60 Hz in the 327 receiver.
Sections:
p1693 pulsar data taken jan07 to 05may07 with wapps
baseband sampled data with ri 03/04may07
conclusions
Plots/images:
P1693
plots
of the total power spectra (.ps) (.pdf
)
P1693
total power dynamic spectra using all 953 files (.gif)
P1693
high
resolution total power dynamic spectra covering 0 to 150 Hz (.gif)
03may07
RI baseband sampled data (.ps) (.pdf):
03may07:
dynamic spectra with .5 Hz resolution
04may07:
baseband sampled data on sky and load (.ps) (.pdf)
05may07:baseband
sampled data with 100 hz offset tone (.ps) (.pdf)
Intro:
A strong 60 hz component was seen in p1693 pulsar
data
taken with the 327 receiver and the wapps around 01may07. It had not
been present in data taken a few months earlier. It went away (the
strong version of 60hz) by the evening of 03may07. To check out this
problem the
p1693 from jan07 thru 05may07 was inspected. Baseband data was also
taken on 03,04,05 may07 using the radar interface (ri).
The p1693 pulsar data
P1693 is doing a pulsar search with the 327 receiver
using the wapps. The experimental setup is:
- dump 3 level acfs with 1024 or 512 channels every 256 useconds.
- combine the polarizations before writing to disc
- The receiver is centered at 327 MHz.
- 25 or 50 MHz bandwidths are used.
- There were 958 files out on disc for the period jan07 to
05may07. 5 files were ignored because of interference.
- Each file held about 2gb of data or about 1 million acf's.
Processing the p1693 data jan07 thru 05may07: (top)
An average spectra of the total power points was
computed for each file. The processing was:
- For each 256 usec acf compute the total power from the 0
lag.
- Compute a 64K spectra of these total power points:
- 64k*256e-6 = 16 seconds. This gave a frequency resolution of
1/16 hz.
- In a 2 gb file there were 16 of these spectra.
- Normalize each spectra to Tsys. This is the Dc value of the
computed spectra.
- Average the 16 spectra of each file.
- When done there were 953 64k long spectra covering
jan07 to 05may07.
Plotting the total power spectra: (top)
The plots
of the p1693 total power spectra (.ps) (.pdf
) show:
- Page 1 has the average spectra for the 953 files. The black trace
is the average, the red trace is the median. 60 Hz is about 1.4%
Tsys in the 1/16 Hz channel width.
- Full vertical and horizontal scale
- Blowup the vertical scale to .02% Tsys to show how far out the
60 Hz harmonics go. The green dotted lines are harmonics of 60 Hz.
- 0 to 300 hz display
- +/- 1 Hz about 60 Hz. The crosses are the samples. The 1/16 Hz
bin width has resolved the 60 hz in the average. Looks like the 60 Hz
wanders by about +/- .1 Hz (over the average of all the files).
The level of 60 Hz should probably be increased by 2 since there are
two channels with about the same value.
- Page 2 shows +/- 1 hz about 60 hz for each of the 953 files. You
can see the occasional jumping of 60 hz by about .05 hz in some files.
- Page 3 takes the maximum value in the range 59-61 hz for
each file. The different wapps have been color coded (to see if the 60
hz is tied to a single digitizer).
- Maximum 60 hz value vs date (AST). The level jumped before
15apr07. It went back down on 03may07 in the evening.
- Blowup of 60 hz level vs date. You can see the individual days
when the 60 hz was worse.
- 60 hz level vs bandwidth used. Both the 25 and 50 MHz
bandwidths have strong and weak 60 hz.
Dynamic spectrum of the tp: (top)
The first image is the
total power dynamic spectra for all 953 files (.gif).
- The image covers the entire 1953 hz with 2 Hz resolution.
- The image has been scaled so that 10 sigma covers the entire
range of black to white.
- You can see harmonics of 60 Hz all the way out to 1900 Hz.
The 2nd image is the high
resolution total power dynamic spectra covering 0 to 150 Hz (.gif):
The resolution is 1/16 Hz.
- The image has been scaled to full range of 10 sigma.
- You can see 60 and 120 hz narrow, spreading, and then getting
narrower again.
- There is a wandering weak birdie near 10 Hz.
Baseband sampled data was taken on 03may07 and
04may07 using the ri (12 bit a/d's) the setups were:
date
|
rfCfr
MHz
|
Bw
MHz
|
Pols
|
Time
secs
|
Notes
|
03may07
|
327
|
1.
|
A
|
120
|
On sky.
|
04may07
|
310
|
.01
|
A,B
|
300
|
sky and load
|
05may07
|
312
|
.01
|
A,B
|
16
|
sky with 100 Hz offset tone.
|
Processing the data:
The data processing steps were:
- Compute the dc offset using a few million samples
- Grab 16 seconds of data at a time, remove DC, and compute the
spectrum. Decimate down to a few Khz.
