[mmaimcal] 183 GHz

[Simon Radford]

>From richard at mrao.cam.ac.uk Mon Jun 28 07:56 MST 1999
From: Richard Hills <richard at mrao.cam.ac.uk>
To: rkurz at eso.org, sradford at NRAO.EDU, mwiedner at randonnee.harvard.edu,
        roy at oso.chalmers.se, belitsky at oso.chalmers.se, denis at oso.chalmers.se,
        R.WADE at RL.AC.UK, JSR at mrao.cam.ac.uk, cgh at ednet.co.uk, LNYMAN at eso.org,
        AOTAROLA at eso.org, gdelgado at eso.org
CC: richard at mrao.cam.ac.uk
Subject: 183 GHz radiometers in Chile
Content-Type: text
Content-Length: 13961

         Report on a visit to the ALMA site - Radiometry aspects.


John Richer, Richard Wade and I visited Chile in the week 20th to 27th 
June and met Guillermo Delgado, Angel Otarola and Lars-Ake Nyman amongst 
others.  The visit served as a general introduction for all three of us to
the situation in Chile and the nature of the site, but my main focus was 
on the 183 GHz radiometers.  Here are my notes on that topic (with actions
marked by **). I am afraid they will be too detailed for some people I am
sending this to but I thought it would at least give you a feel for what 
is going on.

We went up to the Chajnantor site on 23 and 24th of June.  The radiometers
were basically operating well, but the following problems were noted (all
on the East radiometer): 

1) the control micro was hung.  It had apparently failed when the power
had gone down and had not restarted correctly.  It did come up OK after
switching the power on and off again.  It was not clear what was wrong -
perhaps too cold when restart tried? 

**REH to check with Colin Hall whether the watchdog timer is implemented
and whether this would solve such a problem. 

2) the Mylar window had failed.  It was replaced with a piece of thin
styrofoam for now. 

** REH to send Goretex material to replace both.

3) there was also a misalignment (of about 22 mm to the West and 7 mm low)
of the beam on the steerable mirror, M1.  This was corrected by adjusting
M2.  One turn on the screw moves the beam about 30 mm at M1. 

The following work was then carried out on the East radiometer:

1) the case was painted white to keep it cooler in sunlight.  It is hoped
that this will prevent the RF plate from getting above its nominal
temperature in the middle of the day.  If this does not work then some of
the thermal insulation should be removed. 

2) a simple temperature sensor was installed inside the case which can be
monitored to help with the above problem. 

We then worked on the West radiometer: 

1) the temperatures of the Hot and Warm loads were adjusted to 100C and
35C.  (This last value was a mistake in that it was intended to match the
East radiometer but it turned out that this was at 30C.)  In general it is
be better to have the Warm loads a cool as possible to maximise the
temperature difference between Hot and Warm.  The temperatures should
however be above the hottest temperature that ever occurs inside the case
so that the temperature regulation can operate properly.  I would guess
that we could lower it to 25C but I do not think this is worth the effort
at present. 

2) a fault that was causing zero readings on the RF plate temperature
counter was found to be due to a short on the opto-isolator board.  This
was cleared just by lifting the wire. 

3) the hardware limits were checked. One micro-switch had been damaged. 
We bent it straight and it then seemed to work OK.  The range on the lower
"Az"  axis is -23.2 to 169.8 and on the upper "Alt" axis it is -18.4 to
316.0.  The values displayed on the screen are in degrees - clockwise seen
from above - with respect to the zero points (the position the mirror
drives to on startup). 

4)  we also confirmed that the position commands are interpreted as being
with respect to the "home" position - the one where the mirror reflects
the beam back to the detector.  On the West radiometer this is at (Az,Alt)
= (67.4,157.9) and, I think, at (64.1,152.1) on the East one.  At this
point it is not clear to me when these values are established and how
accurate they are.  We found that the definitions of axis 1 and 2 were
reversed between the commands and the screen.  The software limits are
presently restricting the range of motion. 

** REH to consult with Colin Hall about what would be the best way of
tidying up this topic.  We should at least open up the software limits as
soon as possible. 

