IBM Magstar MP 3570 C Series, Информация по обслуживанию

Страница: 368 / 675

Note: FID 97 only occurred in the columns in the right-hand portion of the work sheet, but that is
because it was considered a low probability of failure when taken one FID at a time. If a FID
occurs often, but always falls in a low probability position, That FID may be causing the
problems but due to limitations of error logging, it never shows up as a high probability of
being the failure. This error log analysis procedure is intended to show up some of these
problems.
7. Last, look up the FID number and associated names of the FRUs they represent (in our example, the
three FIDs are listed under Legend). Now use your training, intuition, or experience to decide which
FRU is most likely to be causing the problem. In our example, the likely failing FRU might be the
picker assembly, the loader assembly, or the library control card. A decision at this time might be to
replace the FID 9A FRU (the picker assembly) first for the following reasons:
a. FID 9A occurs most often of all high probability FIDs (5 times), and its position is, in all cases, in
the high probability area (left most three columns). It does, however, have a medium variation.
Note: Variation is used here to define a measure of how often a FID occurs exactly the same
way in a given error log sample. For example, if the FID 9A had occurred as the first FID
after a FID 98 all five times instead of how it did occur, it would have a small variation or a
high correlation. Small variation (similar to high correlation) Error Log FID Entries tend to
point to those FIDs as the most likely failing components. Large variation (similar to low
correlation) Error Log FID Entries tend to point to those FIDs as not the most likely failing
components.
b. The FID BC occurs quite often, but has a slightly lower probability based upon which column it
appears. FID BC has a slightly smaller variation in the Error Log FID Entries, which may point to
it as the failing component.
c. The FID 97 seems to be a medium probability due to its position in the middle columns (that are
used), but has the greatest variation in the Error Log FID Entries.
Had the FID BC had a very small variation in the Error Log FID Entries (such as only one or two
different FIDs), or appeared in a slightly higher probability of failure position (some entries in the
left-most column), it might have been selected as the FRU to replace first. Had the FID 97 had a
smaller variation in the Error Log FID Entries (such as only one or two different FIDs), it might have
been selected as the FRU to replace first.
These decisions must rely heavily on your experience and knowledge of the device, the timing of the
errors that occurred, error history of this device, and your customer environment.
PROC-40
IBM 3570 MI