It has always been my lab's policy that if an OEM calibration procedure is available, use that over a T.O. (or NAVAIR or Army T.B.) procedure for the same unit, because quite often the military writes the T.O.s to fit their needs, takes fewer readings, or more, depending on the unit. Take a Fluke 5500A calibrator for example, where the T.O. overkills some parts compared to the manufacturer's procedure, yet loosens tolerances and skips over some sections. Example: The mfg. procedure calls for readings of square wave AC in decades of 3V at certain frequencies (e.g. 300 mV, 3 V, 30 V @ 10 Hz, 1 kHz, 20 kHz, 100 kHz), whereas T.O. 33K8-4-1093-1 calls for the readings to be in decades of 6V at more frequencies in different ranges (e.g. 600 mV, 6 V, 60V @ 9.5 Hz, 40 Hz, 0.9 kHz, 19 kHz, and 100 kHz). And the T.O. also tests the triangle and truncated sine function of the 5500A, whereas nothing is mentioned in the mfg. procedure. To top it off, the manufacturer's procedure's test point limits are at 90 day spec, whereas the T.O. specs are at 1 year spec.
The T.O. also takes into consideration what should be in the inventory of most PMEL labs (e.g. just about all PMELs have a set of ESI SR-1010 standard decade resistors on hand, whereas very few have Fluke 742A standard resistors, as called for in the mfg. procedure.) Our lab doesn't have 742As, but does have the SR-1010s, so we refer to that part of the T.O. for setup instructions, but use the mfg. procedure for the test points.
Having said that, would we actually have to write an in-house procedure that combines the two in order to satisfy the auditors?