23 sep08 Total power vs time
23sep08 dynamic spectra
24sep08: long recovery vs alfa rotation angle.
The bias voltages for alfa were changed on 19sep08
to improve the recovery time from large pulses. On 23sep08 around 14:00
hours data was taken with alfa on the sky to see how long the recovery
took. The setup was:
- Data was taken at az=360,za=17.23. This was the position
that caused large recovery problems before (see za strips at az=360).
- Alfa was set to 0 rotation angle. The sky center freq was 1375
MHz. The two 172 MHz bands sent to pdev were centered at 1300 and
1450 MHz. The 100 MHz RF filter was not used.
- The pdev spectrometer was with:
- 2 172 MHz Bands per beam
- band centers 1300 MHz and 1450 MHz
- .1 second integrations
- 512 channels were used at az=360. This had no pfb overflows
- ADC overflow blanking was disabled.
The 100 MHz Rf filter (1390-1490 MHz) was not used
so the IF/LO downstream from alfa also saw the large radar
pulses. This is different from the pulsed noise tests that used
the filter (on the other hand, galfacts does not use the filter).
23sep08 Total power vs time:
The total power
vs time for all beams(.ps)
was computed for each .1 second spectra in the 1375 to 1525 MHz band:
- Frequency channels with rfi were excluded (by fitting to rms by
channel and throwing away outliers).
- PolA is black and polB is red. Each step in the plot is .1
- Long recovery times were only seen in polB. PolA did not show any.
- Page 1: Show the 7 beams:
- Beam 0 saw the most recovery.
- The initial jump was up to 5% of Tsys
- It took about 1 second to get back with 1% of the starting value
- It took up to 5 seconds to get back down to the original Tsys.
- Beam 2 showed the large jumps but it recovered quickly
(within .1 secs).
- Beam 3 polB went into compression during the pulse.
- Page 2: blowup of beam0, beam2
- Changing the bias caused sped up the recovery. It now gets within
1% of Tsys in about 1 second
- It can still take close to 10 seconds for the system to
- The selection of only PolB, and mainly Beam 0 is probably a
function of geometry. Rotating the turret could change which beam,pol
23sep08: DYNAMIC SPECTRA:
Dynamic spectra for the az=360, za=17.23 with adc
blanking off were made to show what was present in the
band. The processing for the images was:
- Polarizations A and B were summed.
- The median bandpass was used to flatten the image.
Notes on the images:
- The > 10 second continuum objects are sources drifting through
- When the radars saturate the system, the 90 deg phase for the I,Q
sampling breaks down giving images about DC.
- For the lower RF band, DC is at 1300 MHz.
- 1310 MHz is the image of the 1290 MHz radar
- 1270 MHz is an image of the 1330 MHz radar.
- 1420 MHz this is Galactic HI. It has been normalize to the
- 1350,1450. These are spike from the DC offset of the complex
24sep08: Long recovery versus
alfa rotation angle
On 24sep08 around 17:00 data was taken with alfa and
the pdev spectrometer to measure the long recovery time versus the
rotation angle of alfa. The setup was:
Total power vs time for
rotation angles 0,10,..60 of alfa (each page =1 angle) (.ps)
- telescope sitting at az=360, za=17.23
- 172 MHz bands taken centered at 1300 and 1450 MHz (rf cfr=1375).
- 512 channels, .1 second integrations, and adc blanking enabled.
- Data was taken for 60 seconds with alfa rotations =
- During the data taking lightning was present giving some unwanted
total power spikes.
power vs time for rotation angles 0,10,..60 of alfa (each page =1
Dynamic spectra of beam 0 at different
alfa rotation angles:
- The first set of plots has each page with the 7 beams at 1
- The 2nd set of plots has each page with 1 beam and the 7
- Some of the total power jumps are caused by lighting. The blue
(1330) and green (1290) marks below each plot show when the 1330, 1290
radar peaks occurred).
- As alfa rotates, different beams have larger/smaller recovery
spectra were made for beam 0 at the 7 different alfa rotation angles.
The images have the 1450 image on the left and the 1300 image on the
right. This is the same data as the total power plots above.
- The 23sep08 data showed only polB having the long recover. When
alfa is rotated, polA also shows the problem.