327 band RFI measurements.
last updated 18oct06
18oct06:
looking at rfi created by axis 212 web camera.
03oct06:
614Khz comb.
18apr06:
the 750 Khz and 40 Khz 327 combs come from the sband console in the dome.
16apr06:
750 Khz and 39.45 Khz combs in 327 receiver
13apr06: testing the trenddata gigabit
switch for rfi.
28mar06:
look at new ac units in dome. Compare with comb from old unit.
16jan06:
28oct10 az swings show birdies with 1 min periodicity
15jan06:
wide band look at Dome ac birdies in 327 Mhz rcvr.
14jan06:
3rd harmonics of fm stations appearing in band
10jan06:
placing hipass, bandpass filters before the 1st amp
02nov05
cleaning up the 327 Mhz band.
28oct05:
327 Mhz birdies and their azimuth dependence.
21oct05: 327 Mhz spectral density
at output of 1st amplifier (before the rf filter)
16oct05
1 Mhz comb in 327 receiver from both dome AC units
03jun05
16Khz comb in 327 receiver from alfa motor.
03dec03 327 Mhz comb from sband
Tx Rack in dome
24nov03 comb at 327 Mhz.
24jul02 2 MHz comb in 327 Mhz receiver
01dec03 327 Mhz comb coming from sb Tx room
in dome.
13apr06: testing the trenddata
gigabit switch for rfi. (top)
A trend data gigabit ethernet switch (teg-s41sx) was
tested for rfi. This switch has 1 fiber port and 4 copper ports. During
the testing there was nothing connected to the switch (except the power).
In the screened room rfi was seen from the switch beyond 2 Ghz. The strongest
birdie was at 476 Mhz. It was stronger than 50 db above the noise floor
of the
sbt250 spectrum analyzer (the resolution bandwidth was about 300 hz).
On 13apr06 data was taken with the 327 receiver
on the telescope and the interim correlator. The switch was located in
the parking lot next to the control room inside the rfi van. It had a clear
shot to the telescope through the front window or the van. The azimuth
was positioned to az=242, za=18 (the 1400
Mhz birdie from the wapps is a maximum here). The correlator was set
to 390 Khz bandwidth and 2048 channels. After hanning smoothing the resolution
bandwidth was 380 hz. 180 1 second integrations were made.
The data taking started with the switch on. It was
turned off at second 80. It was turned back on at second 125. It was shut
off again at second 166.
-
The first image (polA) shows the
390 khz band for the 180 seconds (.gif).
-
The birdie is at 335.9367 hz. You can see it come and go as we switched
the switch on and off.
-
The majority of the other birdies are a comb with a spacing of
39.5 Khz
-
The second image is a 50
Khz blowup centered on the birdie (.gif):
-
A plot of the
1 second spectra as well as the total power of the birdie channel
(.ps) (.pdf):
-
The top frame overplays the 180 1 second spectra (with an offset for plotting).
There is a drift of about 1 channel (380 hz) in the spectra. It is probably
a thermal shift when the switch is off and then turned back on.
-
The bottom frame plots the birdie channel (average of 2 channels about
the peak). The black plot is polA while the red plot is polB. The times
when the switch was turned on and off are flagged. Pol A is stronger than
polB. The strongest is about 11 times Tsys in an 380 hz channel width.
Tsys for the 327 receiver is about 120 K.
Conclusions:
-
The gigabit switch with no link active makes rfi through 2 ghz.
-
The majority of the rfi is coming from the fiber to copper converter port.
-
The strongest rfi is at 476 Mhz. It is 50 db above a 700 K noise floor
in 300 Hz resolution bandwidth when a probe is set on the device.
-
When the switch is placed in the parking lot next to the control room,
a birdie at 335.9367 Mhz can be seen in the 327 receiver. It is 11 times
tsys in a 380 Hz channel. It was stronger in polA than in polB. This was
the strongest birdie in the 327 Mhz band when looking at the switch in
the screened room.
processing: x101/060413/rfi.pro
21oct05: 327 Mhz
spectral density at output of 1st amplifier (before the rf filter). (top)
The output of the 327 Mhz receiver after the first amplifier
was measured with the tektronix ybt250 portable spectrum analyzer. The
signal was taken after the first amp and before the rf filter (the
filter limits the band to 30 Mhz about 325 Mhz). The telescope was at an
az of 270 and a za of 8.46 degrees. Data was taken covering 0 to 1000 Mhz
and then 0 to 500 Mhz. At each band a regular sweep and a peak hold of
about 1 minute were recorded. This was done for polA and polB. The plots
show that television
stations are the largest signals in the receiver band (.ps) (.pdf):
-
Top: this covers 30 Mhz to 1000 Mhz. The black and red are the single sweeps
for polA and polB. The green and blue are the one minute peak holds for
polA and polB. The largest birdie in the band is 471.25. This is the video
carrier for tv channel 14. The next two largest birdies are 519.25 (tv
channel 22) and 187.25 (tv channel 9).
