A Study of the Effect of Changing Power Levels
(Lband observing using spectrometer 50 MHz, 9-level mode)
D. Hogg
December 20,2002
Summary
A series of observations explored the effects of changing the
level of the input power to the spectrometer. It is found that
an increase in input power decreases the apparent temperature.
The effects are more pronounced in polarization X than in Y.
The effects are more pronounced when the
input power level is increased than when it is decreased, and
the decrease is more pronounced at the high frequency end of the
bandpass. The effect appears to be more pronounced when observing
a strong continuum source.
Observations and Analysis
I. The Observations.
The observations were made on September 4, 2002 by R. Prestage
and me. The LBAND receiver and the spectrometer were used,
the latter in the 9-level mode (1N2-0A-50-9), over a bandwidth
of 50 MHz, resulting in a frequency spacing per pixel of 6.1 kHz.
The observations were made as a sequence of OFF/ON scans, each
scan of the OFF/ON pair taking 5 minutes. The sequence began
with a carefully balanced pair (balancing done off-source),
followed by several pairs in which the power level into the
correlator was altered by changing the attenuation a known amount.
Several continuum sources spanning a range of flux densities were
observed, as was one galaxy selected from the Fisher catalog.
II. The Continuum Source 2203+6240 (3C440)
This source is in the VLA Calibrator book. The flux density at 20cm
is 2.75 Jy.
After a number of preliminary scans, none of which are used here, a
set of OFF/ON scans was made over a range of converter rack
attenuator settings. The data are summarized in Table 1.
Table 1: Attenuator and Spectrometer Input levels.
Scan State Attenuation Input level
Number dB (duty cycle ratios)
RX1 RX2 RX1 RX2
23 Balanced 17.75 13.25 not recorded
25 -0.5dB 17.25 12.75 1.73 1.80
27 -1.5dB 16.25 11.75 2.02 2.19
29 -3.0dB 14.75 10.25 2.44 2.61
31 -6.0dB 11.75 7.25 3.76 4.28
33 Balanced 17.75 13.25 1.613 1.724
35 Tweaked 18.50 14.125 1.351 1.376
37 +1.5dB 20.00 16.625 1.030 1.121
39 +3.0dB 21.50 17.125 0.768 0.778
41 +6.0dB 24.50 20.125 0.341 0.353
Figure 1 (tday4#23ava.ps) shows the average of the four balanced scans.
I have computed the mean value and rms for seven spectral regions
between 1396 and 1435 MHz, chosen to avoid the obvious interference
spikes. For polarization X the mean amplitude is 5.48 +/- 0.06 K,
implying a system gain of 1.99 K/Jy. For polarization 2 the
corresponding values are 5.84 +/- 0.12, and 2.12 K/Jy. It is noted
that polarization Y is apparently higher by 6.5%, and that the
variation across the band is greater in polarization 2 by a factor
of two.
Figure 2 (tday4#23pla.ps) shows all of the offset scans for polarization X
compared with the mean of the balanced scans. In general there is
good correspondence, although the scan in which the attenuation has
been reduced by 6dB (scan#31) clearly deviates at the high frequency end.
However, the individual scans are difficult to distinguish. A more
useful approach is to focus on the two scans having the largest offset
from balance.
Figure 3 (tday4#23plc.ps) compares the scans that are offset
by 6dB with the mean of the balanced scans, for polarization X.
Figure 4 (tday4#23pld.ps) gives a similar display for polarization Y.
Table 2 gives a quantitative summary of these plots.
Table 2
Frequency Ratio to Balanced Scan
Range Poln 1 Poln 2
(MHz) -6dB +6dB -6dB +6dB
1396:1399 0.9976 0.9753 1.0084 1.0208
1416:1419 0.9658 0.9909 1.0032 1.0230
1432:1435 0.9296 1.0012 0.9785 1.0208
There is a significant change in the shape of the bandpass for
polarization X, (-6dB), with a decrease in Ta* of 7% from the lower
edge to the higher edge. There is probably a decrease of 3%
for polarization Y (-6dB) and a change of +2% for polarization X
(+6dB), but these drifts are of lower significance. The value
for polarization Y (+6db) appears to be greater by a constant
of 2%.
