  

• Page 1: total power vs time.
• The total power was computed over the 50 MHz bandwidth after averaging in time to 1 second.
• The vertical dashed lines are the start of each file.
  
• TOP: Db scale (above min value).
• There are large power jumps in file 8 and  file 18. In file 18 it is about 100 times the system total power (although its probably no longer linear). It lasts for about 10 seconds.
  
• Bottom: total power vs time. Linear scale
• The variation in the first 6 files is from Tsys changing from the  za dependence.
• The total power looks pretty good.. i don't see any jumps.
• For optimum statistics, the level should be around 1. Having it at 2.2 will require an  increase in the integration time of about 40% (to get the same sensitivity as level=1).
  
• Page 2: Averaged spectra:
• An average spectra for each file (178 seconds) was computed.
• TOP: overplot the 13 average spectra.
• the large birdie at 440.6 continues up to about 600 (where tsys average=2)
  
• Bottom:
• Blowup the vertical scale for the average spectra.
• A small vertical offset was added to each spectra for display purposes.
• The large black spikes (in file 8 and 18) are all coming from the 440.6 MHz birdie.
• The other spectra look ok..
  

• The dynamic spectra concatenates all 13 files.
• The full scale range is +/- 2 sigma on the noise.
  
• The steps every 178 seconds for the bottom most files are caused by the za dependence. The spectra from each file were normalized to their median value so this will cause a jump while the za is high.
• The edges 400 to 415 and 443-460 look uniform since there was no signal there (the 420-442 MHz filter cut it out).
• the rfi below 420 or above 442 should be ignored. They look stronger since the bandpass normalization will bring them up. Looking at the average spectra, there is no power there.
  
• The 440.6 MHz birdie is causing the horizontal line at 10 and 1800 seconds.
• There is also  intermittent birdies at 440 and about 437 MHz
  

Summary:

• There is a large birdie at 440.6 MHz.
• it can last for up to 10 seconds.
• The signal  saturates the wapps. I'm not sure where in the chain the intermods are being created.
  
• The interference  is hams talking to each other (we've listened to them). We are probably getting the signal from the repeater (rather than straight from the hams). It's kind of funny that someone would be talking between 3 and 5 am (when the data was taken).
  
• Other than the 440.6 MHz birdie and a few other biridies, the data looks ok..
• I'm not sure why the setup is:
• 3 level summed pols. Why not use 9 level summed pols. The gui supports it.. I don't know whether anyone used it. It would give better signal to noise.. It guess it should be tested first.
• 50 MHz. There is not 50 MHz of bandwidth available. 25 MHz would cover what is there (and it might cut down the birdies on the edge a little.
• The power level should be closer to 1. Maybe this is a problem trying to get the vlbi, asp, and  wapp power levels all adjusted at the same time. Being at 2.2 requires an increase of about 40% in the integration time for the same sensitivity.
• The reported instability in the 430 MHz data.
• I have an idea what might have happened to make the continuum (ri) total power data look unstable:
• The ri data was taken with a 5 MHz filter to get rid of unwanted rfi. This filter is centered at 260 MHz.
• The wapps in 50 MHz mode center the band at 275 MHz (432).
• The band at 275 is not flipped so 15 MHz below 432 would be 417 MHz.
• Taking 5 MHz around 417 would probably give funny results. The 420-442 rf filter has filtered out much of the power, and there are a few interference sources down there.
• Of course i'm assuming that the ri data was taken with the same band as the wapp configuration.. i don't really know if this is what happened.
  

 home_~phil