-
Bottom: the 30 Mhz to 500 Mhz measurements.
On 20oct05 we saw intermittent intermods/harmonics
in the 327 Mhz band that included the fm station 107.3 (3rd harmonic
at 322) and intermods with a tv station and 107.3 (we heard the fm station
in the IF signal). The power levels measured above do not look like
they should be driving the first amp into compression.
The 1 minute peak hold has values that are up to
10 db stronger than the single sweep values so the levels are time variable.
There is most likely an azimuth dependence to the birdies. It still looks
like a good idea to put an rf filter before the first amplifier (since
the receiver isn't cooled).
processing: x101/051021/327.pro
03dec03 327 Mhz comb from
sband Tx Rack in dome. (top)
The 327 Mhz comb (with
372 kHz spacing) is coming from the sband transmitter cabinet in the dome.
Rey velez noted that the fingers for making contact between the door and
the cabinet were not working. Data was taken with the correlator while
the sband rack was on and off. 1 second dumps were done using a 24 kHz
resolution and a 190 Hz resolution (after hanning smoothing). A bandpass
correction was fit to an average of the band passes when the rack was powered
off. The plots show the 327
MHz comb when the sband rack was power on and off. The door from the
sband transmitter room to the stairwell was open during these test.
-
Fig 1 . A 30 second average with the rack on. The top plot has 24 kHz resolution
while the bottom plot has 190 Hz. The vertical units are Tsys.
-
Fig 2. The 30 second average with the rack on (black) followed by a 30
second integration when the rack was off (red). The top plot has 24 kHz
resolution while the bottom plot has 190 Hz. You can see that turning the
rack on created many birdies and it raised the entire continuum level by
about 10%.
processing: x101/031203/doit.pro
24nov03 comb at 327
Mhz. (top)
(note: this comb is coming from the sb
tx console in the dome)
A comb appeared at 327 Mhz sometime in oct or nov03. Data was taken
on 24nov03 with the interim correlator using a 25 Mhz bw, 2048 channels,
1 second sampling, and centered at 327 MHz. An az swing was done from az=290
(za=10) clockwise to an az at 660 using a rate of .4 deg/sec. The az was
then swung counter clockwise from 660 to about 540 (where is stopped because
the cable car was in use). The plots show the
comb at 327 Mhz:
-
Fig 1 Top is the average of the clockwise spin. Black is polA, Red
is polB. The units are Tsys (about)...Bottom: This is the average
of the counter clockwise spin. The vertical scale is blown up to show the
smaller birdies.
-
Fig 2 This shows the spacing between the birdies. A baseline was
fit to the non-comb values and then removed. All measurements above 1.5
Tsys were flagged as comb members. Combs in adjacent freq channels were
collapsed to the channel with the largest value. The frequency difference
was then computed between all of the adjacent comb members. The data was
plotted with the comb intensity versus the comb frequency step. The points
clustered at multiples of 372 Khz with the largest number occurring at
327*2=745 Khz. Black is polA, Red is polB, and green lines are multiples
of 372 kHz. The top plot is the clockwise spin, the bottom plot is the
counter clockwise spin.
-
Fig 3: This is the strength of each comb element vs azimuth position
for all of the comb members found. The top plot is the clockwise spin,
the bottom plot is the counter clockwise spin. Colors differentiate different
comb members (note that the colors on the top plot do not necessarily correspond
to the same frequency for that color on the bottom plot).
The comb spacing is 372 kHz with the strongest values at 745 Khz
(actually 371.97Khz generates the comb at 327 MHz). There is variation
in the power for each element vs azimuth but it does not look like something
that would be coming from a tower far away. The dropouts and increases
do not repeat with azimuth implying that they are a time variation rather
than an azimuth dependence. The birdie is probably on the platform or even
in the dome. It would be interesting to take some data with the doors to
the turret room open and then closed.
processing:x101/031124/doit.pro
24jul02 2 MHz comb
in 327 Mhz receiver (resolved).
(top)
A 2 Mhz comb was found in the 327 Mhz receiver. It is at the
odd Mhz 327,329,331,.. Mhz. It is offset from the Mhz by +3.737Khz. Looking
more carefully you can also see a birdie offset by the same amount at 326,328,330,....
The comb must be a 1 Mhz square wave with the even harmonics non-zero because
of imperfections in the square wave. The birdie goes away when the receiver
is put on load so it is coming in through the front end. The first lo was
shifted and the comb remained at the same rf frequency so it is not coming
in after the first mixer.
