Reviews for ALMA Memo 505
ALMA Memo 505: Bandpass Calibration for ALMA Bacmann and Guilloteau; September 30, 2004
Review from Robert Lucas
There is so much work in this memo that I felt I had to read it
deeply; unfortunately I had not yet enough time to check the side
band ratio thoroughly enough. I will continue by I send you this
now, sorry for being so late.
- p.3 Sec 1 eqn. (3) line above: I would say "so that the power response..."
- next paragraph: can we totally neglect change of chromatism due to
focus tracking? I think a focus move should be compensated by a
delay correction. If I understand well you mention that any
additional change in chromatism is negligible?
- bullet 3. Digital filters are assumed pre-calibrated. By the way do
you know whether the calculation has been done and by who?
- p. 4 Sec 1.1 1 line after eqn. 10: It is not only needed to predict
the freq. dependence of atm. transmission, but to *correct* for it
at this point. This is worth mentioning I guess.
- p. 4 sec. 1.2 I think the goal for ALMA should be to correct the
actual atmospheric *delay* fluctuations, not only the phase
fluctuations. This due to the large bandwidths. This may be
- the WVR predicts a delay as a function of time
- the fast switching if based at the low frequency where
Delta(nu)/nu is large can also measure a delay on the calibrator
in addition to the phase, and thus apply it on the source at the
observing frequency (getting rid of 2pi ambiguities).
I think that if this can be done, then the effect you mention will
- p. 5 eqn. 14: nomenclature: I would not call J_sky the sky
emissivity; it has contribution from the sky but also from
ground-like sources. The sky emissivity is rather what is multiplied
by g*eta in this equation. Why not `apparent sky RJ equivalent
temperature'? or may be apparent emissivity? (while P_sky is the
output power on sky).
- p. 5: can we neglect the variation of eta with frequency and side band?
- p. 6, 7: It is conceptually disturbing that using 2 loads one gets a result
worse that when using one load only. In fact the single-load method
is using 2 loads, one being the blank sky. So the method with
two hardware loads should in fact use the blank sky too.
One possible solution is to use a variant of the single-load
described here, replacing the load signal with a linear combination
of the signals on the two hardware loads. This will mimic the signal
of a load at an intermediate effective temperature. One can then
choose this effective temperature so that the expression of Tcal the
dependence in Tau (thus in frequency and airmasses) is minimized.
This was the idea of the 'cooled chopper' method of B. Ulich
(Astrophysical Letters, 21, 1, 1980, p. 21-28).
Using eqn. 21, when the image sideband effect is neglected, the appropriate
combination is to have for the equivalent load:
J_load = (1- eta) J_spill + eta J_m
which gives T_Cal = J_spill -J_bg + 1./eta (J_load - J_spill) pretty
much independendtly of the optical depth, and even easier to
interpolate with frequency and airmass. Naturally J_m changes a
little with tau, so it would be good to perform an actual
- p. 17 You do not need to know the flux but you need the spectral
index, which is probably still more difficult to get!
- p. 19 can we assume the optical depths are identical on all antennas
in the extended configurations?
- p. 21 closure relations are not required to determine antenna based
gains: one can do it for N=3, and there is no amplitude closure
relation for N < 4.
- p. 22 3.3 I'm not sure that the wide bandwith ensures the flat
rejection spectrum; one could have some modulation of g and only
small ripples in the passband.
- p. 24 Are there actual plans to calculate/measure the response of
the bandpass filters (including the many available configurations
with the tunable filter design)?
Review from Rick Perley
Well, on my rather cursory review, I certainly didn't detect any
patently incorrect items. My mostly cosmetic comments are:
1) So far as I could tell, there was no text reference to either Fig. 2
or Fig. 3. And, for those two figures, the figure captions are identical,
although clearly the figures themselves are different. Furthermore
(and this comment applies to all the later figures), the difference between
yellow and aquamarine shading is not explained. (I realize that ALMA
experts will know the difference, but peons like me don't). The y-axis
labels for both these figures are different (TCAL S and TCAL D), but
I couldn't figure from these just what these figures meant.
2) While it was clear to me the need to work out the effects of
elevation and water vapor change, I was puzzled by why the effects of
changing frequency would be considered. AFter all (I reasoned) isn't
it obvious that one calibrates the bandpass (or whatever) at the frequency
of observation? So I consulted with Bryan, and learned that there is
apparently some rationale to calibrating the BP from lower frequency
observations. It sounded semi-reasonable ... Now, whether such rationale
should be explained in a memo depends on whether you're addressing the
ALMA experts who are already familiar with the history of the
argument, or you are trying to reach a wider audience. I don't know the
answer to this, but if it's the latter, then I suggest the memo begin with
an introduction giving this rationale, and perhaps some rought order-of-
magnitude estimates of the importance of these variables. The reason for
the latter suggestion is that this is *not* an easy memo to read! It's not
for the faint of heart.
I hope these minor comments are useful.
Review from Dave Woody
The paper looks fine. I didn't check the formulas in detail,
but the calculations and conclusions seem reasonable.
- 15 Oct 2004