LE>> particularly impressive. Actually there was an interesting
LE>> discussion here (or zone3_tech?) about the fact that RS232
LE>> is NRZ, so therefore doesn't have a "baud rate" per se
LE>> possibly.

IS> RS232 is Non Return to Zero?  I don't believe so.  As far as I know, it's
IS> straight 0 (start bit), data bits as they come, then 1 (stop bit), per
IS> byte.

IS> NRZ (or NRZI) is usually used on self-clocking data streams, is it not?
IS> RS232 as we use it is referenced to fixed (local) Rx/Tx
 IS> clocks, clocked at mid-bit.

 DH> The data stream doesn't change from bit to bit, if the data
 DH> value doesn't. There is no embedded clock transition.

Right.  Quite so.  Ooops.  Dunno what I was thinking of there.

 DH> NRZI is similar, with the data being toggled for a 1 bit,
 DH> and left static for zero.  For that one, a constant stream
 DH> of zeros gives constant polarity of whichever was showing
 DH> when the constant stream started.  Still no embedded
 DH> clocks.  The spectrum had some advantages on lines that
 DH> could go quiescent in mark state, which is where it came
 DH> from.

Indeed, giving average DC levels much closer to zero too, Tanenbaum and
Black tell me.  I should have hit the book first ..

 DH> Embedded clocks don't show up until you get to bit
 DH> stuffing, or go to FM or MFM encoding.

 DH> NRZ is (N)on (R)eturn to (Z)ero encoding.  The reason for
 DH> the start and stop bits is that when the format was
 DH> designed, 8 data bits was just about the limit for accuracy
 DH> of the local sampling clock.  (It was mechanical..:-)

:)  Indeed, remembering setting up a 50 (or was it 45?) Baud Teletype at
one stage, my first printer.  5 data bits.  Drivers were =real= drivers in
those days (current loop circuit off a single output bit, UART done in
software :)

I also recall we worked out how to pulse dial using the right 300bps
pattern, for a modem design using the AMD7910 critter, about 10 years ago. 
We thought ourselves sooo clever .. can't quite remember how it was done
now, but 300bps gave just the right mark/space ratio for it, after
stripping start/stop bits.

Poor old 'baud', eh?

Cheers, Ian

--- MaltEd 1.0.b5