A production build of 42 Paroscientific
barometers was available for testing.
These were divided into a test group of 23 barometers and
a control group of 19 barometers.
The test group was cycled repeatedly up and down through
test pressures of 827, 965, 1069, 965, and 827 hPa.
The control group was held continuously at
965 hPa, and was connected to the test group by opening a valve
only when the test pressure was 965 hPa.
Thus, the test group would experience hysteresis while
the control group would not.
The control group could therefore be used to monitor and
correct for real changes in delivered test pressure at the 965
hPa test points.
To eliminate temperature effects on the
barometers, all tests were conducted in a temperature-controlled
during the tests varied by less than 0.04 degrees C, and
residual temperature effects on the barometers are less than
Test pressures were applied with a CEC Model
6-201 precision dead weight pressure standard.
This is an absolute pressure standard with weights
spinning inside a vacuum bell jar.
True delivered pressure was calculated by correcting for
dead weight piston temperature, bell jar back pressure, and
local gravity. At
each test pressure, delivered pressure was continuously
monitored with another Paroscientific barometer sampling once
per second to detect weight bounce and other short term pressure
fluctuations, similar to techniques used by Wearn and Paros,
Reference 4. These
fluctuations were typically ± 0.005 hPa during data taking at a
single test pressure. The
test group of barometers was cycled twice through the test
pressures before taking data.
Data from the subsequent three cycles were used to
For each transducer, hysteresis was
calculated as the average indicated pressure at 965 hPa taken on
the decreasing pressure part of the cycle minus the average
indicated pressure at 965 hPa taken on the increasing pressure
part of the cycle. Corrections
were made for differences in actual delivered pressure because
of changes in piston temperature and bell jar back pressure.
Data from the control group of barometers
indicated actual delivered pressure at 965 hPa on the various
cycles varied by ±0.008 hPa.
The measurements from the control group were used for a
further correction to actual delivered pressure in calculating
hysteresis for the test group of barometers.
The effect of this correction was a shift in calculated
hysteresis for the test barometers by 0.0034 hPa.