alfa performance at OH with 10db amps in

comparison of the system performance for alfa and lbw at OH (.ps)  (.pdf) (no amps in)
comparison of the system performance for alfa and lbw at OH (.ps)  (.pdf)   (with amps in)

    A filter bank was installed in the alfa receiver in mar05 that allowed switching between the 300 Mhz filters and filters that let you look at the OH line.  On 01apr05 the system performance was measured using the continuum source CTA21 (B0316+162. 7.2 Jy at  1666 Mhz). After these measurements   amplifiers (about 10db)  were installed before the filter bank to improve the noise figure of the system. The system performance was remeasured on 06may05 using the continuum source B0428+205 (3.8 Jy at 1666 Mhz).  For each of these dates the procedure was:

• do 1 minute on/off position switching on the continuum source
• The data was centered at 1666 with a low side 1st lo.
• The wapps were setup for 1 second dumps, 6.25 (apr) and 3.125 (may) Mhz bandwidths.
• on/offs were also done with lbw using the 1500 to 1800 Mhz filter.

---

01apr05 with no amplifiers installed:

•  Fig 1: Average on source and off source. Black is polA, red is polB. The top plot is alfa while the bottom plot is lband wide. These plots include the system temperature. The za's for the measurements were alfa:7.1, lbw:11.5 degrees. The power levels were adjusted to the optimum levels (1.) on source. The off source levels dropped to about .4 of optimum.
• Fig 2: This shows srcDeflection/Offsrc. This is G/T. The two polarization's of alfa differ by about 40%. This source is not polarized at lband. Using the fluxes of 7.2 Jy, the sefd for the measurements are: Alfa: 4.3,6.0  Lbw: 2.4,2.3 Jy.
• Fig 3: This shows (G/T alfa)/(G/T lbw) The average values are:  .55 PolA, .37 polB.

---

06may05 with the amplifiers installed:

Fig 1: This shows srcDeflection/Offsrc (G/T). Black is polA and red is polB. The top plot is pixel 0, the middle plot is pixel 1, and the bottom plot is lband wide. The two polarization's of alfa differ now differ by about 30%. This source is not polarized at lband. Using the fluxes of 3.78 Jy, the sefds for the measurements are: Alfapix0: 4.4,5.7 alfaPix1:6.2,9.9  Lbw: 2.5,2.4 Jy. By comparing the lbw sefd's between 01apr05 an 06may05, the sefd's (source fluxes) probably have an error of 5%.

Fig 3: This shows (G/T alfa)/(G/T lbw) for pixel0 (top) and pixel 1 bottom. The average values are:  .56,.42 pixel 0 and .5,.25 pixel1.

---

Summary:

    Adding the amplifiers may a small improvement to polA and a larger improvement to polB.
The polB/polA performance ratios are:

• noAmps    pix0: .67
• with amps Pix0: .75
• with amps Pix1: .63

    The original reason for the oh filters were for comet observations that do a hexmap about the comet nucleus.  With alfa this could be done 7 times faster (if the outer beams were in the correct position). The effectiveness is determined by the square of the ratios of the sefd's. The numbers to use are (using the data with the amps in):

• sefdLbw: (2.5+2.4)/2 = 2.45
• sefdAlfa: Assuming pixels 2-6 have the same performance as pixel 1 the average for alfa is: ((4.4+5.7) + 6*(6.2+9.9))/(2*7)=7.6
• For alfa to match the sensitivity of lbw alfa needs to integrate (7.6/2.45)^2 = 9.6 times longer.

• lbw time: 7*1.05=7.4
• Alfa time: 9.6
• lbw is more sensitive than alfa by  sqrt(7.4/6.5) = 1.14

• The sensitivities are about equal (lbw is 1.14  time more sensitive).
• Alfa allows simultaneous measurements. This is an advantage if things are changing on the time frame of a hexmap pattern.
• Lbw  has the advantage that you can position the off positions anywhere you want, while alfa restricts you to rotations at the beam spacing.
• Calibration using the cals are available with lbw while it does not work for alfa. This has meant that some of the alfa observation time, lbw has to be used to get an idea of the calibration.
• Polarization information is available with lbw but probably not with alfa.
• If the source is extended then the comparison should be with Tsys and not G/T
• polB of alfa performs worse than polA by 30 to 40%. Pixel1 polB could probably use a higher power level.