Subject: Notes from discussion with Rob Douglas
From: Kate Scholberg (kate@neutrino.kek.jp)
Date: Tue Jan 11 2000 - 07:32:04 EST
Hi timing folks,
Here are my notes from the discussion with Rob Douglas
(Clarence and Alec also in attendance).
Kate.
============================================================
The overall idea is to develop a robust traveling time calibration
system that could move from one laboratory to the next to check the
timing, using a consistent, reliable system.
The proposed system has three components:
1. GPS receiver for surface tests
2. Time reference signal (stable oscillator)
3. Time interval counter
Also needed: 2 laptops, one for outside and one for inside.
All of these must be robust and portable. Another major item on the
budget will be travel expenses for personnel who will carry the
equipment, train onsite personnel, and perform calibrations.
Rob Douglas gave general suggestions as well as advice for specific
equipment to consider. (There may also be funding available from NRC
to buy some or all of these, since the problem we have -- of
determining time accuracy of systems remote from a clear GPS view --
is a general one they have an interest in developing general solutions
for).
GPS receiver:
The purpose of this is to verify the external time systems (usually
GPS based) of the respective laboratories. It is important that this
equipment be robust for traveling.
Rob suggests a NovaTel Millenium system, which costs about $9000 US.
The uses two frequencies for ionospheric corrections, and has way
better precision than we need (~20 ns). It can do even better with
further calibration (local multipath corrections). It has other bells
and whistles, too, such as convenient output that geodesists can use
to determine precise positions. Rob prefers this particular model
because they have experience with it at NRC and they know it to be
very reliable; also it is designed for portability. However, many
other models would be fine, too.
2. For the time transfer standard (the "alarm clock", formerly known
as the "time bomb"), Rob suggests a rubidium oscillator. (Ovenized
crystals may not be quite good enough).
His specific recommendation is a Datum MRT. This contains an "FRK" Rb
oscillator. Cost is about $15000 for the fanciest model, with nifty
features like a pps output. However there are ways to save money:
a used one of these costs about $2500, and the bare Rb oscillators can be had
separately for as little as $500 (used). The used ones are sold under
warranty.
Some other information about these oscillators:
-- The Datum MRK weighs 30 lb, most of which is battery...
with better batteries, this weight could be reduced?
-- The bare Rb oscillator weighs aobut 2.9 lb, and is about 4" on a side
-- Power requirement (dependent on external temperature) is less
than 30 W (less than a laptop).
-- Frequency stability is somewhat sensitive to pressure: df/f
~5e-13 per mbar. Pressure differences are potentially a problem for
Sudbury, for which there is a 300 mbar pressure change going
underground (compare to 100 mbar change for a big weather change) but
it should still be OK. (SK and LNGS would have less of a problem).
-- There's also some sensitivity to magnetic fields: df/f ~
3e-11/gauss, so we would have to be careful about getting too close to
major powered machinery... at very high fields (such as would be
dangerous for someone wearing a pacemaker), there can be permanent
magnetization of the shielding and persistent frequency shifts. This
is probably not a problem for us.
-- Other portable Rb clock models are available... Rob prefers those
without tricky microprocessors inside, which allow less flexibility
and sometimes give unreliable output.
-- Rb clock lifetimes are about 5 years of continuous running. Bulbs
can be replaced for about $500. However we don't have to operate them
continuously, so this isn't much of an issue for us.
Time Interval Counter:
A possibility previously discussed was to have the time standard be
internal to the time interval counter (TIC models are available with
both ovenized crystal and Rb internal oscillators). However Rob
recommends against this because it would give us less flexibility; he
prefers a TIC with an external input so that it can act as a slave to
an external time standard, such as a separately purchased Rb clock.
A promising model is the Stanford Instruments SR620, which is
available for about $5000 ($9000 with ovenized crystal internal
oscillator, $14850 for Rb).
Computers:
A laptop with serial port connections to be used for data acquisition
will be needed. Any old one with a 486 chip or better will do fine.
We need two-- one for work on the surface, and a second one
to bring inside the laboratories.
Other knickknacks:
A few other odds and ends will be needed (partly dependent on
what specific stuff we get): carrying cases, batteries, connectors,
etc.
Travel:
A significant part of the cost may be travel, particularly in
the initial stages of the program.
Rob suggests an initial "flurry" of calibration measurements, after
which we can (we hope) demonstrate to the community that laboratory
timing is reliable. The flurry should include data taken during both hot
and cold weather. Then, we can make measurements at regular, but
longer intervals. Possible schedule: once per month for 8 months or
so, spanning winter and summer, followed by a regular check every 6
months or so. We don't want to wait too long between checks or people
will forget details of the procedures.
We should develop a standard procedure to follow. For the
first few checks, personnel at each site can be trained in the
procedure by traveling expert(s). After that, there should be a
regular review (every year or so?) to check that procedures are
consistent.
Rob suggests that the international astronomical community may feel
more comfortable with the results if one of the stops on the round of
checks is at the BIPM in Paris, for GPS receiver calibration. He
thinks they will be happy to do this for free. This check need not be
done for every round, but perhaps at the beginning and the end of the
"flurry" would be good. BIPM is more comfortable with dealing with
pps signals.
A stop at NIST for a good technical calibration might be worthwhile.
They may charge for this service (apparently suggesting to them that
USNO might do it instead might help here!)
This archive was generated by hypermail 2a24 : Tue Jan 11 2000 - 07:32:09 EST