Testing the QRFH on the telescope:jun22
July22
Links
QRFH
power levels on telescope (warm receiver)
check the pointing with
crosses on the moon
Tsys using on,off the
moon
sefd using on,off on
CasA
sun: on/off position
switch and cross.
wideband
receiver main page
Abbreviations:
- QRFH Quad ridge flared horn . 2 to 14 GHz
- FF Field
Fox spectrum analyzer.
intro
Hamdi mani brought a wide band
QuadRidgeFlaredHorn (QRFH) to arecibo during jun22. It was
tested on the 12meter using a warm receiver and the field fox
spectrum analyzer.
Data taking routines to drive telescope and fieldfox
the tcl datataking scripts are in /share/obs4/usr/p12m/wbrcv/
Timeline of what we did.
date
|
event
|
220609
|
log periodic ant (1-10GHz) on camera
tripod looking at rfi all night long on pad next to
12meter
|
220613
|
qrfh feed on camera tripod on pad next to
12meter
- sky, absorber (yfactor) measurement
- monitored rfi all night with qrfh feed pointing
south.
|
220614
|
qrfh feed installed on telescope
- long cables from feed down to pedestal used.
- y factor measurements
- had lots of trouble with dynamic range from rfi.
- onoff 3C274.. didn't see anything
- elevation scan. step 10->80deg in 10deg steps.
az=5
|
220615
|
- more testing trying to optimize dynamic range by
changing the postamp amplifiers'
- still had trouble with dynamic range
- on,off position switch on CasA
|
220616
|
Field fox moved adjacent to the
feed. solved the dynamic range problems
- yfactor measurements
- onoff crab nebula
- 2.8 GHz hipass filter installed
- setup was:
- lna->biasT->4ft cable->biasT->2.8GHz
HiPassFilter -> fieldfox
- Horn at 21.6 inches (capton to apex of secondary).
- azswings (stepping az,el)
- horn at 21.6 inches, start 18:10
- preamp on, FF attn set to 5db.
- az :-90 to +270 deg with azstep of 45 deg. 4
elevations starting at el-8deg then stepping by 25
deg in el.
- 20 recs/position
- cmdfile: onoff_moon_220617.tcl
|
220617
|
- on,off position switch on moon.
- 2 days after full moon., el 41 to 37 deg. start
04:59 am.
- preamp on, attn=0
- on,off position switch on casA . start 5:45 am
- attn=0,preamp on, 4 on,offs done.
- on,off position switch on sun. 14:44 -> 15:32
ast
- elevation steps. start: 18:57
- az=5, el 10->80 deg i 10deg steps.'
|
220618 |
- cross on moon
- 11 points/strip. -50 to +50 amin in 10 amin
steps
- cross on sun
- 11 points/strips -50 to _50 amin in 10 amin
steps
- azswings (stepping az, el)
- preamp on, FF attn set to 5db.
- -90 to +270 az using 24 deg az step. (10
recs/position)
- step in el 8 to 68 deg in 20deg steps
- el steps
- el =10 to 80 deg in 5deg steps. az=180.
- preamp on, attn=0
|
220619
|
- cross on casA start 23:54
- 7 points/strips 10amin steps
- 2 patterns , but only a strips ok.
- cross on moon.
- 11points/strips, 1 pattern, only az strip ok
|
220620
|
- on,off position switch on moon. 1 pattern.
- Switched to 2.5 GHz hipass filter
- azswings (drive in az, step in el)
- -90 to +270 at .2 deg/sec.
- 1 sweep of FF took about 6 seconds, so 1.2
degrees motion /spectra
- el: 6 degrees to 81 deg in 5 deg steps
- took about 8 hours.
|
220621
|
- cross on moon
- preamp off, attn=0 (because of 2.5GHz hipass
filter)
- az and el strips ok, but off moon we are close
to the ff noise floor.
- Switched to no HiPass filter
- azswings (drive in az, step in el)
- -90 to +270 at .2 deg/sec.
- 1 sweep of FF took about 6 seconds, so 1.2
degrees motion /spectra
- el: 6 degrees to 81 deg in 5 deg steps
- took about 8 hours.
|
|
|
Check the pointing with crosses on the
moon.
Crosses were done on the moon to check the
pointing of the system.
The crosses had:
- an az strip (great circle) followed by n elevation strip.
- On 18,19 june the elevation strips were incorrect (i had the
wrong offset coord in the script).
