User Guide

E-8 Ampex 1308911-02

Model No.

Running Head

Operational Statistics Parameters 2 (Log Page 33h) Statistics

2. Events With Parity Errors. Parity errors are detected at scan (5 ms) intervals. There may be more

than one parity error per interval. All parity errors within the same interval are counted as just

one event. Parity error indications are automatically reset at the start of each interval.

3. Statistics Related to Zero Erasure Counts. The C1, C2 and C3 EDACs provide status for each

processed codeword, relating to the number of erasures ﬂagged within the codeword. However

a count of zero erasures may come from either of two sources. Either no erasures have been

ﬂagged while receiving a codeword, or too many (11 or more) were ﬂagged, which may lead to

a codeword failure. If codeword failures due to erasure (E) overﬂow occur too often,

adjustments to the erasure source may be advised (such as turning erasure input select to OFF,

or increasing the PBUS value on the preceding EDAC). On the other hand, if codewords that

had no erasures ﬂagged still required correction too often, then the erasure source may not be

detecting data errors adequately. Lowering the PBUS value on the preceding EDAC may help

in this case.

4. Read Recovery/Retry. When a read PB attempt failed, but analysis of C1/C2 generated erasures

would indicate that the read could have been successful, then a De-shufﬂe Buffer read retry may

be advisable as a recovery method. Here, an attempt is made to retrieve the PB data once more

from the De-shufﬂe Buffer (instead of off-tape), and C3-decode it, with the expectation that this

time it may pass to the MSI error-free (or without C3 generated erasures or PB checksum

failure). The ratio of such successful attempts to all recovery attempts may be an important De-

Shufﬂe Buffer reliability indicator.

5. C1/C2 Generated Erasures. C1/C2 generated erasures are erasure ﬂagged Reed-Solomon code

symbols that failed the C1 EDAC and also the C2 EDAC. The determination of C1/C2 generated

erasures is done separately for each of the two product code arrays, by Logical AND in the

hardware and multiplication by software of each array’s C1 and C2 generated erasures and

erasure counts respectively.

6. Head-related Error Counts. These counts are accumulated from the same source as those used

for transport setup. They are indicative of the random error rates per head, and are calculated for

each track as:

C1 Corrections + (C1 Generated Erasures • Average C1 Corrections per C1 codeword)