cell phone harmonics in lbw while xmiting from inside
People have been asking how cell phones transmissions
affect observations done on the telescope. To get an unambiguous signal,
cell phone transmissions were done from inside the dome (on the stairwell
leading into the turret floor).
The telescope was parked at az=279, za=10.
lband wide was set to linear polarization (no hybrid).
Three 50 Mhz bands centered at 1625, 1675, and 1725 Mhz were sampled once
a second using the interim correlator. 2048 channels gave 48Khz resolution
(after hanning smoothing).
The first attempt at connecting from inside the turret room (with the doors
closed) failed because of no signal. Stepping out onto the stairwell worked
The cell phone used was from cingular.
Data was taken for 440 seconds. During this time there were two separate
transmission cycles (turn on, connect, talk, turn off).
Data from the first 90 seconds (when there was no cell
phone transmission) were used for a bandpass correction. The processing
The median of the first 90 seconds was computed
corinterprfi() was used to interpolate across rfi in the median bandpass.
Each 1 second spectra was divided by this band pass correction.
Channels 1200 to 1300 (which were free of rfi) were averaged giving 440
samples. These values were subtracted from the normalized data. This was
done to remove 2 continuum sources that drifted through the beam.
The plots show the cell interference in the lbw receiver.
Plotting the data:
spectra of 3 50 MHz bands (.gif)
This shows the 3 50 Mhz bands: (1600=1650 Top), (1650-1700 middle), and
(1700-1750 bottom). The cell phone transmission rfi occurs in the center
image. You can also see glonass at 1609, and iridium at 1620 to 1628
spectra of the 1650-1700 Mhz band (.gif)
The cell phone transmissions are from (110 to 185 seconds) and (290
to 390 seconds).
There are two frequency bands:
1653.4 to 1667.4 containing 35 400 Khz channels
1690.3 to 1692.6 containing 6 400 Khz channels.
spectra plotted (.ps) (.pdf):
Fig 1. These are the average band passes during the cell phone transmissions.
The top is transmit1 while the bottom is transmit 2. Black is pol A, red
is pol B.
Fig 2. The average band passes have been normalized. The y axis units are
now tsys. You can see the frequency bins of the transmitted signals.
Fig 3. The 1 second spectra are over plotted for the 1653 to 1667 band
of transmit 1. Color differentiates each 1 second sample. The bottom over
plots 10 of these 1 second spectra with offsets for plotting. You can see
the time variation of the transmissions.
The allocated cell phone bands are:
We must be seeing the second harmonic of the 824-849 uplink. The cell phone
is actually transmitting on:
824-849 uplink mobile to base station
869-894 downlink base station to mobile
The lbw wave guide will not let 800 Mhz enter the dewar, so the harmonics
must be coming from the phone itself.
826.7 to 833.7
845.15 to 846.3
The frequency bin width of 400 Khz at the second
harmonic becomes 200 kHz at the fundamental. This is the channel width
of the GSM format. The range 826 to 833 must be the frequency channels
allocated to the tower that the phone was talking to. The 6 channels at
845 to 846 look like the 5 common control channels that are used for control.
Gsm does a frequency hop as well as tdma. Some of the time variation
in a channel is from that. The output power is also regulated depending
on how much talking is being done.
The individual 1 second spectra show max levels of
1.5 times Tsys (about 25K*1.5). The peak value is larger than this since
the device does not transmit continuously in 1 freq channel.
These harmonics lie on top of the 1665/1667 OH line.
Next thing is to see if we can see any harmonics when transmitting
from the visitor center.
You can see more
gsm info here.