setup
sources
processing the data
plotting the averaged
recomb line by src
Line summary
Diagnostics:
rms/mean by freq channel
Tsys vs sample, el, ra,
local time for 09Feb22 data
Tsys vs sample, el,ra
for 14Feb22 data
diagnostics
summary
Intro
On 09 and 14Feb22 5min on/off position
switches were done on various sources looking at recombination
lines at 8gaz with the 12meter antenna.
Setup:
The setup was:
- On/off position switch:
- 5 minutes on source, 5 min off source, and then 10sec
salon, 10sec cal off in the off source position.
- The off source position retraced the same AZ,el as the
on position.
- The beam is about 10 arc minutes.
- a 5 min off source is 5*15*cos(dec) =75
main*cos(dec). so about 7.5 beams (assuming a point
source).
- 10 sec cal on,off
- This was taken in the off position.
- If/lo setup
- recomb lines looked at:
-
#
|
H
|
C
|
Notes
|
89a
|
9173.320653
|
9177.897014
|
outside band
|
90a
|
8872.568216
|
8876.994538
|
|
91a
|
8584.820642
|
8589.103413
|
|
92a
|
8309.382525
|
8313.527887
|
|
93a
|
8045.602609
|
8049.616376
|
outside band not used
|
- 4 frequency bands were used.
- the topocentric band center was set about 5MHz below
the restfreq of the line
- This was to stay away from the dc spike in the
middle of the backend spectra.
- The band centers used were:
- 9168.32 , 8867.57, 8579.82, 8304.38 MHz
- No doppler correct was performed.
- The 5ht line at 8045.60 was not used since the if/lo
has a 1 GHz IF filter.
- mock backend setup
- 4 boards used
- bandwidth 160MHz/ 3=53.33 MHz per spectrum
- 2048 freq channels gave 26 Khz channel width.
- the data files are stored as
- /share/pdataN/pdev/x101.20210209.bMs1g0.nnnnn.fits
- M = 0,1,2,3 for the 4 boards
- nnnn steps by 100 each scan
- A single on/off pattern will have (assuming we start
at nnn=0)
- nnn=0 on src
- nnn=100 off src
- nnn=200 salon
- nnn=300 cal off
- Each of the nnn will have 4 files M=0,.3 for the 4
freq bands. file nnn=0
- x101.20220209.b0s1g0.00000.fits (9168 band)
- x101.20220209.b1s1g0.00000.fits (8867 band)
- x101.20220209.b2s1g0.00000.fits (8579 band)
- x101.20220209.b3s1g0.00000.fits (8304 band)
- The data for 14Feb22 has the date switched to 20220214
in the filenames.
Sources used:
The first table shows the on/off patterns done on
09Feb22. Each line is a 5 minute on/off with the 4 file
numbers used on the left
On this day all sources had 2 5min on,off patterns done.
Sources for 09Feb22
source
|
RA(Hr)
|
file numbers
|
| NGC6334 |
17.344 |
0 100 200 300 |
| NGC6334 |
17.344 |
400 500 600 700 |
| NGC6357 |
17.427 |
800 900 1000 1100 |
| NGC6357 |
17.427 |
1200 1300 1400 1500 |
| M8 |
18.063 |
1600 1700 1800 1900 |
| M8 |
18.063 |
2000 2100 2200 2300 |
| w31 |
18.157 |
2400 2500 2600 2700 |
| w31 |
18.157 |
2800 2900 3000 3100 |
| w33 |
18.236 |
3200 3300 3400 3500 |
| w33 |
18.236 |
3600 3700 3800 3900 |
| m17 |
18.341 |
4000 4100 4200 4300 |
| m17 |
18.341 |
4400 4500 4600 4700 |
| w43 |
18.794 |
4800 4900 5000 5100 |
| w43 |
18.794 |
5200 5300 5400 5500 |
| w48 |
19.030 |
5600 5700 5800 5900 |
| w48 |
19.030 |
6000 6100 6200 6300 |
| w49 |
19.171 |
6400 6500 6600 6700 |
| w49 |
19.171 |
6800 6900 7000 7100 |
| w51 |
19.395 |
7200 7300 7400 7500 |
| w51 |
19.395 |
7600 7700 7800 7900 |
| dr7 |
20.469 |
