Hilltop rfi monitoring system
oct 2002
Intro
Daily plots
images for last 8 days.
Monthly plots of ism band
pictures of system
History
System modification
history for high freq system
Introduction:
The hilltop rfi monitoring
system
consists of an omni directional antenna covering 10Mhz to 1400 Mhz and
a log periodic antenna that covers 1.7 Ghz to 10 Ghz. The antennas are
mounted on the hilltop above the control room and feed a receiver
consisting
of switches, filters, amplifiers, and a spectrum analyzer (100Khz to 22
Ghz). The system runs 24 hours a day under computer control. There are
19 frequency bands that it steps through covering 10 Mhz to 10 Ghz. At
each band a 1 minute peak hold is taken, read back by the computer, and
then the setup is switched to the next band. Each frequency band
is sampled once every 19 minutes.
The data is written to disc and the
daily
plots are generated at 4 am on the next day. The plots created are:
The
current daily plots are:
- Daily plots encompassing the previous day. There are 4 sets of
the plots:
- The average plots
are just the
average
over the day for each frequency bin.
- The image plots are
frequency vs
time
for each band.
- One plot has no bandpass correction
- The 2nd set has
the
minimum channel over the day used to flatten the baseline. This
distorts
the true bandpass for channels that always have interference but is it
good for removing the standing waves in the cables for the higher
frequencies.
- rms/mean.
- For each channel of each band, the rms is
computed for the entire day and then divided by the mean value. This is
a sensitive indicator of interference (except for rock stable
tones).
- .pdf plots are created
(2 plots per file) which give
the best resolution and can be blownup.
- .gif plots are also
provided for users that prefer them
(although they are much lower resolution).
- Daily average
- Freq by time:
- Freq by time with bandpass correction:
- Rms/Mean by channel:
images by frequency for last 8 days:
- The 0 to 10Ghz data is split into two files
- Low freq (0 to 1450 Mhz)
- High freq (1600 to 10000 Mhz)
- Each page holds the 8 dynamic spectra for a single frequency
(coming from the last 8 days).
- There is one set with no bandpass correction and a second with
the bandpass correction.
- The image stretch (0 to 255) is done on the logarithmic
data.
- the date is printed on the upper left of each dynamic spectra as
yymmdd.
- The freq is plotted as the x axis in Mhz
- Not much rfi is seen above 6500 Mhz
- the gain in front of the spectrum analyzer is not
sufficient to overcome the spectrum analyzer noise.
Notes on the images:
- 3200-3240. This rfi has
the same time dependence as the aerostat radar.
- 30nov10.
- 3200-1260 = 1940 .. so the aerostat may be mixing with the cell
phone rfi.
- This is probably in our system...
- 6500 noise step.
- The spectrum analyzer uses
different bands.
- There is a band edge at 6500 Mhz. The higher band looks like it
has a higher noise temp.
- 1400 Mhz band standing wave:
- period is 8.02 Mhz
- This is a standing wave distance of 18.7 meters
(61.3feet) in free space.49 feet if index of refraction=.8.
- It may be the cable from the antenna to the box. The bpc
version removes this.
- 4000-6500 Mhz extra noise during
day (08:00 to 18:00)
- 30nov10
- more easily seen in bpc version.
- above 6500 it probably disappears because Tsys of the spectrum
analyze goes up.
- I wonder it this is a heating problem?
- if so why does it start at 4000? The system uses the same
components 2000 -> 10000.
2300-2650 monitoring the ism band
The ism band (2400-2500Mhz) is used by a number of
unregulated devices. The lower edge of this band is adjacent to the
band used by the 2380 Mhz sband transmit/receive system.
Plots have been made from the hilltop monitoring
system showing the daily usage of 2350-2650 Mhz. The parameters are:
- The plots are updated 4am each morning for the previous day/month.
- Each file contains one month worth of data
- Each page shows one day's worth of data
- The spectra are taken about every 20 minutes. They are plotted on
a db scale with an offset for display purposes.
- The bottom trace is the first plot of the day
(usually midnite start of the day).
- The top trace of the day is usually midnite (end of the day
.. unless the hilltop monitor system has trouble during the day).
- Vertical dashed lines have been drawn at 2380 and 2400 Mhz (for
reference).
- The data read from the hilltop monitoring at sband has an offset
of 3.6 Mhz.
- This was determined by looking at the center of the 2380 xmter
signal.
- If you look at this data with imgui, the interference will be
3.6 Mhz higher than these plots..
Monthly plots of ism band from hilltop
monitoring system:
Notes 2011:
- dual band rfi 2360-2340 seen when our 430 Mhz radar is
active.
- jan: 110112: 2380 Mhz rfi appears (micro wave link from radio rendentor tower).
- feb: radio
rendentor tower turns off misbehaving micro wave link
22feb11. 2380 Mhz rfi goes away.
- mar: 24-28mar Ao430rdr.
- may:Ao430rdr on:05,19,26,
- jun: weaker rfi 2380
- jul: Ao430rdr :07-09,12,14,16,21,27-31
- aug:
- Ao430rdr: 01-11,19-20,25-30.
- 2408 band shifts down by 10 Mhz for 20 minutes on 31aug
- sep:
- oct:
- 05: 2390 band appears for about 3 hours. appears when 2530 band
goes off.
- Found licences for 2525-2530 and 2535-2540 (ana G Mendez tv
broadcast) but no one at 2532.