5)  data were taken with ambient and LN2 load placed in front of the
window to check calibration and stability.  We found that the conventional
technique of holding flat Eccosorb in front of the window was giving high
values of apparent temperature (although this had seemed to work OK on
previous occasions). A somewhat better-designed load, with a "corner" of
Eccosorb viewed through the side of a polystyrene bucket, gave values
close to those expected.  This seems to imply that the Eccosorb we were
using is not very "black" at 183 GHz. We should try to make some better
loads for calibration purposes. 

6) we looked for the reasons why the count rates were low (~60 kHz) on the
1.2 GHz channel.  There was no evidence of bad connections, etc., and it
appeared that any differences between channels were appearing in the
v-to-f's.  It was noted that the present count rates are really too low
relative to the noise level expected from the radiometer equation. 

** REH to check if the rates can be raised. 

7) we were able to detect the Sun by first moving the mirror to shine the
visible light into the window and then peaking up the counts on the 7.8
GHz channel.  A scan across the Sun gave a beamwidth (FWHM) of about 2.5
degrees.  (Some error in this will have been introduced by the Sun's
motion of ~0.25 degrees per minute.) This value was significantly less
than I had expected but is actually close to that obtained from a more
careful check on the optical design. It seems that making such pointing
measurements on the Sun under control of the Labview PC will be the best
way of checking alignment of the whole system. 

**GD to consider what software will be needed to do this. (Note that it
should also be possible to check the coupling to the beam from the
measured antenna temperature of the Sun.) 

8) We measured the beam size at various points in the system. This was
done by a) measuring the coupling to Eccosorb discs of different sizes
(40, 75 and 100 mm diameter) and b) moving a straight Eccosorb edge in
from the sides and finding the separations between the positions where a
quarter of the power was absorbed.  Neither of these are very accurate
(especially when carried out at 5000 m) but the results appear consistent
with the design values. These give a (1/e amplitude) radius of 20 mm at
the window, 13.5 mm at the waist between M2 and M1 and about 28 mm at M1. 
It was noted that the beam was displaced by about 5 mm to the West at the
window, which is not very significant but could indicate that the flip
mirror adjustment needs to be tweaked. 


In further discussions after the visit to the site the following plan for
the next phases of work emerged: 

1) The radiometers should already be collecting valid data which can be
compared to the phases gathered by the 11.2 GHz interferometer.  We agreed
that GD would continue to process the radiometer data for the moment and
that as soon as additional effort is available in the UK (in a month or 
so?) this should be focussed on getting set up to process interferometer
data and be in a position to make comparisons with the radiometry. 

**REH to contact Simon Radford to find out what he is doing on the
interferometer data and what software exists, e.g. for removing drifts due
to movement of satellite, etc.

2) At present the agreement between the interferometer and the radiometer
on short timescales is not expected to be good because they are not
looking at exactly the same paths through the atmosphere, for several
reasons: 

a) there is a pointing offset of a few degrees due to the software limits.

b) the radiometer beams are displaced horizontally by about 3.3 metres
from those of the interferometer. 

c) the radiometer beams have a width of about 2.5 degrees compared to
about 1.2 degrees for the interferometer.  (Note that, because the
interferometer is detecting a coherent signal, the observed phase will be
weighted by the Amplitude pattern of the individual elements at the
distance of the screen that is introducing the phase fluctuations.  The
radiometers are, by contrast, measuring the incoherent emission from the
water molecules so it is the Power pattern that is relevant for them.) 

**REH to contact Simon Radford to get better information on the
interferometer antenna patterns. 

We agreed to deal with these points as follows:

a) AO to install new micro software with limits on mirror movement opened
up on next visit to site.  (If other less critical improvements are
available from Colin Hall by then they could also be installed.) 

b) AO to organise relocation of radiometers so that beam passes closer to
the edge of the interferometer dish.  A spacing of about 0.5 metres seems
to be a good goal.  It is suggested that the radiometers be re-oriented
with their axes North-South which should provide a convenient geometry and
simplify the alignment.  This move will probably take place when the Cal
Tech instrument is being installed and lifting equipment is available on
site.  This is expected to be in August. 

c) REH to check optical design and provide a new pair of mirrors M1 with
curved surfaces which will give a  beamwidth which is a good match to that
of the interferometer.  (If necessary the mirrors M2 will be replaced as
well.)  The goal is for these to be available by August. 