- each strip had 11 points covering -50 to +50 Amin in 10Amin
steps
- At each point had 20 recs of 3 avg sweeps each.
- the horizons ephemeris was used to track the moon.
Data was taken on:
- 220618,19 az strip only
- 2.8 GHz hipass filter in, capton at 21.6 inches
- FF preamp on, attn=0
- 220621: az and el strip
- 2.5 GHz hipass filter in, capton at 21.6 inches.
- FF preamp off, attn=0
- off moon, was close to noise floor of the FF
Processing the data:
Try and exclude freq channels that have rfi
when computing the total power.
For each of the 11 points in a strip:
- input the 20 spectra (each an avg of 3 sweeps).
- Compute a mask of good freq channels.
- compute the rms/mean by channel for the 20 spectra of a
point.
- Do a robust linear fit to the rms/mean. Any fit residuals
> 2.5 sigma were marked as bad.
- average the 20 spectra for the point
For each strip:
- exclude spectral points that stand out using a median filter
- for each of the 11 spcAvg of a strip:
- compute spcMed = medianfilter(spcAvg) using a filter
length of 120 freqChan (10001 total)
- divide :spcAvg/spcMed
- any freq channels > 1.2 are marked at bad
compute the total power for each point of a
strip.
- create a single mask for the strip combing the median
filter mask with the rms/mean masks
- only use 2.8 to 10 GHz when computing the total power.
- for each point compute the total power using the good channels
of the single mask
Fit a Gaussian to each strip
- Normalize the total power points of each strip by
dividing by the average of the 2 endpoints of each strips (off
the moon)
- fit to the function :
- z=(xoff - A1)/A2
- fit=A0*exp(-z^2/2) + A3 + A4*xoff
- where xoff are the offsets for the 11 points from the center
of the moon.
- This fits the beam width convolved with the width of the moon
(so A2 is not the beam width of the ant)
The plots show the results of the crosses on the moon
(.ps) (.pdf):
- Page 1: spectra off the moon.
- Top: 18,19 jun22.. az strips
- black: 18jun22, green : 19jun22
- 2.8GHz hipass filter in, preamp on, attn=0
- red: Noise floor of the FF. Above 10GHz the FF noise floor
approaches the spectral values.
- Bottom: 21jun22 az an el strips
- Black: azstrip, green el strip
- 2.5 GHz hipass filter in, preamp off, attn =0
- red: noise floor of FF. It is within a few db of the off
moon spectra.
- Page 2: Gaussian fits to the strips.
- Top: 18jun22 az strip it
- 2nd: 19jun22 az strip fit
- 3rd: 21jun22 az strip fit (preamp off)
- bottom: 21jun22 el strips fit (preamp off)
Fit values:
strip
|
amplitude
[Tsys]
|
PntOffset
[Amin]
|
fwhm fit
[Amin]
|
az18jun
|
.41
|
-.22
|
29.0
|
az19jun
|
.42
|
-1.23
|
29.0
|
az21jun
|
.12
|
-1.90
|
30.9
|
el21jun
|
.10
|
2.53
|
33.0
|
Summary
- The moon deflection of .41 Tsys
- the average az pntOffset was:-1.1 Amin
- the el pnt offset was: 2.53 Amin
- Adding them in quadrature gives a pointing error of 2.7amin
- This is computer over 2.8 to 10 GHz.
- the beam width probably varies from 40Amin to 9amin, with most
of the power probably coming at the lower end.
- This pointing error means that we can use the position
switch patterns without worrying about large pointing
errors.
processing: x101/220621/crosses.pro
On off position switching on
the moon was done measure the system temperature. The moon is about
30 Amin. The beamwidth of the 12meter varies from 40Amin to about
8amin.
- 220617: 2 on,off position switches done.
- 220620: 1 on,off position switch done.
- capton at 21.6 inches, 2.8 GHz hipass filter in, FF preamp in,
attn=0
- The horizons ephmerides were used to track the moon center
- the off position was 10Min Time. With a dec of -24deg, this
was a great circle offset of about 137 Amin.
Processing:
- The off
- The 20 spectra for each on,off were averaged.
- Tsys was computed as:
- (on-off) = moon Deflection
- off = Tsys
deflection
- MoonDefl/TsysDefl= (on-off)/off = (on/off -1)
- TsysDefl=MoonDefl/(on/off -1)
- If no size correction then MoonDefl=Tmoon.
