MILPRF50884F
APPENDIX C
C.5.3.2 Significant digits. Unless otherwise specified (see A.3.1.1), the significant digits to be retained of an
observed value shall be in accordance with the resolution requirements of C.5.1.1.
C.5.3.3 Rounding method. For purposes of determining conformance with the specification limit, an observed
value, or a calculated value, shall be rounded "to the nearest unit" in the last right-hand significant digit to be retained
in expressing the specification limit, in accordance with the rounding method of ASTM E29. The significant digits to
be retained of an observed value or calculated value shall reflect the resolution requirements of C.5.1.1.
EXAMPLE 1: The minimum specified etchback requirement is .0002 inch. The resolution required of
the measurement device in accordance with C.5.1.1 would to the ±.00001 inch. The significant digit to
be retained would be one digit to the right of requirement, in this example, ".000RS" inch, where "R" is
the requirement and "S" is the significant digit. In a situation where a test personnel uses an instrument
capable of resolution to ±.000001 inch, a measurement of .000186 inch is determined for etchback on
that test specimen. This value can be rounded to .00019 inch or reported as taken, .000186 inch. The
observed value of .000186 inch cannot be rounded to .0002 inch.
EXAMPLE 2: The maximum specified printed wiring board thickness requirement is .090 inch. The
resolution required of the measurement device in accordance with C.5.1.1 would to the ±.0001 inch. In
this situation, the test personnel uses an instrument capable of resolution to ±.0001 inch, and records
measurements of .0899, .0900, .0898, and .0897 inch for the printed wiring board thickness on that test
specimen. The test determination of (.3594 / 4) .08985 inch is calculated. This test determination can
be rounded to .0898 inch or reported as calculated. Rounding this value to .0899 inch would not be in
accordance with rounding procedure of ASTM E29.
C.5.4 Control based on uncertainty. Test processes that have complex characteristics are best performed and
controlled by the application of uncertainty analysis. The overall uncertainty in a test or measurement process shall
be determined and the impact of said uncertainty on the product parameter tolerance shall be taken into account. The
methods used for determining uncertainty shall be defined and documented. The method selected may use any or all
combinations of the following forms:
a.
Arithmetic addition (linear), normally produces an overly conservative estimate and reflects a highly
improbable situation in which contributing errors are at their maximum limit at the same time and same
direction.
b.
Root sum square (RSS), normally applied where the errors tend to fit a normal distribution (Gaussian) and
are from independent sources.
c.
Partial derivatives, used where complex relationships exist.
d.
Monte Carlo simulation, used in very complex situations where other methods are not easily applied or do
not fit.
e.
Standard reference material (or controlled correlation device) testing providing observable data. NOTE:
Observable data from a controlled device may be relied upon to provide feedback that confirms process
performance is within statistical limits.
f.
Analysis of systematic and random errors, applying corrections as applicable.
g.
Any other recognized method of combining errors into an expression of uncertainty substantiated by an
engineering analysis.
69
For Parts Inquires submit RFQ to Parts Hangar, Inc.
© Copyright 2015 Integrated Publishing, Inc.
A Service Disabled Veteran Owned Small Business