97-98 t0 preliminary results
Preliminary t0s from 7-May-98
These results include the corrections from the interstring analysis,
and they are accurate to ~5 ns.
Here is a detailed description
from Christopher of the problems in the standard analysis.
Use < shift-click> to download with Netscape Navigator.
97-98 Fiber lengths 24-Feb-98
The 97/98 fiber lengths are derived from several sets of measurements
made at UCI (Oct-95 and Oct-96) and the south pole (Jan-97, Dec-97, Jan-98,
and Feb-98). All of these measurements have been cross-calibrated, and
normalized to the complete set of measurements made in Dec-97. The results
are reported in the
multimode length table and
single-mode length table.
I make three assumptions in my analysis.
-
The calibration factor for the Dec-97 measurements is
constant, because I made the measurements over a short period of time.
- The calibration factor for any one string
is constant, because the measurements are also done over a short
period of time by one person.
- That there are no DC offsets in the 850 nm OTDR measurements.
I compute the ratios of the measurements for each fiber with respect
to the Dec-97 measurements. The string-by-string
ratio plot and the
MM and
SM
ratio histograms shows the calibration differences.
Based on these results, I group the measurements as shown by the dotted lines
in the string-by-string plot, calculate the
statistics
for each group, and rescale the measurements with the median
to the Dec-97 scale.
The master length results use the first of
- the rescaled Jan-98 measurements,
- the Dec-97 measurement,
- or the rescaled UCI measurements.
After rescaling there are still some discrepancies. These are
listed in string/fiber order
or sorted by magnitude.
Please note that string 9, fiber 3, MM & SM are particularly unreliable,
because they seem to have changed length during 1997.
The
multi-mode residuals and
single-mode residuals
are the differences of the original sets from the final master set.
- 'o' indicates measurements with no measureable return pulse.
- '+' indicates measurements with (0 < return <= 4)
- 'x' indicates measurements with (5 < return)
The Dec-97 single-mode residuals are plotted with and without
the 'o' data points.
97-98 Fiber transit times 24-Feb-98
The fiber transit times are determined by calibrating the previous
fiber length measurements with the 532 nm OTDR. They are available
A zero entry indicates no data for the fiber.
The individual ratios are
histogrammed here,
and the measurement statistics are
here.
At this stage of the analysis I am convinced that the multi-mode
results are more reliable than the single mode results, and that the
multi-mode fibers should yield better t0s as long as there is enough
data to characterize the LE distribution.