- TsysK=Tmoon*sizeCor/(on/off - 1)
- The sizeCor was an attempt to correct for the extended
beam on the moon.
- The moon is taken as a constant brightness temperature
of 220K.
- The beam will spill off the moon (sidelobes at lower
freq, main beam+sidelobes at lower freq.)
- The temperature increase caused by the moon will decrease as
the spill over increases.
- The plot shows a means to correct for this (.pdf)
- The curves show the decrease in temp as a function of the
ration of HPBW/diameterObect and by the pointing offset as a
fraction of the angular radius.
- I used the pointing offset measured by the crosses on the
moon: 2.7Amin
- 2.7/15Aminmoonradius=.18
- The blue line shows the pointing error that was used.
- The curves only go down to hpbw/diamthobject=1.
- unfortunately our hpbw/30amin goes to 1.3 ..
- instead of computing the convolutions to this extent. i
just eyeballed and extrapolated the .9->1.0 step
- So the lower freq correction has been fudged.
- The page is taken from the
paper by Ho et al (.pdf)
The plots show the results of the on,offs (.ps)
(.pdf)
- Page 1:17jun22. 2 on,off spectra
- Top: spectra on,off moon
- black: on moon (2 spectra over plotted)
- red : off moon (2 spectra over plotted)
- middle: on/off -1
- The spectra has been smoothed to 31 MHz.
- black: 2 patterns over plotted.
- red: average of 2 patterns
- bottom: Tsys vs frequency (with no size correction)
- the blue line is 500K
- This assumes that the moon 220K temp fills the beam
and the sidelobes
- Page 2: 20jun22 on,off on moon (1 pattern)
- the 3 frames are the same as page 1 (except only 1 pattern.
- Page 3:try to correct for the beam size
- Top: Tsys vs freq with no size correction.
- black is the average of the 2 17jun22 measurements
- red is the 20jun22 measurement
- Middle: computed correction factor for the beam size
- This was taken for the figure in the Ho paper
- I used 2.7 amin as the pointing error
- fwhm=1.02*lambda/D=12 was used for the beam widths
- I also extrapolated to ratios > 1.
- Bottom: Tsys vs freq after size correction
- i average the 3 on/off values and then applied the sizeCor
in frame 2 to Tmoon.
- The dashed lines are at 200,300,400, and 500 Kelvins.
- Their is still lots of rfi in the plot
Summary:
- on,offs on the moon were used to compute the Tsys for the QRFH
on the 12meter
- 220K was used for the moon's brightness temp over 2 to 12 GHz
- a size correction was used to decrease Tmoon seen
by the system.
- the lower freq correction was extrapolated. It not as
reliable as the higher freq correction.
- Tsys goes from 200K at 3GHz to > 500K at 10GHz.
- These values are larger that those measured using the
absorber and sky yfactor measurements.
- When computing the size correction i should have used a beam
map with the sidelobes and all (but one was not available with
this horn.)
- For the Yfactor values I used Tflange +
Tscattered + Tsky=10K
-
Tsys values
Measurement
|
3-3.5
GHz
|
8-8.5
GHz
|
10-10.5
GHz
|
Moon
|
230K
|
430K
|
600K
|
yfactor sky,abs
|
202K
|
245K
|
393K
|
processing: x101/220617/onoff_moon_both.pro
On, off position switching on CasA was used to
measure the SEFD (System Equivalent Flux Density.. or Tsys/Gain).
Data was taken on 17jun22. 4 on/off patterns were done. The crosses
on the moon showed that the pointing error was not too large.
To compute the SEFD you need the flux of casA. Turns out the flux of
CasA is changing with time.
2 separate flux spectra were used:
- The absolute spectrum of Cas A: baars, et al 1977 (.pdf)
- The flux was corrected to 2022.
- An accurate flux density scale 50 to 50GHz. perley et al, 2017
(.pdf) (with
no correction to 2022).
- they claim the casA fit good to 4 GHz although the plots
look pretty good to higher freq. This limitation may have been
because of the interferometer beam size.
The plots show the on,off spectra and sefd: (.pdf)
- Page 1: on, off spectra
- Black is the average of the 4 on source spectra
- red is the average of the 4 off source spectra.
- Purple is the noise floor of the FF.
- Page 2: on/off, flux and sefd
- top: on/off -1
- black are the 4 on/offs-1 over plotted
- red is the average of the 4.
- middle: casA flux
- black: baars flux
- red : Perley flux
- The solid line are the standard values.
