wideband rfi in alfa during galfacts
Galfacts (a2130) reported rfi during their run on 14feb13
(and into the morning of 15feb13). Their setup is:
The first plots show
the total power cal deflections for the 1450 band (averaged over
the 172 MHz centered at 1450) (.pdf)
- alfa covering all of the receiver bandwidth using 2 172 MHz
bands (4096 chan each) in the mocks
- "Stokes" is taken : polA,PolB, stokes U and V.
- data is dumped at 1 millisecond.
- The cal is winked every 20 milliseconds (on then 20
- Two sets of files are created after datataking:
- 1 millisec resolution decimated to 128 freq chan over 172
- 200 millisec resolution calon, caloff bufs with full freq
- They typically run for 6-8 hours at the meridian with the dome
nodding from 2 to 19.5 degrees at close to slew rate.
- Part of the data monitoring plots the cal deflections avged to
200 millisec resolution.
- The plotted power is averaged over the 172 MHz of the
band centered at 1450 MHz.
- The vertical scale is linear in power, the horizontal scale is
minutes since start of experiment
- Frame 1,2 (polA,polB) total power vs time. black is
calOn, red is calOff. time resolution is 1 second
- Frame 3,4 (polA,polB) calOn - caloff 172 MHz band centered at
1450 MHz. The data has been averaged over the 172 MHz
- The rfi is present from minute 160 -> 200, and minute
The dynamic spectra show spectral
density for polA, calOff data for file 0955 (.gif):
The next plot show the .2 second averaged spectra for seconds
27 thru 28.8 of file 955 (.ps) (.pdf)
- The top image is the 1375 MHz band, the bottom image is the
1450 MHz band.
- The time resolution is .2 seconds. The image durate is 64
- White is maximum power. The image has been flattened using the
These plots show bm0 polA total
power vs time and za for files 950 thru 986 (.ps) (.pdf)
- Page 1: PolA, Page 2: PolB
- For each page:
- top 1300 MHz band, bottom: 1450 MHz band
- Each .2 second spectra is over plotted (bottom to top) with
an off set for display)
- The rfi covers most of the band. It is strongest in polA from
1300 to 1460 Mhz. for polB is it a bit wider.
- This comvers minutes 165 thru 203 of the first set of plots.
- top frame: za vs time for the filess
- each pass up or down in za is plotted in a different color
- 2nd frame: 1300 MHz band total power vs time. the tsys
increase with za is visible
- 3rd frame: 1300 MHz band total power vs za. Each pass is
overplotted in a separate color
- bottom frame: 1450 MHz band total power vs za.
- The strong rfi seems to group at various za's.
- There are some za's where the rfi is not present for all the
The final plots look at how close the
gps satellites were to the dome during the rfi (.ps) (.pdf)
- The gps satellites transmit a strong signal at
- L1= 1575.4 Mhz. This is outside the alfa band and could
cause saturation in the dewar without the primary showing up
in the data.
- L2C=1227 Mhz. this is inside the alfa band and would be seen
if it was a gps satellite transmitting this signal.
- We did not see the 1227 signall
- PRN 03 is svn 33 (blkIIA) launched in 1996. It does
not have the L2 signal
- PRN 19 is svn 50 (blkIIR) launched in 2004. This
satellite is also before L2 was turned on
- So not seeing L2 does not rule out the rfi being a gps
- Page1: za,az plots for gps constellation for 130214,130215
- gps03 transits ao close to za=5 at 22:15:58 AST
- gps19 transits ao near za=5 at 23:07:00 AST
- Page 2: angular separation between satellite and dome beam on
- The data is taken from galfacts files number 950 to 986
(minutes 165 - 203 of the first set of plots).
- Top: separation vs time for sat GPS03 (black) and GPS19
- Bottom: bm0, polA 1450Mhz band total power vs time.
- You can see compare this with the angular separate of the
- Page: Total power bm0,polA,1450Band vs angular separate sat,
- Top is gps03
- bottom is gps19
- The data is not grouped at smaller angular distances.
SUMMARY:what is it.
- It is not the mock spectrometer having some of the high order
output bits flipping:
- This normally happens in 1 fgpa chip. The rfi is in all
- It doesn't happen when the full bandwidth is used (the
problem is in the digital filter).
- The rfi is broad band.
- it stretches across most of alfa's 300 MHz.
- There is no "strongest peak" in the spectra.
- The radars don't look different than normal during this rfi.
- If the rfi was out of band
- it would get filtered by the waveguide at the low
- The frequency above 1520 is filtered after the dewar by a
- So a strong birdie above 1520 (say 1575) could be
saturating the dewar.
- Nothing was seen during this time in the hilltop monitoring
- but.. the hilltop system has a high tsys and it doesn't see
satellites very well.
- It probably is not in the dome.
- If it was in the dome, i would be surprised to see no rfi at
a few za's over multiple passes (unless the rfi period was
synced with the za strip period).
- A GPS satellite:
- the 1575 Mhz signal could have saturated the alfa lna
without being seen in the bandpass. Since the satellites
didn't transmit L2, not seeing L2 does not rule out gps
- Previous gps close
approaches caused saturation, but in one alfa beam at a
time. Here we see it in all alfa beams.
- The rfi strength versus angular distance does not show
stronger signal for smaller angles.
- It's a real coincidence that we had the two satellites
passing close overhead when the problem occurred.
- I tried looking at azimuth dependence of the gps satellite
to see if the signal was strongest when the satellite was
perpendicular to a beam of the triangle. The correlation
was not consistant.
- So the problem could have been a gps satellite...
- A meteor.
- The rfi started at minute 160 utc.
- A meteor hit in russia at minute 9:20 local russian
time (3:20 utc, or 200 min UTC).
- The russian site was at utc + 6, we are at utc -4 so
we are 10 hours behind the point of contact.
- Preliminary orbits calculations show that the meteor orbit
was inside the orbit of earth prior to hitting.
- The meteor did not pass over AO 40 minutes prior to
- The meteor was moving at 13-20 km/sec prior to impact.
- The rfi started at UTC 160 minutes.
- Using 18km/sec for its velocity, the meteor would have
travelled 40*60*18=43000km from utc 160 to 200 (this is
relative to the earth).
- Don't know the exact orbit of the meteor.
- Things that are 40000 km from earth
- average earth radii: 6371 km.
- van allen belts:
- outer belt 3-10 earth radii (from surface)13000 to 60000
km above earth's surface
- inner belt: 1.1 to 3 earth radii
- geostationary satellites:
- 35768 km above the earth (5.64 earth radii above