- Average these spectra till the end of the data set.
- Normalize the spectrum to the total power integrated over the
entire spectrum
Plotting the data:
03may07 data:
The first plot shows 60
Hz in the 03may07 baseband sampled data (.ps) (.pdf):
This was 1 MHz baseband data centered at an rf frequency of 327
MHz. Only polA was sampled.
- Top: +/- 300 hz about the center. 60 Hz is about 5% of the total
system power.
- Center: Blowup around 60 Hz. The 60 hz component is at close to
63 Hz. This was not seen in the wapp data. I'm not sure where this
offset is coming from. It is also moving about a fair bit over the 300
seconds of data.
- Bottom: The 3 harmonic of 60 hz. The vertical scale is about 1/10
th of the middle plot.
A dynamic
spectra with .5 Hz resolution was made from the same data set.
- There were about 150 2 second spectra in the 300 seconds of data.
- +/- 200 hz are displayed.
- The image has been scaled so that full range in the image is
6*sigma of the noise.
- At 10 seconds and 110 seconds the 60 Hz narrows for a few seconds.
04may07 data
The 2nd plot shows 60
Hz in the 04may07 baseband sampled data (.ps) (.pdf):
The data was 10 Khz of baseband data centered at an rf frequency of 310
MHz. Both polA and polb were sampled.
- Top: PolA +/- 300 Hz about the center. 60 Hz is .01% of the
system total power. This is a factor of 500 less than the data sampled
on 03may07.
- Middle: PolB +/- 300 Hz about the center. 60 Hz is .02% of the
system total power. -60 Hz is also visible. This mixer may not
have had it's i/q balanced perfectly.
- Bottom: Blowup around 60Hz. It is offset from 60Hz by about 2 Hz.
05may07
The 2rd plot shows 60
Hz in the 05may07 baseband sampled data (.ps) (.pdf):
The data was 10 Khz of baseband data centered at an rf frequency of 312
MHz. Both polA and polb were sampled. A tone at 312MHz + 100 Hz was
added for a frequency reference.
- Top: PolA +/- 300 Hz about the center. 60 Hz is .1% of the
system total power. This is similar to the wapp data.
- 2nd: PolB +/- 300 Hz about the center. 60 Hz is .3% of the
system total power. This matches the wapp values (the wapp data adds
polA and polB).
- 3rd: Blowup around 60Hz. It is centered at 60 Hz unlike the
previous 2 days.
- Bottom: The +100 hz offset tone shows up at -100 Hz. The spectrum
is flipped because of the 3 high side mixes and i have corrected for it.
The image shows they dynamic
Conclusions: (top)
- The 327 receiver typically has a 60 hz component of about
.3% of the integrated total power. Around 15apr07 this level jumped up
to about 5% (2.5% in a 1/16hz channel times two channels). On 03apr07
somtime in the evening it went back to the .3% value.
- When looking at dynamic spectra of 50MHz bw by time, the 60 Hz
was modulating a large fraction of the and pass. It was strongest in
the lower half of the band (centered at 327 MHz).
- The large 60 Hz was in seen in wapp 2, wapp 4, the oscilloscope
in the control room, and the ri sampled data so it is
probably not tied to a particular downstairs piece of equipment.
- Harmonics were seem up through 1900 Hz.
- The amplitude of the 60 Hz was:
-
|
Strength
[total pwr]
|
Notes
|
p1693
|
2.5%High
.3% Low
|
15apr07 to 03may07
After 03may07, before 15apr07
|
03mayRi
|
4.5%
|
62Hz
|
04mayRi
|
.03%
|
62Hz
|
05mayRi
|
.3%
|
60Hz
|
- The 60 Hz was resolved by the 1/16th hz channels. The total
contribution of the 60 hz was probably twice the value in the table
(since it covered about 2 channels).
- The ri data gave different results on 03may07 and 04may07. On
05may07 the values matched that of the wapp data. There may have been a
configuration error on these two days causing the samplers to run at a
different rate??.
- This experiment used the wide band 327 filter.
- On 3may07 in the morning, ganesh and dana went up and looked at
the receiver in
the dome. The 03may07 ri data (where the 60Hz was still strong) was
taken after this.
processing:
x101/070503/wappcmptp.pro, wappproctp.pro,327ri.pro, 070504/327ri.pro
home_~phil