At 10:49 PM 1/27/99 +0000, you wrote:
Bill Pechter <pechter(a)monmouth.com> wrote about
1/2 mag tapes and drives:
Tape Density Encoding method
DEC Tape Drive
200BPI (NRZ?)
556BPI (NRZI?) TU10?
200 and 556 were only used for 7-track. Even from a single vendor such
as IBM, some 7-track systems used odd parity and others used even. I'm
not sure what recording mode the even parity systems used, as NRZI wouldn't
work with even parity unless you could otherwise guarantee that there are
no all-zeros data conditions. See below.
800BPI NRZI (Non-Return to Zero Indescrete)
TU10/16/TE10/16
800 BPI was used for 7-track and 9-track. NRZI is Non Return to Zero
Inverting. On a 9-track, each "line" on the tape consists of 8 data bits and
an parity bit. The "I" (Inverting) means that a flux transition is recorded
on the tape if a data bit in the line is different than the same bit position
in the previous line. So unless the data is all zeros, each line will have at
least one flux transition. Unlike DECtape, there is no "clock" track, so it
is essential that there is always at least one track with a flux transition,
from which a read clock can be recovered. The odd parity guarantees that if
all 8 data bits are zero, the parity bit will be one, so there is still one
flux transition.
It was conventional on 7 track drives to use the modes "BCD" and
"Binary"
to describe the parity of the written characters. In BCD mode, each character
was written in even parity, while "binary" mode wrote odd parity. One
interesting side effect of this was the character code for '0' (usually six
bits of 0) couldn't be represented, since. as described above, it would have
been encoded as no flux changes for that field - i.e. a "blank" spot which
would have been completely skipped by the read electronics. For this reason,
some other character (often an octal 12 - often ':' in BCD) was chosen as the
replacement character, so in BCD mode, often the ':' couldn't be represented.
This code chosen varied from manufacturer to manufacturer.
In "binary" mode with odd parity, all 6 bit codes could be represented since
there would never be a field of all 0's. I've never found out why there
needed to be both of these modes rather than just use binary mode for
everything. Anyone know the reason for this?
I should point out that my experience on this was on Control Data 3300
equipment (and some early CDC 6000 equipment with 7 track drives as well.)
We also used a number of Kennedy (sp?) small stepper-motor 7 track 200BPI
drives and, as far as I can tell (I have an old 7 track drive I use for
reading legacy tapes) all densities 200, 556 and 800 BPI use NRZI. In fact
for my small drive, setting "read" density is a No-op. It only affects
the clock generator for writing. So I assume the only difference is how
tight the bits are written to the tape.
Gary