DH Instruments GFS2102, Руководство по эксплуатации и обслуживанию, Страница: 70 / 206

GFS2102™ OPERATION AND MAINTENANCE MANUAL
© 2007 DH Instruments, a Fluke Company Page 54
operation and programmed start times allows a user to complete a test overnight using a
cylinder freshly filled at the end of the workday, when the user would otherwise have to wait
until the next day. (See Section 5.5.2.4)
4.4.3 COMPARING REFERENCE AND DUT MEASUREMENTS
USING FLOW VS. MASS CALCULATIONS
The GFS2102 system makes a direct measurement of the mass that flows from the reference
gas cylinder during operation. At the end of an accumulation, an accurate measure of the
total accumulated mass, or the average mass flow over the total elapsed time, is available.
GFS Tools software reads and records the flow output from the device under test during an
accumulation and calculates both the average DUT flow and the total mass measured by the
DUT by integrating all of the mass flow values from the DUT. Our assumption is that the
output of the DUT is a mass flow measurement, not a totalized mass. So the user has an
opportunity to compare the reference and DUT measurements in either total mass or mass
flow terms. A pure view of the gravimetric nature of the GFS2102 system may lead one to
suspect that using total mass eliminates some time-related uncertainty from the GFS
measurement and is a more correct choice. But when the logging of measurements from
both the reference and DUT are considered, it is apparent that using the average mass flow
is likely to produce the best result.
Although every effort is made to integrate the DUT flow over a time period that matches the time
used for the GFS measurement, the update rate of the readings from the DUT will force some
interpolation between DUT values in order to match the DUT timing to that of the GFS. Also, any
delay or error in timing the start and stop times of the GFS and DUT will result in some finite error
that can be significant. The case is worst at higher flows. When the average DUT flow is used
instead, a very good measure of the average DUT flow over a time period that is very close to
that of the GFS measurement can be obtained. The time measurement-related errors that affect
the GFS average flow measurement are handled within the GFS system to produce very good
results when flow output is used for comparisons.
4.4.4 LEAKS
Leaks are of great concern when using the GFS2102. Any loss of gas mass between the
Reference Gas Cylinder and the DUT(s) will impinge upon the accuracy of the
measurements. It is recommended to test the flow path for leaks often, particularly with
every hardware setup change, i.e. every time a different molbloc, MFC, molstic, Reference
Gas Cylinder or catenary tube is used, or a heat exchanger is installed in the flow path.
This section describes three methods to check for leaks: an automated leak check in a test
definition; a semi-automated leak check of the flow path up to the isolation valve; and a semi-
automated leak check to a external manual isolation valve placed in the flow path by the
user. Regardless of the method used, the user must establish what the maximum
“acceptable” leak rate can be for the gas and flow rate to be tested. When the uncertainty of
the GFS accumulation is analyzed, if a leak was detected and could not be eliminated, the
error contributed by the leak should be treated as an asymmetrical distribution because a
leak would always be in one direction: gas can leak out of the system, but it cannot leak in.
To establish the significance of a leak, or what amount of leak would be acceptable during a
test, the user must consider the gas being tested, the minimum flow point of the gas during
the test, the uncertainty target being sought, and the amount of time to get a valid leak test
result. Using a 1E2-L molbloc as an example being tested in Nitrogen, a typical minimum
flow of 10% of full scale is 10 sccm (2.084E-1 mg/s), and if the user is seeking less than
0.02% effect from a potential leak then a leak rate under 0.002 sccm (4.168E-5 mg/s) would
be acceptable. In order to ensure a valid leak test result, the leak test should run a minimum
of 10% of the accumulation time of the lowest flow point: if the accumulation takes 26 hours
the leak test should run for AT LEAST 2.6 hours, if the accumulation takes 10 hours the leak
test should run for AT LEAST 1 hour.