----------------------------------------------------------------------------
A Comparison of Correlator Modes.
D. Hogg and F. Ghigo
November 4,2002
Summary
The bandpass shape for two correlator modes is found from observation
of a moderately strong calibrator. There is excellent agreement in
the shape for spectra made with Mode 1N4-0A-50-9 in two IF's, and
excellent agreement in shape for spectra made with modes 1N4-OA-50-9
AND 1N2-0A-50-9. For polarization #1 (XX) the amplitudes of the signal
in the two modes is also in excellent agreement, but for polarization
#2 the amplitude of the signal observed in mode 1N4-0A-50-9 is consistently
lower than that observed in 1N2-0A-50-9, by approximately 1.5%. Since
the polarizations in 1N4-0A-50-9 show a similar difference in each IF,
the problem may lie in this configuration.
The fact that the spectral features are similar in the two correlator
configurations, and remain fixed in frequency with a shift of the LO,
indicates that they arise somewhere before the IF mixer.
Analysis
In the spectrometer tests of October 4,2002 (TSPECTEST_09) it
was noted that the spectral features in an LBand 50 MHz
observation did not shift in frequency when observed in a
two-IF mode, with the IF's separated by 4 MHz. To explore this
further, additional observations were made on October 30,2002 as a part of
TSPECTEST_11. The raw data are found in the gbt area under that file.
The source is NVSS0106+013 = VLA 0108+015, having an LBand flux of 3.30 Jy.
Scan #8 was made with the correlator in mode 1N2-0A-50-9. The bandwidth
is 50 MHz, and there are 8191 channels of spacing 6103.51562 Hz. Scan #12
was made with the correlator in mode 1N4-0A-50-9. The bandwidth is 50 MHz,
and there are 4095 channels of spacing 12207.0312 Hz. In this mode there
are two IF's. IF#1 ranges in frequency between 1395.41220 and 1445.38779 MHz,
while IF#2 ranges between 1391.41220 and 1441.38779 MHz.
Figure 1 (tday11#8_12pl.ps) compares the spectra from scan #8 with those of
scan #12, IF#1. The spectra for polarization #1 agree quite closely
in amplitude, and all significant features are seen in each spectrum.
The spectra of polarization #2 agree in shape, but differ consistently
in amplitude, by about 1.5%. Since the spectra from the two IF's of
scan #12 agree quite closely in amplitude and shape (see Figure 2.
tday11#12pla.ps), and since in scan #12 the two polarizations differ
in amplitude by about 1.5%, it appears as if there may be an effect in mode
1N4-OA-50-9.
The following table demonstrates the effect quantitatively.
Here are the average antenna temperature in each configuration for
four spectral regions chosen to be free of interference.
Table 1
Average Antenna Temperature, in K, for Different Modes
Frequency Range Correlator Mode
(MHz) 1N2-0A-50-9 1N4-0A-50-9 1N4-0A-50-9
IF#1 IF#2
RX1 RX2 RX1 RX2 RX1 RX2
1405.3-1409.3 5.271 5.270 5.288 5.194 5.296 5.184
1410.7-1411.7 5.366 5.353 5.385 5.267 5.395 5.272
1415.5-1420.0 5.270 5.304 5.290 5.217 5.291 5.234
1421.0-1430.0 5.413 5.413 5.430 5.331 5.422 5.431
Average value 5.330 5.335 5.348 5.252 5.351 5.258
StdDev 0.071 0.062 0.071 0.061 0.067 0.066
K/Jy 1.62 1.62 1.62 1.59 1.62 1.59
Ratio 1N2/1N4 0.997 1.016 0.997 1.015
Ratio IF#1/IF#2 0.999 0.999
It is important to note that since the spectral features are similar in
the two correlator configurations, and remain fixed in frequency with a
shift of the LO, they must arise somewhere before the IF mixer.
The value of K/Jy is somewhat lower than in earlier test runs. In Test_09
on October 4,2002 the value was 1.835. A pointing observation was made
immediately before taking the spectra and appeared to be normal, although
it was noted at the time that the amplitude of the calibrator was smaller
than expected.