- 08-12: 2408 band shifts down by 10 Mhz.
processing:
x101/111012/ismrfi_hilltop.pro
Pictures of system
History
- 080520: unused kl2500 control board 11 switched for bad board 11.
System back online.
- 080506: kl2500 switch controller board 11 error.. system brought
down to lab
System
modification history for the 1 - 10 Ghz
receiver:
The figure below shows the layout for the 1 to 10
Ghz
receiver. A, B, and C are amplifiers. They have been installed,
removed,
changed, etc.. over time.
Notes:
Noise floor 2.2 Ghz is the daily average connector to the full
receiver.
The cable is 50 feet of low loss heliax. 3.2 db loss at 2 Ghz to 8db at
10Ghz.
Since apr02 there have been harmonics of the 1950 cellular phone band.
At 3900 Mhz the level was -25 dbm in oct, -33 dbm in june, and not
there
before 29apr02 (with a noise floor of -58 dbm).
Log of changes to high freq receiver
| date |
ampA |
ampB |
ampC |
R1
db |
R2
db |
Noise
Floor
2.2Ghz |
Noise
Flr
3.5Ghz |
Notes |
17jan08
|
|
|
|
|
|
|
|
system reinstalled on hilltop.
monitoring restarted.
|
15may07
|
|
|
|
|
|
|
|
spectrum analyzer brought down
for maintenance during painting shutdown.
|
| 10jan06 |
|
|
|
|
|
|
|
New ac in rfi shack. Old one was failing. |
| 19sep05 |
replaces |
replaced |
|
replaced |
|
|
|
noise floor looks ok, but signals absent. |
| 26aug05 |
blew |
blew |
|
blew |
|
|
|
general power outage,lightning |
| 29oct02 |
|
|
|
|
|
|
|
Found that 50 ft cable connector at box had come
loose (the shield). Tightened it up. |
| 23oct02 |
g=25db |
g=25db |
0 |
3 |
0 |
-58 |
-60 |
didn't change by 20 db.. |
| 22oct02 |
g=25db |
g=dbdb |
g=30db |
3 |
10 |
-44 |
-45 |
added swept elbows box to specAnalyzer
added R1
replaced short cable R1 to input of box. |
| 9sep02 |
g=25db
1.7Nf |
g=25db
1.7Nf |
g=30db
2.8Nf |
0 |
10 |
-41 |
-42 |
back on the air, system saturates
with 1.9Ghz birdie |
| 22aug02 |
|
|
|
|
|
|
|
amps blown, off the air |
| 01aug02 |
|
|
|
0 |
|
-50 |
-42 |
|
| 16may02 |
|
|
|
0 |
|
-50 |
-48 |
back on the air, not sure which amps were installed |
| 29apr02 |
|
|
|
0 |
|
-58 |
-57 |
blew out 30apr02 |
| 25feb02 |
|
|
|
0 |
|
-55 |
-57 |
|
| 24sep01 |
|
|
|
0 |
|
-55 |
-56 |
|
Noise floor:
On 28oct02 we measured the noise floor of the spectrum analyzer using
peak
hold. The noise floor is a function of the input antenuation,
resolution
bandwidth, and the reference level. For the measurement we had the 10db
attenuator in. The resolution bandwidth and the reference level used is
in the table.
We then measured the noise floor with the high frequency antenna,
ampA,
and the 50 foot cable plugged into the spectrum analyzer (the 10db pad
was always in) to see how much more gain we needed to get above the
spectrum
analyz noise floor. Finally we plugged in the whole high freq
receiver
(ampA,cable,3db pad, ampB(25db) and recorded the noise floor. The
spectrum
analyzer steps bands near 6500 Mhz.
Noise floor measurements 29oct02
| freq (Mhz) |
ref
Level
dbm |
rbw
Mhz |
spectrum
analyzer
floor(dbm) |
antenna
ampA,cable
floor(dbm) |
antenna
full rcvr
floor(dbm) |
| |
|
|
low freq |
|
|
| 100 |
0 |
.1 |
-78 |
|
|
| 500 |
0 |
.1 |
-76 |
|
|
| 1000 |
0 |
.1 |
-75 |
|
|
| 1300 |
0 |
.1 |
-75 |
|
|
| 1400 |
-20 |
.1 |
-79 |
|
|
| |
|
|
hi freq |
|
|
| 1000 |
0 |
3 |
-63 |
-63 |
-64(800Mhz) |
| 2000 |
0 |
3 |
-62.5 |
-62 |
-57 |
| 3000 |
0 |
3 |
-62 |
-62.5 |
-60 |
| 4000 |
0 |
3 |
-63 |
-61 |
-61 |
| 5000 |
0 |
3 |
-62.5 |
-62.6 |
-60 |
| 6000 |
0 |
3 |
-62 |
-62 |
-61 |
| 7000 |
0 |
3 |
-58 |
-58 |
-58 |
| 8000 |
0 |
3 |
-56 |
-56.5 |
-56.2 |
| 9000 |
0 |
3 |
-56 |
-56 |
-57 |
| 10000 |
0 |
3 |
-56 |
-56 |
-57 |
Amp A with the 50 foot cable is not above the spectrum analyzer
noise
floor.
Amp A, cable, 3db pad, ampB(25db) is above the noise floor at 2ghz
but by 3 ghz the noise floor is being set by the spectrum analyzer. We
probably could use 10db more but may then have problems with harmonics.
home_~phil