- The dashed lines are corrections to the flux because CasA
is extended
- the extended source correction: (fwhm^2)/(fwhm^2 +
srcWidth^2)
- is used 5 amin for casA width
- fwhn came from 1.02*lambda/Diam
- bottom: sefd
- The SEFD= Tsys/Gain
- on/off -1 = casADeflection/TsysDeflection
- casADeflection= fluxCasA*gain(k/Jy)
- sefd=
TsysDeflection/casADeflection/fluxCasA=fluxcasa/(on/off -1)
- I've used the corrected flux (frame 2)
- black i baars flux
- red in perley flux.
- The data was smoothed to 31 MHz.
- Page 3: compute sefd in 500 MHz bw steps.
- Top: casA srcDeflection/Tsys in 1 500 MHz steps
- rfi channels were excluded by:
- using robust fits to the rms/mean by chan
- and dividing by a median filtered spectra and clipping
at 1.2 Tsys.
- Middle : sefd for the 500MHz totpwr pnts
- black used the baars corrected flux.
- red is the Perley corrected flux
- there is lots of rfi between 5 and 6 GHz (UNI band).
- bottom: aperture efficiency for various Tsys values.
- The aperture efficiency is Tsys/SEFD*2760/(PI*DIAM/2)^2
- A gain of 1 K/Jy has 2760 m^2 (apertureEff=1=1.)
- The different colors are the efficiencies for various Tsys
values
Using the tsys measured with the moon on,offs and the sky absorber
measurements give the following aperture efficiencies:
Aperture Efficiency from CasA and Tsys for Yfactor
and moon.
measurement
|
3-3.5
GHz
|
8-8.5
GHz
|
moonTsys
|
.75
|
.8
|
Yfalctor Tsys
|
.68
|
.4
|
The efficiency from the moonTsys is giving an
efficiency that is higher than expected. The probably means the
moon Tsys is too high.
processing: x101/220617/onoff_casa.pro
On off position switch on sun
On 17jun22 we did 2 on off position switches
on the sun. the setup was:
- qrfh at 21.6" from secondary
- horn->lna->biasT->4ft cable->biasT->2.8ghz
Hipassfilter->field fox
- fieldfox, 0db attn, pre amp on. rbw 1Mhz,avg 3 sweeps\
- 20 recs on, 20 recs off source
- off source was 137 amin (great circle).
The plots
show the on,off position switch results (.ps) (.pdf)
- Page 1: on,off spectra
- Top: on sun, off sun spectra
- Black in on sun, red is off sun, purple is the noise floor
of the field fox
- all of these have been moved to the output levels of the
horn.
- two on,offs have been over plotted.
- Bottom: (on-off)/off (sun deflection in units of Tsys)
- black: 2 patterns overplotted
- red: average of 2 patterns
- blue: average smoothed to 31 MHz.
- The dips are rfi in the off spectra.
- Page 2: computed temp of the sun
- A model for the sun temperature was used:
- Tsun=A*freqGhz^(-2.1) + Tchromosphere
- Blackline: A=140077K , Tchrome=10880
- Red line: force Tsun to be 9000K at 8.5GHz
- This is a typical value have been measuring with
the12meter solar maps.
- Tsys in degK using the deflection on the sun and the above
models.
- This 100% coupling between the incoming radiation and the
12m output (itexcludes the aperture efficiency.
- Page 3: Solar power vs freq
- use the 2 previous models
- the freq channels are 1Mhz rbw
- The total integrated power is about -57 dbm
- The cryo elec lna's for the new system will have a 1db
compression of -35dbm.
Were we pointing at the sun?
On 18jun22 a cross was done on the sun (but
only the azimuth strip was valid. The setup was:
- horn,field fox same setup as 220617 above
- the azimuth strip was -50 to +50 arcminutes (great circle)
with 11 10Amin steps
- each point:
- average 3 sweeps/rec then average 10 recs.
The plots
show the azimuth strip results (.ps) (.pdf)
- top: average spectra for the 11 points across the azimuth
strip
- each color is a different point.
- The power levels has been translated to the output of the
qrfh
- Bottom: Total power vs az strip offset
- Each * is one of the 11 points
- The vertical scale has been normalized to the value at the
strip edges (tsys).
- The actual peak is probably 5amin offset from the expected
center.
The pointing error of 5amin --> that the
total power computation on 17jun22 is probably not biased by
pointing errors.
processing: x101/220618/suncross.pro
12meter_home
home_~phil