8000 8100 8200 8300 |
| dr7 |
20.469 |
8400 8500 8600 8700 |
| dr21 |
20.650 |
8800 8900 9000 9100 |
| dr21 |
20.650 |
9200 9300 9400 9500 |
| NGC7538 |
23.225 |
9600 9700 9800 9900 |
| NGC7538 |
23.225 |
10000 10100 10200 10300 |
| w3 |
2.428 |
16200 16300 16400 16500 |
| w3 |
2.428 |
16600 16700 16800 16900 |
| w3N |
2.448 |
17000 17100 17200 17300 |
| w3N |
2.448 |
17400 17500 17600 17700 |
| S206 |
4.054 |
17800 17900 18000 18100 |
| S206 |
4.054 |
18200 18300 18400 18500 |
| orion_A |
5.588 |
18600 18700 18800 18900 |
| orion_A |
5.588 |
19000 19100 19200 19300 |
| M43 |
5.592 |
processing the data 19400 19500 19600 19700 |
| M43 |
5.592 |
19800 19900 20000 20100 |
| orion_B |
5.696 |
20200 20300 20400 20500 |
| orion_B |
5.696 |
20600 20700 20800 20900 |
The 2nd table shows the sources done on 14Feb22. Some of the
weaker sources from 09Feb21 were repeated multiple times.
Sources for 14Feb22
source
|
RA(hrs)
|
file numbers
|
| w48 |
19.030 |
0 100 200 300 |
| w48 |
19.030 |
400 500 600 700 |
| dr7 |
20.469 |
800 900 1000 1100 |
| dr7 |
20.469 |
1200 1300 1400 1500 |
| dr7 |
20.469 |
1600 1700 1800 1900 |
| dr7 |
20.469 |
2000 2100 2200 2300 |
| dr7 |
20.469 |
2400 2500 2600 2700 |
| dr7 |
20.469 |
2800 2900 3000 3100 |
| dr21 |
20.650 |
3200 3300 3400 3500 |
| dr21 |
20.650 |
3600 3700 3800 3900 |
Processing the data.
The idl route masposonoff() was used to
process the data:
- Input the on and off scans. bon[300],boff[300]
- average each 300 second set.
- Use the median rather than the normal mean to help
exclude outliers from rfi
- bonAvg,boffAvg
- Average the calon, caloff records using the median.
- Compute bonAvg/bOffAvg
- Convert from Tsys to Kelvin units using the cal difference
and the cal value in degK
- average the calOn,calOff over freq using a mask to
exclude rfi channels.
- Compute the mask:
- cal Ratio=calOnAvg/calOffavg
- flatten cal Ratio by dividing by a smoothed version
of calRatio/smooth(cal ratio,.03*Nanchang)
- robust fit a linear function, and flag all points that
are greater than 3 sigma from the fit
- When starting the fit , exclude 6% of the channels
on the edges.
- Compute
- CalonTpCnts=mean(calonSpc[freq]) using mask to exclude
bad freq channels
- CalOffTpCnts=mean(calOffSpc[freq]) using same mask
- DeltaCalCnts=CalOnTpCnts - CalOffTpCnts
- Get the calValueK vs freq. average over freqs using
the good channels from above mask.
- CntsTodegK=calValK/(deltaCalCnts)
- Scale spectra Cnts to degK
- The cal scale factor is in specCnts
- Bon/Boff is in tsys units.
- Scale Bon/Boff by the mean value of Boff (using the
good channels we used for the cal)
- This puts bon/Boff back to spectraCnts
- multiply Bon/normalized(Boff) by CntsTodegK to get
Kelvins.
- Average multiple onoffs of the same source
- Use the median if 3 or more on/offs
Plotting the averaged recomb lines by
source:
The average line profiles were computed and
plotted vs frequency and then lsr velocity. The computation
consisted of
- avg position switch on,offs, convert to kelvins using cal.
- compute on/off-1
- average the pols
- remove a baseline from each averaged src/line. linear fit
+ 4ht order harmonic.
- smooth by 5 channels.
The plots show the averaged line profiles taken
on each day.
- Each page is a separate source.
