Progect a2335 (GASS)
reported wideband rfi while using lband wide and the interrim
correlator (with 1 second dumps) on 04may05. While investigating the
problem, the rfi was seen to have a 1.2 Hz periodicty. This pointed to
the cross head which has a 1.2 hz period.
Data was taken with lbw and the mock spectrometers
on 05may09 to investigate the problem with higher time resolution and
bandwidth. The setup was:
- Telescope sitting at 270,10.
- 6 172 Mhz bands were taken centered at: 1200,1300,1400,1500,1600,
and 1700 Mhz.
- 8192 channels (20Khz resolution) was used for the low time
resolution (.1 second dumps).
- 4096 channels (40Khz resolution) was used for the high time
resolution (.001 second dumps).
.1 second time resolution:
Data was recorded for about an hour. One of the
worst 60 second spans of data was then selected to show how the
interference appeared across the lbw band of 1100 to 1750 Mhz.
Dynamic
spectra
- The total power was computed for each 172 Mhz band and then
plotted vs time (.1 second resolution).
- The total power was transformed and the magnitude plotted
vs frequency to check on the time periodicity of the rfi.
- Each page has 2 frequency bands.
- Each frequency bands shows the total power vs time followed by
the spectrum of the total power.
- For some band the rfi in the 172 Mhz band was greater than 5*tsys.
- The spectra show a peak at 1.2 hz and 2.4 hz.
.001 second time resolution.
1 millisecond by 4096 channel, 172 Mhz data was also
take for about 60 seconds. The rfi was not as strong during this data
set.
- The data were averaged to .1 second and then dynamic spectra were
made to show the frequency extent of the rfi in the 1115-1285 Mhz
region where the total power was computed.
Total power vs
time and periodicty of rfi (.ps) (
.pdf):
- Page 1: total power vs time:
- Black is polA and red is polB
- The dashed purple lines are the section used for the following
plot.
- Top: The full 60 seconds with .1 second time resolution.
- It was strongest during the time 45 to 55 seconds.
- Middle: 45 to 55 seconds with .01 second time resolution.
- You start to see regions where the rfi goes away completely.
- Bottom: 48.5 to 49.5 seconds with .001 second time resolution.
- The rfi turned off for about .15 seconds at 49.1 seconds.
- Page 2: spectrum of total power for 1211 to 1225 Mhz.
- Top: 500 hz .08 Hz resolution
- 2nd: 125 to 135 Hz .016 Hz resolution showing spikes at 128.5
and 129.5 hz.
- 3rd : 232 to 242 Hz .016 Hz resolution showing spike at 237.8
hz.
- 4th : 0 to 10 hz showing 1.2 hz cross head frequency and it's
harmonics.
Summary:
- Wide band rfi is coming from the cross head or at least being
caused by the crosshead motion.
- It was first seen in the x111 data around the first of May. It
was not there on the 20th of april.
- the rfi showed the following frequency dependence:
- 1.2 hz and it's harmonics. This is the frequency of the
cross head.
- 128.5, 129.5, 237.8 hz. These might be higher frequencies that
were aliased by the 500 hz max frequency.
- When the cross head driver cable was removed, the rfi went away.
When the cross head driver cable was plugged back in, the rfi started
again in about 10 to 30 seconds.
- Switching cross head drivers from the nrao driver to
another one made no difference. The rfi remained. This also switched
the cross head driver cable.
- If the rfi was not present, disconnecting and reconnecting the
driver cable could make the rfi appear again.
- Banging on the lbw can had little affect on the rfi so it is
probably not something loose inside the dewar. Banging close to where
the cross head cable plugged into the cross head did make the rfi go
away.
- We covered the mouth of the lbw horn with a mesh. This did not
affect the strength of the rfi. So the rfi was not coming in through
the horn.
- High frequency spikes were seen on top of the 1.2 hz signal of
the cross head. When these spikes went away, the wideband rfi
disappeared.
- High time resolution (.001 secs) showed that the rfi would
go away completely for up to 100 milliseconds.
- 07may09 afternoon...Replacing the cross head fixed the problem:
- There was still a 1.2 hz variation in the total power.
- The high frequency spikes were gone and the wide band rfi went
away.
processing: x101/090505/masdata.pro,
masdata_hr.pro
