Summary from Richard
(
Responses from Geoff and Todd on Oct 7 written in blue.)
I visited the NA-FEIC on Friday 26th Sept 2008. Here are some notes.
1) Calculation of efficiencies, etc.
a) We found the cause of the remaining discrepancy between Andrey
Baryshev's calculations and those done using NSI software plus the
spreadsheet and corrected it.
b) Josh will continue to press NSI on the "vector addition" problem.
They are still working on this.
c) Fred Schwab agreed to complete his rigorous calculations using the same
assumptions about the location of the subreflector as is being made by the
other approaches. This will provide a check on whether the approximations
being made by both Andrey and the NSI/spreadsheet treatments are adequate.
d) The remaining topics here are the correct treatment of defocus errors
and the calculation of the beam squint. Both require the combination of
data from Pol0 and Pol1. We sketched how to do this. Josh and Geoff will
make a first cut at implementing it.
2) The hardware.
a) I noticed to my horror that the whole source, including the rotator, is
hanging off a single short linear bearing. This means that, with the FESS
horizontal the torque due to the (offset) weight of the source, and that
due to any variable loads coming from the cables, etc., is only resisted by
the twisting stiffness of the linear bearing. These bearing are not
designed to carry loads in that direction and it is very soft, as can
easily be determined be pushing gently on the source. There is no chance
of obtaining the ~micron level accuracy required with such an arrangement.
(1 micron of z-movement is roughly 1 degree at 900 GHz.) I fear that this
needs to be modified, certainly if we want to get meaningful results about
changes in beam direction as the FE is tilted. This may not be a very big
deal and it probably need not delay us from getting on with basic beam
measurements. Until this is fixed exceptional care needs to be taken in
the way that the cables are supported.
We have changed the mounting for the cable to the stage (Todd). This has
been installed, and should support the cable in a way that lessens the
effect on the stage. This has been sent to NSI who are looking into
improvements.
b) The there was some sort of "flicker" on the signals being measured,
which we could see on the spectrum analyzer. This looked like a loop
jumping out of lock but Geoff pointed out that with this set-up (that for
the room-temperature measurements of the standard horn) there are no phase
locked loops in the system. This is probably not at a level that will
prevent data being taken.
This has been investigated (see attached note from Morgan McLEod). We now operate with the M&C off and do not see this any more.
c) There is something in the measurement system that is limiting the signal
to noise ratio to on the signal arriving at the V/A ~40dB in 1MHz. This is
not present on the IF direct out of the receiver. Again it is probably not
a limit at the moment but it should not be there and it's origin should be
found. Linearity should also be checked.
Linearity is always checked when setting up on a new band or receiver. The
"extra IF plate" appears to be the limit. On further investigation the
output of the IF of the receiver through the IF processor was not linear.
The max range (for the Schottky receiver) was about 50 dB in 1 MHz.
d) The integration time being used (3.3msec) is very short compared to the
time between samples (typically 125msec). If the s/n ratio is very high
this may not matter, but it is certainly worth experimenting with longer
times.
Todd and I tried different integration times (from 3 up to 333 msec), one after
another, and there was an improvement in the ripple on the main beam (see this plot). Higher integration times (~100msec)
will be used in the future as there is no change in scan time.
e) The standard horn has been placed rather a long way from the source
which will mean that a large area needs to be scanned. It should be moved
up if possible.
The receiver was moved up, the distance between the front of the open ended
waveguide and the front face of the lens on the horn is 163 mm.
REH 29th Sept 08
--
ToddHunter - 29 Sep 2008
*
beam2.pdf: Comparison of scans 29 and 30 on Sep 30, 2008