III. The Standard Continuum Source 3C48
This source was observed in three OFF/ON scans == #55 in balance, #57
with 3dB removed, and #59 with 3dB added.
The three scans are displayed in Figure 5 (tday4#55pla.ps) (polarization X)
and Figure 6 (tday4#55plb.ps) (polarization Y).
The results are summarized in Table 3. The average
value over the balanced spectrum is 29.55 +/- 0.51 and 30.86 +/- 0.67
for polarizations X and Y respectively. The corresponding system gains,
in K/Jy, are 1.79 and 1.87 for an assumed flux density of 16.50 Jy.
Table 3
Frequency Balanced Ratio to Balanced Scan
Range Values (K) Poln 1 Poln 2
(MHz) RX1 RX2 -3dB +3dB -3dB +3dB
1396:1399 29.90 31.78 0.9693 1.0340 0.9946 1.0189
1416:1419 29.60 30.75 0.9486 1.0161 0.9908 1.0139
1431:1434 29.58 31.50 0.9315 1.0319 0.9806 1.0102
The trends are similar to those seen in 2203+6240 above. A decrease of
3 dB introduces a decrease of intensity with increasing frequency, of
3.5% for polarization X and 1.5% for polarization Y. Increasing the
attenuation increases the intensity by 3% in polarization X, and
about half that in polarization Y, without introducing a significant
slope to the baseline.
IV. The Normal Galaxy UGC2173
This object was observed as a test of the impact of varying power
levels on a spectral line. The galaxy is one which has been observed
by Fisher.
Two scans, #76 and #84, are OFF/ON pairs made with the input to the
correlator balanced. Each phase of pair was observed for five minutes.
Figure 7 (tday4#76ava.ps) shows the 50 MHz spectrum obtained by averaging
the two scans. There appears to be a galaxy in the reference position
at 1408 MHz, or a redshift of approximately 2600 km/s. There is galactic
HI at 1420.4 MHz. There is a strong signal in the second polarization
only in the region around 1442 MHz. The signal from UGC2173 is at
1416 MHz.
Figure 8 (tday4#76avb.ps) focuses on the galaxy profile.
Table 1 compares the properties of the HI profile as seen in this
figure with those reported in the Fisher catalog. In deriving the
flux scale I use the values found for 3C48 in scan#55. These values
are 1.79 K/Jy and 1.87 K/Jy for the two channels. The agreement
between the present observations and those of Fisher is good.
Table 4
A Comparison Between Fisher and the Mean of Scans 76,84
Quantity Fisher IF#1 IF#2 Mean
Peak Intensity (Jy) 0.340 0.370 0.360 0.350
Line Width @ 20% (km/s) 412.1 411.9
Systemic Velocity (km/s) 991.9 992.9
Flux Integral (Jy km/s) 101.1 102.5 101.8
I then made OFF/ON observation of the galaxy with several values of the
attenuation at the input to the correlator. The results are summarized
in Table 5. In computing the areas the bad channels associated with
the birdie at 1414.89 MHz has been clipped.
Table 5
A Comparison of Fluxes Measured at Various Power Levels
Scan Attn Input DutyC. Profile Area K km/s Ratio to Avg
RX1 RX2 RX1 RX2 RX1 RX2
Avg 0dB -- -- 182.03 191.91 --- ---
#76 0dB 1.411 1.406 182.45 192.82 0.9968 1.0047
#78 -3dB 2.203 2.434 186.93 193.91 1.0213 1.0104
#80 -6dB 3.408 3.640 201.87 194.06 1.1029 1.0112
#82 +6dB 0.352 0.340 179.11 192.70 0.9786 1.0041
#84 0dB -- -- 183.62 191.00 1.0032 0.9953
Polarization X is sensitive to changing power levels. A reduction of
attenuation of 3dB increases the apparent amplitude of the profile
by 2%, and a reduction of 6dB increases the amplitude by 10%. That this
is a real change in the profile is illustrated in Figure 9 (tday4#76avc.ps)
where the data for scan#80 (-6dB) are superimposed on the averaged
balanced profile.
Polarization X, scan#80, in the dark blue, is the
highest signal, and is clearly significantly higher than the red trace
for polarization X on the average.
An increase of 6dB results in a decrease in profile integral of about 2%.
