Project: TBASErdn021210
Roger Norrod

First try of new test noise source, which incorporates a TTL controlled
noise cal with adjustable level. Injecting it into the AF Rack SF modules
1-4. Set bandwidth to 200MHz. Initial test with DCR.

tau=1sec; T=120sec; TPwCal; recording A9-A12.

Scans 1-20 (adjusted cal attenuator during scan 3). Data (Figure 1)
shows ripple due to Equipment Room temperature cycling. (See TBASErdn021209).

Began using spectrometer at scan 21. tau=20sec, T=120sec, 2 phases,
1sec switching time, blanking time 0.010sec, => 6 integrations (records) 
per scan.

Mode 1W1-00-200: One sampler, HSS0 from SF1, Bank A, Quadrant 1,
16384 lags, 200MHz.
Scan 21 st @ 15:35:55
Scan 30 fi @ 15:57:05

Figure 2 plots the mean spectral power (10-190MHz) for each record, 
normalized to the "mean of the means". Figure 3 shows the spectrum in 
units of dB with respect to the mean power, for scan 21 CalOn phase. 
Figure 4 shows the unnormalized CalOff and CalOn spectra (linear scale).

Figure 5 shows Ta/Tsys, or (Sig-Ref)/Ref, taking scan 21 as Ref and 
scans 22-30 in sequence (starting at the bottom trace) as Sig. The plots 
are offset by 0.01 units for clarity.  A mean is NOT subtracted from the
traces. The standard deviation of the nine traces is (14.87+/-0.1)X10^-4.
For the bandwidth and integration time, the radiometer equation predicts
(Sigma/Tsys = 1.235/sqrt(1.21*(T/2)*(BW/Nlags)) = 13X10^-4.  The tenth
(topmost) trace in Figure 5 is the average of the nine lower traces. The
noise did not integrate down, but I realized later that is to be expected
since the common reference spectrum produces correlated noise in the 
difference spectra. (See discussion related to Figure 12 below.)

Reconfigured spectrometer to Mode 1W2-001-200: Two samplers, HSS0 from
SF1 and HSS1 from SF5, Bank A, Quadrant 1, 8192 lags. Same integration 
and scan times.

Scan 31 st @ 16:04:02
Scan 40 fi @ 16:25:15

Figure 6 shows the mean spectrum power for each integration. The red
trace is from HSS0 and the green trace is from HSS1. Figure 7 shows
the normalized spectra in dB. Note the mode label in the title is
incorrect.  Figure 8 shows the Ta/Tsys traces.  The radiometer equation
predicts Sigma/Tsys of 9.3 X 10^04, while the nine difference spectra
have standard deviation of 10.6+/-0.1 X 10^-4.

Reconfigured spectrometer to Mode 1W2-023-200: Two samplers, HSS2 from
SF2 and HSS3 from SF6, Bank A, Quadrant 1, 8192 lags. Same integration
and scan times.

Scan 41 st @ 16:26:53
Scan 50 st @ 16:48:05

Figure 9 shows the mean spectrum power for each integration. The red
trace is from HSS2 and the green trace is from HSS3. Figure 10 shows
the normalized spectra in dB. Figure 11 shows the Ta/Tsys traces.  
The radiometer equation again predicts Sigma/Tsys of 9.3 X 10^04, 
while the nine difference spectra have standard deviation of 
10.5+/-0.1 X 10^-4. Figure 12 is Ta/Tsys but taking alternate scans
as reference and signal spectra. The upper trace is the average of the
other five, and its standard deviation is 4.7 X 10^-4, exactly 
1/sqrt(5) the noise in the components.

Conclusion: The three modes checked today worked without a hitch. The
difference spectra show excellent flatness. The noise is about 13% higher
than theoretical, perhaps due to gain fluctuations in the SF modules or
in the samplers.