- the lines: 89a,90a,91a, and 92a are plotted
- the dashed vertical lines in plot:
- freq: black,purple - the topocentric line frequencies
for H, C lines (no doppler correction)
- vel: purple , the expected lsr velocity.The dashed
purple line is the expected lsr velocity (for the velocity
plots)
- the integration time is printed on the top of each page in
minutes.
- the channel width is : 26Khz before smoothing, 130Khz
after smoothing.
- the expected and measured rms are printed in
red/green as well as the average Tsys.
- the rms is computed across the freq band (excluding +/-
50 km/sec about the line velocity.
- the measured rms is at least a factor of 2 greater than
the expected value.
220209 data
220214 data
the sources w48, dr7, and dr21 were
re-observed to improve the signal to noise.
Combine 220209 and 220214 common sources
This combines the 220209 and 220214
sources (w48, dr7, and dr21).
These plots have integration time 10 minutes
longer than 220214.
Line Summary:
- 19 sources where checked for recombination lines on
09Feb22.
- the Alpha lines 89a,90a,91a,92a where measured.
- on off position switching was used with 10 minutes on
source.
- On 14Feb22 3 weak sources were observed again to help get
a detection.
- Most of the lines were detected (in at least 1
transition).
- after removing a baseline (1st order poly, 4ht order
harmonic) the baseline rms was at least a factor of 2
greater than the expected value.
processing: x101/220209/recomb_get dat.pro,recomb_plot.pro.
220214/recomb_get dat.pro,recomb_plot.pro
Diagnostics
Rms/mean by channel
Each on and off had 300 1 second spectra
each with 2048 freq channels. The rms/mean was computed for
each channel across the 300 spectra. The resulting value is
determined by the radiometer equation:
- expectedRms/Tsys = deltaT/T=1./sqrt(bw*tm)
- bw=53.333MHz/2048 = 26.041 KHz/channel
- tm=1 second
- expectedRms/Tsys=.0062
This was done for each freq channel of each on,off
scan.
220209 data:
The first plots show rms/mean
spectra for a few selected sources (.pdf)
- Each page has 4 frames (1 for each frequency band)
from a single source.
- black is polA, red is polB, the purple line is the
expected rms/mean
- These spectra had rms/mean values that was larger than
expected in at least 1 freq band
- Page 1 NGC6334
- 8868 band had lots of interference. It was present
for the first 6 sources done.
- 8650-9000 MHz is allocated to radiolocation.
Aviation and private land mobile.
- 8320-8330 also had rfi. This was seen the first
7 and last 5 sources.
- 8025-8400 is allocated to earth exploration satellites
(space to earth).
- Page 2 dr7
- PolA rms is larger than polB rms
- It is largest in highest band and decreases with freq
(see next set of plots).
- This is probably the gain changing with time that has
been seen intermittently in polA.
- Page 3 dr21
- polA rms larger at higher freq bands. Similar to page 2
dr7
Average/median rms/mean vs source
(.pdf)
- For each on or off, the rms/mean was computed and then the
mean or median value across the freq band was computed.
- Page 1 has the average mean/rms across frequency.
- Page 2 has the median mean/rms across frequency.
- sporadic rfi in the band will be excluded by the median.
- Each page has 4 frames for the 4 frequency bands
- The x axis is the source number (listed on the right
side of the plot)
- The colors are for the various 300 scans that were done.
- black polA on, red polA off, green polB on , blue polB
off, purple expected rms.
- there are 2 sets for each type (since we did 2
on/off's per source).
- 8868 band
- the first 6 sources had large rms/mean using the mean
and the median (so the rfi was > 50%of the band).
- increasing rms/mean with freq band.
- source 11,12 (dr7,dr21) show a larger rms/mean as the
freq band increases.
- The previous plots showed that this was across the
entire freq band (rather than isolated rfi).
- This is probably a gain variation during the 300 second
scan and it is larger at higher frequencies.
- 8304 band
- the average rms/mean is large while the median
(rms/mean) is close to the expected value
- this is the 8320-8330 rfi seen in the first set of
plots.
Median rms/mean for each on, off
done (.pdf)
- there are 19 pages (one for each source)
- the 4 frames per page show the first on,off and then
the 2nd on,off done for each source
- Black in polA, red is polB
- The x axis shows the 4 frequency bands recorded.
- the median (rms/mean) across each band is plotted.