This is marginally significant if the inherent accuracy of the measurement
is about +/- 0.5%, as judged by the balanced scans #76 and #84.
The second polarization is clearly much less sensitive to changing power
levels. The decrease of 6dB resulted in an apparent increase of 1.1%, but
this is not significant, and is in any case a factor of nine less than
polarization 1. This is also illustrated in Figure 9, where
the two traces for polarization two (light blue, green) agree so well that
they are difficult to distinguish.
V. The Strong Continuum Calibrator 3C147
This source was observed in a set of five OFF/ON pairs, with the power
levels balanced, and with attenuation changed by +/- 3dB and +/- 6dB.
The balanced scan, #95, was broken into nine spectral regions, each of
width 2 MHz, and the average Ta* was measured in each region. The
average value and it dispersion are, for CH:1(XX) 40.39 +/- 0.54 K,
and for CH:2(YY) 42.24 +/- 0.93. As has been observed in other tests,
the second polarization has apparently higher gain, in this case by
a factor 1.046, and somewhat greater baseline structure, in this case
by a factor of 1.7. Not surprisingly the dispersion in the value of
the apparent Ta* is dominated by the structure in the baseline; the
receiver noise is expected to be 14 mK.
If I adopt 22.5 Jy as the value of the flux density of 3C147 at this
frequency the values of Ta* above imply system gains of 1.795 and 1.877
K/Jy for polarizations X and Y, respectively, in excellent agreement
with the values found from 3C48.
Figure 10 (tday4#95pla.ps) and Figure 11 (tday4#95plb.ps) show the
five individual spectra for polarization X and Y, respectively.
It is immediately obvious from these figures that the effects of the
change in input power level are much greater than were seen for the
weaker sources. For example, a reduction of 6 dB (an increase in the
input power) reduced the apparent Ta* at 1435 MHz by 25%.
To quantify these effects I have computed the mean Ta* in nine frequency
bands for each spectrum. I then normalized the results for the data taken
with altered attenuation by the values observed with the system balanced,
in the appropriate frequency bin. Thus if the change in power level had no
effect the ratio is 1.0; if the band shape is unchanged the slope, measured
as a percentage change over 40 MHz, is 0.0. The results are summarized in
Table 6.
Table 6
CH:1(XX) Polarization X
Scan Nominal Attn Duty Mean Dispersion Slope
Number Reading Cycles %/40MHz
103 -3dB 14.25 4.412 0.928 0.015 -4.7
97 -6dB 11.25 5.967 0.796 0.048 -18.3
101 +3dB 20.25 1.990 1.032 0.008 +1.9
99 +6dB 23.25 1.152 1.041 0.011 +2.7
CH:2(YY) Polarization Y
Scan Nominal Attn Duty Mean Dispersion Slope
Number Reading Cycles %/40MHz
103 -3dB 10.125 4.553 0.987 0.008 -2.2
97 -6dB 7.125 7.043 0.952 0.057 -18.1
101 +3dB 16.125 1.814 1.016 0.002 +0.4
99 +6dB 19.125 1.093 1.027 0.004 +1.1
The important features are as follow:
a) The effect of increasing the input power (-6dB) is much greater than
that of reducing the input power. On average an increase in power
in polarization 1, the worst case, results in a decrease of Ta* of
20%, whereas a reduction of input power by 6dB increases the
average value of Ta* by 4%.
b) Polarization X is much more affected than is polarization Y.
c) Increasing the input power changes the shape of the spectrum,
in the sense that the reduction of Ta* is greater at the high
frequency end. In the example of -6dB the spectral slope is
18% (over 40 MHz) compared with the balanced scan.
d) The effect of varying the input power level is much greater in
the spectrum of 3C48 than is found in the previous studies of
the weaker sources 2203+6240 and UGC2173, where the changes are
typically 2% or less, and where even at a change of -6dB the
effect is about 7%. Unfortunately 3C48 was not observed at
+/- 6dB so a true comparison can not be made. However, the
trends are the same, in that an increase in input power decreases
the apparent temperature, the effects are more pronounced in
polarization X, the effects are more pronounced when the input
power level is increased, and the decrease is more pronounced at
the high frequency end of the bandpass.
--------------------------------------------------------------