Alfa slow recovery times when hit by 1350 radar

03mar05 SECTIONS:
28feb05: project A9163 drift scans.
03mar05 project A2010 drift scans.
03mar05: problem is occurring before the downstairs if/lo.

links to plots:
    Scan 506000381: The first 480 sec drift (.ps)   (.pdf) (a1963)
    10 drifts over plotted by pixel (.ps) (.pdf) (a1963)
     total power versus time (.ps)   (.pdf) (a2010)
     dynamic spectra of the first 100 seconds (.gif) (a2010)
    the jumps/recovery changed when we swapped the if2 cables (.ps) (.pdf)

Intro:

Previous measurements have shown total power jumps and then slow recovery times when the alfa receiver is hit by the faa radar:

More recent data taken on 28feb05 by project a1963 and 03mar05 by project a2010 also had this problem (although not as severely).

28feb05: A9163 drift scans. (top)

A1963 is doing drift scans of 480 seconds centered at 1417 Mhz with a bandwidth of 12.5 Mhz, 1 second sampling, and using 9 level sampling. This setup has a 50 Mhz IF centered at 275 Mhz. On 28feb05, 10 of these drift scans were done. The plots show that some of these scans had total power jumps with slow recovery times. The jumps repeated at a 12 second period (the radar rotation period). The main beam of the radar sweeps by the observatory in about 80 milliseconds. With one second sampling, the strong saturation should not last for more than 2 one second samples.

Scan 506000381: The first 480 sec drift (.ps) (.pdf)

Fig 1: The total power over the 12.5 Mhz bandwidth vs time for the pixels 0 thru 6. Black is polA and red is polB. The total power has been divided by the median total power for the strip to make the y axis fractions of Tsys. The positive going 13 sec bumps are continuum sources drifting through the beam.

Pix0 polA (black) has large positive spikes every 12 seconds with a few second return. polB is clean

pix2 polA (black) has negative going spikes that recover quickly. This is normal compression.

pix3 polA and B both have positive spikes with long recoveries.

pix 4 polA has some positive going spikes with slow recovery.

Fig 2: this is a blowup covering 100 seconds of data. The blue dashed lines are every 12 seconds.

The 10 drifts over plotted by pixel (.ps) (.pdf)
Each plot has all 10 drifts over plotted by pixel. The drifts have been offset for plotting. Black in polA and red is polB.

The filter setup used is:

rf freq: 1417 mixed to first IF at 275 Mhz with hi side lo.
50 Mhz IF filter centered at 275 Mhz.
1350 Mhz is 67 Mhz below 1417 Mhz. Since the IF is flipped this puts 1350 and 317 Mhz in the IF. This is 42 Mhz above the edge of the IF Filter.

Compression should make the rest of the band decrease in power instead of increase in strength. Looking at the dynamic spectra for this data, the increase is across all frequency. There is either an increase in gain or an increase in noise while the radar points at ao. It is strange that is takes so long to recover. This must be some kind of thermal response.

processing: usr/a1963/28feb05_satur/28feb05.pro,28feb05_01.pro,28feb05_2.pro

03mar05 project a2010 drift scans. (top)

Project a2010 was doing 600 seconds drift scans with alfa. The setup was 100 Mhz bandwidth, 3 level sampling, 1 second sampling, with the center frequency close to 1384 Mhz. This experiment included the 1350 radar in the sampled band.

The first set of plots computes the total power between 1383 and 1420 Mhz and then plots the total power versus time (.ps) (.pdf) for the first scan of the evening.

Fig 1: Total power vs time for the first scan. The data has been divided by the median value to scale to Tsys. PolA pix0 and pix 3 show the jumps and slow recovery.
Fig 2: a blowup show seconds 200 through 300. The blue dashed lines are every 12 seconds (radar rotation period).

The second plot is a dynamic spectra of the first 100 seconds (.gif) of the above scan.

The 4 images are (top to bottom) pix0,1,2,3 polA. The power every 12 seconds in pixels 0a,3a extend across all frequencies.

03mar05 Problem is occurring before downstairs iflo.

On 03mar05 data was taken with alfa sitting at az=372,za=18 degrees, and alfa rotation angle 19 degrees. 100 Mhz bandwidth centered at 1380 Mhz was sampled once a second using 3 level sampling for 60 seconds. Total power was computed over the frequency range 1378 to 1415 Mhz and then normalized to Tsys.
The power jump and then slow recovery time was seen in pixel 0 polA. The downstairs IFLO starts at the fiber chassis. The signal then goes to a power splitter that sends the data to the if/lo racks and to the alfa monitoring chassis. The polA, B cables were swapped at the output of this power splitter. The polAD (dewar) now went through the polBIf2 part of the downstairs if/lo and wapps. The polBD went through the polAIf2 part of the equipment. The plots show how the jumps/recovery changed when we swapped the cables (.ps) (.pdf) :

Fig 1 top. This is the straight through mode. The jumps are in polB
Fig 1 bottom: The cables are now switched. The jumps moved to polA

The problem switched, so it is before the power splitter in the downstairs iflo that follows the fiber optic receiver chassis.

It is interesting to note that the pixel/pol that has the worst problem is changing with time:

28feb05: (a1963) pix0a, pix3a,b
03mar05:(a2010) pix0A, pix3A
03mar05:daytime test: pix0B,pix3A,pix4A

The faa radar strength getting into the feeds is a function of az, za, and alfa rotation angle. Different pixels/pols are having the problem depending on the input strength. So the problem is not isolated to one or to pixel/pols.

processing: x101/050303/alfa.pro

home_~phil