The 47 Mhz receiver uses a set of dipoles that are mounted around the carriage 430 Mhz line feed. The system was originally built to be used with the 47 Mhz  when studying meteors.

Facts:

• The First LO is a fixed synthesizer upstairs which is normally set to 76.8 Mhz (46.8 + 30). When using the system you need to make sure that the synthesizer is on and set to the correct frequency (verify this with the birdie maker).
• The IF comes out at 30 Mhz in the receiver room on the rightmost rack of the 30 Mhz racks. You need to switch the IF selector to 47A, 47B manually.
• If you cable up left to left, left to top, base band mix, and then plug into the Ri you will get   the real signal  in the Q digitizer (rightmost of pair), imaginary  in I (leftmost). If you use riget with the complex option and then compute the spectrum, the frequency will be flipped. When looking at 30 Mhz on the spectrum analyzer the spectrum is flipped.

---

Sections:

---

History:

• 20jan06: Receiver is off frequency by about 150 hz. About 01feb06 found that the 5 Mhz reference was not working. It has now been fixed (01feb06).

---

Calibration measurements:

---

Miscellaneous

---

20jan06: On, off 3C348.0 (Hercules A) Tsys,gain,Beamwidth,sefd2: (top)

• Source rising, move  off source by -2 degrees in za. Tracking 2 degrees off, fire the cal for 14 seconds.
• Move on source and sit for awhile
• From -2 deg za offset drive to +4 deg za Offset (passing through the source) at slew rate (.04 deg/sec).
• Move from +4 deg za offset to -4 deg za offset passing through the source at slew rate.
• Fire the cal while off source.

• input the data, compute power and median filter to 16 milliseconds.
• Interpolate the telescope 1 second az, za  positions to the center of each 16 millisecond sample.
• Compute  the az, za positions for the source at the time stamp of each  data sample.
• Compute the Telescope Za - srcZa for each data sample. This is the source offset since we only offset in the za direction.
• For plotting purpose median filter the total power to .5 seconds.
• Compute scale factor of power counts to Kelvins using the two cals (at the start end end of the run). The scale factors agreed to better than 1% (taken at the beginning and end of run).
• Fit a gaussian to the power versus zenith angle offset to get Tsys, srcDeflection, and beam width.

• Fig 1 top,middle: The total power vs time for the 700 seconds of data (top is  polA, middle is polB). The green dashed line is the system temperature. The red dotted lines are where we crossed the source. The jumps in power at the start and end are the cal being fired. The dip in power at 400 seconds is when we started to spill over and see the ground (which is much colder than the sky).
• Fig 1 bottom: The zenith angle position of the telescope. The green * were on source.
• Fig 2: Plots the power versus za offset for the data (top polA, bottom polB). The bumps at +/- 2.5 degrees are the first sidelobes. The dip at +4 degrees is the Tsys going down because of the spill over (Tsys >> Tground). The green line is the gaussian fit. The sidelobes and above + 2 deg were not used in the fit.

    The table summarizes the results:
 

cal value	polA	polB	Notes
calValues used (K)	5700	4792	from electronics. These are old values.
Tsys (K)	6990	6486	The Tsys from the src fit is higher. This may be a sidelobe problem.
beam width (deg)	2.1	2.1	Scaling from 430 linefeed gives a beam width of 1.5 degrees.  The 47 dipoles are under illuminating the dish by 25% (the are probably also off a little).
Tsys/Tsrc	2.5	2.8	Using Tsys from the src fit
Flux src (3C348.0) (Jy)	1332	1332	Kuehr et al A&A suppl. vol 45, sep 81, 367-430.
SEFD (Jy)	3330	3729	this does not depend on the cal value.
Gain (K/Jy)	2.1	1.7

Notes for future cross scans:

• When redoing with cross scans try not to spill over so the baseline remains flatter.
• To move 4 degrees in za takes about 1.6 minutes at slew rate. You will probably want to go a little slower so the telescope can catch up.
• The pointing model is that of the line feed. Looks like it might be a bit off. The cross scan should tell us that.

---

20jan06: A first look at the 47 Mhz bandpass,dynamic spectra, rfi.

Average spectra and rms noise (.ps) (.pdf):

• Fig 1 top: Spectra of 4096 channels (244 hz resolution) were computed and then averaged to .2 seconds. This gave 95 spectra for the 20 seconds. The median bandpass was computed, normalized to the median power,  and then plotted. The birdie at 46.8 Mhz is the Dc offset of the a/d converters. The birdie at 46.92 Mhz comes from the birdies maker.
• Fig 1 middle: This is a blowup of the median bandpass. The left half of the band is a lot rattier than the higher frequency half.
• Fig 1 bottom: The total power versus time is plotted. There are peaks every 5.3 seconds.
• Fig 2: The rms/Mean was computed fore each channel of the 95 spectra and plotted. The bottom plot is a blowup. The expected value is determined by the radiometer equation. In this case the value should be .141. The bottom plot shows that the noise statistics are what we expect (with a small rise as we move toward the edge of the band (probably coming from the filters). The largest variability is at 46.689 Mhz.

    Dynamic spectra were computed for the 20 seconds using .2 second averaged spectra. A robust fit of a 7th order polynomial was fit to the median band passes and then used to remove the bandpass shape of each spectra. The spectral channels between 46.61 and 46.67 were averaged and then used to flatten the image. The images was then  averaged to 2048 channels. You can see :

• 5.3 second rfi at 46.689 Mhz
• the birdie maker birdie at 46.92
• A birdie at 47 Mhz with side bands
• A birdie at 46.66 that wanders around.

Fig 1 Top: This is blowup of the 46.92 Mhz birdie generated by the birdie maker (15.64*3 Mhz). The birdie is offset from where it is supposed to be by 153 Hz. This offset was also seen on the spectrum analyzer in the control room. I tested the synthesizer used to send the frequency to the birdie maker and it is locked to the station clock. It looks like the synthesizer for the 47 Mhz is either unlocked or someone dialed in the wrong frequency.

Fig 1 center: A blowup of the birdie at 47.000452 Mhz. It was not resolved in the 1 hz channels. The sidelobes are 11.403 Khz from the center.

Fig 1 bottom: The 5.3 second birdie at 46.689 Mhz. This has been averaged to 31 Hz. It is about 8. Khz wide. You can see that it is wandering about in frequency by a few Khz during the 5 seconds.

• The noise statistics for the receiver are correct.
• The receiver's 1st lo is probably unlocked. (note 01feb06.. 5 Mhz reference was not working. It has now been fixed).
• There is a birdie with a 5.3 second period near 46.689 Mhz. It is 8 khz wide and wanders by a few Khz every period.
• There is a narrow birdie (< 1 hz) a 47.00452 Mhz with side bands spaced 11.403 Khz from the center.
• The low frequency half of the band is a lot noisier than the upper half.

home_~phil