- The vertical axis is limited to .01 so some bands with
large rms/mean go off scale.
- The first 6 pages show the large rms/mean in the 8868 MHz
band
- page 11,12 (dr7,dr21) show the increase in rms/mean with
freq in polA.
- The increase is not contiguous in time:
- dr7: 2nd on source has increase, but off source does
not.
- dr21: the first on source has no increase, but the first
off source does have it.
- Page 17 orion_a
- larger rms/mean for 2nd on source.
- This is probably a gain variation during the 300 second
scan.
220209 Tsys vs sample, elevation, ra,
and local time
The cals for each on,off pattern were used
to convert the 1 second spectra to degrees Kelvin and then the
median value for each spectra was computed.
The plots show Tsys vs sample, elevation, RA, and
local time (.pdf)
- Each frame is a separate frequency band.
- The vertical scale is Tsys in degrees K
- Page 1: tsys vs sample
- there are 300 samples/scan. We did 2 on/offs per
source and had 19 sources.
- The vertical dashed lines separate 300 sec scans.
- The x axis lays out the 300 samples times the 2*19
on or offs
- the colors show:
- Black: polA on source
- Red : polB on source
- green: polA off source
- blue: polB off source
- Tsys for polA is higher than Tsys for polB (assuming the
cal values are ok:).
- The differences for PolA on,off and polB on,off occurs
for sources with continuum flux.
- scans 11,12 - M17,
- last 5th,6th scans: orion_A
- 8868 band. for the first couple of scans there is lots
of rfi, then it stops.
- Page 2: Tsys vs elevation
- the off source median Tsys was used.
- Black is polA, red is polB
- From elevation 45 to 52 there are measurements at the
same elevation with different values.
- the next plots show that the higher values were taken
in the plane of the galaxy.
- Page 3 Tsys vs Right ascension
- We see lower tsys's away from galactic time.
- Page 4: tsys vs local time.
- It peaks around 13 ast (this is also around the galactic
time).
220214 Tsys vs sample, elevation, and
RA
Similar plots were made for the 220214
tsys data (see 220209 above for a description of the plots)
220214: Tsys
vs sample, elevation, RA, and local time (.pdf)
- 3 sources were used: w48,dr7, and dr21
- 10 on/off pairs were done
- The 4 frames on each page are the 4 frequency bands.
- Page 1: tsys vs sample
- the x axis plots the 1 second Tsys values from each
scan.
- black polA on, red polB on, green polA off, blue
polB off, purple lines separate 300 second scans.
- the srcnames for each scan are printed on frame 2.
- there was not as much rfi as the 220209 data.
- The 2nd dr7 on,off showed a jump in polA.
- The jump was larger for higher frequencies.
- examining the median rms/mean by freq band show a
large increase in rms with freq band.
- The cal is taken at the end of the on,off pair. If the
gain changes during tine 10 minutes, then this will
distort the Tsys measurement.
- Page 2: blowup showing the jump between the first and 2nd
dr7 onoff pair.
- Page 3: Tsys vs elevation
- Black in polA, red is polB
- the polA Tsys slope vs el for the el=81 and
83 deg scans are the dr7 jumps.
- Page 4: tsys vs ra
- This data did not have a large ra span (19.1 to 20.8
hours).
Diagnostics Summary:
- The rms/mean by freq channel was used to examine the data
- The 8868 MHz band showed lots of rfi for the first 6
on,off pairs and then it stopped (airplane?)
- rfi was seen at 8320 to 8330 MHz
- 220209 > 50% of the on,offs
- 220214 not present
- Pol A had gain changes within a 5 min on or off with
higher frequency bands having more change.
- when looking at the rms/mean this showed a larger rms
across the entire band (since the rms includes the noise
and the gain change)
- for Tsys measurements, this showed as jump in tsys
(since the cal was only taken at the end of the on, off
pair).
- Tsys for polA was 5 to 8 K higher than polB
- We saw about a 6 degK Tsys change when looking in and out
of the plane of the galaxy.
- For out of the plane tsys varied (for polB)
- 93K at 8304 MHz to 117 K at 9168 MHz
processing: x101/220209/rms_getdat.pro,plot_tp.pro,
220214/rms_getdat.pro, plot_tp.pro