Data was taken with the correlator in 195Khz/2048
channels (with hanning smoothing). This has a 190 hz resolution.
Band passes were centered at 327,329,331,and 333 Mhz. 1200 records were
then taken while various devices were turned on/off and doors opened, closed.
The 327 MHz birdie was then mixed down close to
base band and sampled with the radar interface at 10 Khz. This measured
the width and drift rate of the signal.
The plots show the results of these measurements (note.. the plots
call it a 2 MHz comb since the odd harmonics were so much stronger):
-
Comb
offsets from 327,329,331, and 333 MHz The comb is offset
3.737 Khz above 327,329,331,333, etc. The spacing is 2 Mhz within the measurement
resolution.
-
Drift
rate of birdie at 327 Mhz. The for the 157 seconds of the
measurement the signal drifted at .041 hz/second. At 327 Mhz this is a
stability of .041/327e6= 1.25e-10 fractional change/second. 3.2 seconds
of data were taken for each spectra giving a resolution of .3 hz. It has
still not been resolved in the last spectra.
-
Data was taken for 1200 seconds using the correlator. During this time
the doors going out to the stairwell were opened and closed a few times.
The doors to the iflo racks were also opened and closed. The plot shows
the birdie
at 331 Mhz -Tsys versus time for the 1200 seconds. The only thing
that correlates is the opening and closing of the outer doors. Opening
and closing the iflo racks did not make a large difference.
Resolution: Conducting
gasket in huffman box degrades with time.
After turning off a number of equipment it was found
that the birdie was coming from the power meter that is used to measure
the power in the 430 Mhz wave guide. It is located on the ceiling of the
turret room inside a huffman box. It has an rf cable going to it from the
wave guide, a fiber optic cable sending the signal downstairs, and an ac
power code that goes through an ac power filter at the input. Inside
the box is a pc card with a VtoF converter on it. This card uses a 1 Mhz
crystal. There is another of these boxes in the carriage house measuring
the 430 Mhz output power. At first the ac power filter was suspected
as the culprit. The box in the carriage house was brought down and tested.
The main problem was the conducting gasket around the inside top of the
huffman box. It was supposed to improve the contact between the top and
the box when the top was closed. The gasket was a rubber (neoprene?) material
with a conductive coating. Where the gasket was in contact with the box
(top or bottom) the gasket was still conducting. The edge of the gasket
had a dark strip which was no longer conducting so the top of the gasket
was no longer electrically in contact with the bottom. When the box was
closed the gasket was acting as an insulator for the top. The guess is
that the part of the gasket in contact with the air oxidized the conducting
material. This is a serious problem since most of the huffman boxes we
have use this gasket material. It is probably why the color camera birdie
was not there when the cameras were installed in their new shielded boxes
and then reappeared 6 months later.
processing:x101/020724/327xxx.pro
01dec03 327 Mhz comb coming
from sb Tx room in dome.
On 01dec03 data was taken with the 327 receiver
while tracking blank sky. The correlator was set for 2048 channels over
25 MHz (centered at 327 MHz) and 2048 channels over 195 kHz (centered at
321.7 Mhz. The data was dumped at 1 second intervals and hanning smoothed
to give resolutions of 24 kHz and 190 Hz. While the data was being taken,
the doors to the sband transmitter room was opened and closed as well as
the door to the turret room. Data was taken for 2200 seconds.
We knew that the signal was coming in through the
horn (since it went away when put on load). The doors should give some
shielding to birdies generated inside the rooms. If the birdie strength
increased when the door was opened, then it was probably coming from inside
the room. The plots shows the
327 comb data while the room doors where opened and closed.
-
Fig 1 This is the average of the 2200 seconds of data. Black is polA and
red is polB (polB has been offset in frequency by a few channels for display).
The top plot is the 25 MHz band (24 kHz resolution while the bottom plot
is the 195 kHz band (190 Hz resolution). The 190 hz resolution data did
not resolve the comb.
-
Fig 2. This is the time series for the 321.75 Mhz, 190 Hz wide channel.
The top plot is the entire 2200 seconds (it took me awhile to get up to
the dome to open/close the doors). The bottom plot is a blowup of 300 seconds
when the transmitter room doors were opened/closed twice. The dashed green
line is when the transmitter door was opened. The blue line is when the
door was closed. (these times could have been off by a few seconds. You
can see that the power in the channel went up by almost a factor of 7.
This shows that the comb is coming from inside the room.
The dip at 1250 seconds is when i opened the door
rapidly and then closed it to get into the room for the first time. When
I closed the door, it must have made better contact than previously.
processing: x101/031201/doit.pro
rfi measurements home.
home_~phil
