Date: Sat, 28 Feb 2009 22:38:26 +0000
From: Richard Hills <firstname.lastname@example.org>
To: Todd R. Hunter <email@example.com>
Cc: Kamaljeet S Saini <firstname.lastname@example.org>, Bojan Nikolic <email@example.com, Baltasar Vila Vilaro <firstname.lastname@example.org>
Subject: Total Power stability
I was stimulated by a question from Bojan (about doing Moon scans to measure
the far-out sidelobe pattern of ALMA dishes) to look into the issue of what
total power stability we might reasonably expect to get from our SIS
I picked some samples - actually these are the better ones - from the test
report that I have on FE number 1 and made this spread-sheet that estimates
the effective bandwidth (or alternatively the effective noise temperature).
I was somewhat dismayed at the results. Assuming that the actual bandwidth
is 4GHz, the receivers are a long way from ideal. This means that if we are
going to do real total power measurements OTF we are going to have to scan
fast. Even when we use the nutator (and assuming that the nutator is indeed
able to achieve the hoped-for 10 Hz chop - which means that the 0.05 second
Allan variance is the relevant data point) then the effective bandwidths are
only ~200MHz for bands 3 and 9 about ~45MHz for band 7.
I wonder whether people thinking about how we are going to use ALMA for
total power measurements have taken this into account in modelling the
problem. I see that there is a typically thorough and convincing memo by
Larry D'Addario in the ALMA series - number 466
- which argues that we
should be able to get to the point where we are limited by the atmosphere
and not the receivers. I am a bit skeptical about this, however, since we
have been using bolometers with an 8Hz nutator on Mauna Kea for years and
most of the time the limit is not the atmosphere. I would have thought
those detectors were more sensitive than the ALMA SIS ones.
I just wanted to check what we know about the stability of the measurement
system. I see from the test procedure (pasted into the spread-sheet for
reference) that we are just using a power meter. What does the Allan
variance look like with a reference source on the input?
Also do we know where in the system the fluctuations are occurring?
What happens if we do something like turn off the LO and turn up the mixer
bias to above the gap so the junction is a shot-noise source?