3) A key experiment is to put the two radiometers as close together as
possible and find how accurately they agree on the water vapour
fluctuations.  It is suggested that this experiment be carried out at the
time when the radiometers are being relocated anyway.  The radiometers
should if possible be run in this mode for a long enough time to sample a
range of weather conditions.  With luck this might occur in a week or so
in August/September. 

**AO and GD to consider how to do this.  It may be possible to send a
post-doc out from Cambridge who could provide an additional pair of hands
if it is thought that this would be helpful. 

4)  In order to check on the general functioning of the instruments and
the sky conditions, it will be worth implementing a better sky-dip
procedure running under Labview.  It is suggested that this be implemented
as a scan, essentially from horizon to horizon, on the small circle which
passes through the position of the satellite used by the interferometer
and which produces a constant angle of incidence on M1.  Such a scan with
say 15 positions should only take a couple of minutes and could be run
every 2 or 3 hours.  Note that it would probably a good idea to issue a
"home" command to the radiometer position drives before starting such
scans and again before returning to the satellite postion.  We should also
chek the status of the mirror drives after each move.. 

**DG and REH to sort out the details of this.

5) It was noted that information about cloud cover would be important to
demonstrate how much different types of cloud effected the results.  We
should set up at least a standard TV camera covering the part of the sky
the experiments are looking at and should find out about the possibility
of using an IR camera to see clouds at night. 

**REH to contact Tim Hawarden at UKIRT on IR camera costs and practicalities.

6) We need to make sure the timing of the radiometer and the
interferometer samples are accurately matched.  A "synch" command should
be issued at an accurately known time before the start of each run.  If
possible the radiometer sampling interval should be decreased from 2
seconds to 1. 

**DG and AO to implement and check.

7) We also discussed various details of the radiometer data reduction
including "Planck" corrections to the radiometry formulae, fitting for
detector non-linearities and smoothing of the data values obtained on the
Hot and Warm loads before they are used to convert the sky values to
temperatures. 

**DG to implement these with input from REH when required.

8) It was agreed that other possible experiments with the radiometers such
as 1- or 2-D sky scans should be given low priority for now.  These should
enable us to learn more about the altitude of the turbulent layer(s).  It
is possible that one or more sky- mapping radiometers mounted seperately
from the antennas will be useful when ALMA is in operation.  The main goal
for now, however, has to be the test of how well the phase correction can
be made to work.  I would however like to point out that, because we have
beam-steering on the radiometers and two interferometers pointing in
slightly different dirrections, there is an opportunity to demonstrate
that we can use atmospheric parameters "learned" on one satellite to make
a "blind" correction to the phase on the other one.  This seems to be a
rather good simulation of the operational situation where we will be able
to use observations of calibration sources to find the best values to use
in the expressions relating the radiometer measurements to the
interferometer phases under the current conditions. 

9) Finally I should record a continuing worry that I have over whether or
not the flip-mirror is in fact behaving with sufficient accuracy over the
wide range of temperatures required.  Both the motor and the control
electronics are attached fairly directly to the metal case which must be
subjected to large temperature swings.  This was not what was expected
when the design was done and it is known that the motor and position
sensor are somewhat temperature sensitive.  If we find evidence for
temperature- dependent behaviour this is the most likely cause of
problems. 


Timescales

As noted above, some of the existing data should be useable for looking at
the more slowly varying components and that being taken now may be
somewhat better.  We should however be cautious about over-interpreting
what we get until the steps listed above to get good beam overlap between
the radiometer and the interferometer have been taken.  We agreed that a
reasonable goal at this stage would be to produce an initial report by the
end of this year. 

Richard Hills                                                  27 June 1999