dhb (04/21/83)
NRZI (Non Return to Zero, I forget what the I stands for) and
phase-encoded are two methods of doing the physical recording
on the tape. The difference has something to do with zero bits
but the complete definition escapes me at the moment (it's been
a while since I concerned myself with tape drive hardware
questions). There is also another method called GCR (Group
Coded Recording) which is frequently used for 6250 BPI.
Anyway, I suggest you get ALL the hardware manuals if you want
to attempt to understand the difference between them. It also
helps if you are a practicing electrical engineer.
Dave Brierley, {decvax!brunix, foxvax1!brunix, allegra}!rayssd!dhbsmh (04/22/83)
I seem to remember that the I in NRZI stands for Non-Return to Zero Inverted, or some such thing. The difference is in whether a zero or one causes a transition, but I for get which.
johnl (04/24/83)
#R:sii:-27000:ima:18600001:000:1821
ima!johnl Apr 23 15:49:00 1983
The difference among the various tape formats is not all that complex.
First, the useful info:
7 track drives: 200,556,800 BPI are all NRZI
9 track drives: 800 BPI is NRZI
1600 BPI is PE
6250 BPI is GCR
So if you need to read a PE tape, it just means that you need a 1600 BPI
drive of which there are many.
Slightly more technically, NRZI (Non Return to Zero IBM, if you work for IBM,
Non Return to Zero Inverting, otherwise) is the traditional tape format.
The seven or nine bits of each tape frame are recorded in parallel tracks.
Whenever there is a 1 bit the magnetic flux in the relevant track is flipped,
when there is a 0 the flux stays the same. Since characters are always odd
parity (except on some ancient 7 track tapes) there is always at least one
1 bit per frame, so the drive can find the frames when reading pretty easily.
The problem is an analog rather than digital one; the frequency of the data
in a given track can range from 0 (all zero bits) to 800 flips/inch (all
ones) and building amplifiers with wide and uniform frequency response is
hard.
PE stands for phase encoding. The idea here is that there is always at least
one flip per frame in each track, and sometimes more. This means that the
number of flips/inch now ranges from 1600 to 3200, which is a one-octave
range, much easier to deal with and so more reliable at higher densities.
At 6250, you can't count on not having errors, so you record a whole
group of characters along with extra error correction stuff, giving you
Group Character Recording.
I'm a little rusty on this, but the essentials are right. Unless you
manufacture or repair tape drives, the details are not likely to be
terribly important anyway.
John Levine, decvax!yale-co!jrl, ucbvax!cbosgd!ima!johnl,
{research|alice|rabbit|floyd|amd70}!ima!johnl