I had to document this from scratch, as most of the stuff I
had kept was large spiels about the double-sending problem.
I am now making this available for FREQ too.  I suppose that
all of these problems can be written off to operator error
or bad phone lines, just like all the Courier problems...



Written 1996-02-21 by Paul Edwards

As a long-time owner of a Spirit II, then a Spirit Thunder,
using a multitude of ROMs, I am documenting the following
problems I experienced over the last 1-2 years of use.

1. When connected to a Rockwell-based modem (using the full
Rockwell solution) at V42, the Spirit sends every block twice,
the MR light flashing like crazy.  Pausing the computer until
the buffers are flushed makes the problem go away.  The
problem also happened with a Maestro SE 9600, but it was not
reproducible with other people's systems for unknown reasons
(presumably because of the time-sensitivity).

2. When connected to a Rockwell-based modem at MNP4, the modem
initially works fine, but after about 5-10 minutes into the
transfer, it starts to double-send.  Pausing also fixes the
problem.  For unknown reasons, I was never able to get someone
else to reproduce this problem (as far as I know, only one
person ever bothered to try it).  It was pretty readily
reproducible here, 90% or something success (of seeing the
double-sending in 5-10 minutes).

3. The MNP4 implementation was very slow compared to other
modems, and to their V42.  The Spirit Thunder fixed this up.

4. The Spirit Thunder sometimes gets a "CONNECT 19200" from 
another Spirit Thunder (and possibly other brand modems,
can't remember), when I have V32terbo specifically DISABLED
via AT*N6.

5. The reason for the V32terbo being disabled, was because
Rockwell based modems would fail to connect to the Spirit
Thunder most of the time, unless they forced MNP4 or forced
V32bis or something (can't remember the change required on
the Rockwell side to work-around the problem).  This was not
a problem on the Spirit II.  Disabling V32terbo on the 
Spirit Thunder did not make the modem as good as a Spirit II
though, there were still failed connects.

6. There are a lot of failed connects from Rockwell and USR
Sportsters to the Spirit Thunder, somewhere in the order of 10%
from one Sportster (but other Sportsters had virtually no 
problems whatsoever, so there was probably a line-noise factor
involved there).  There was never any workaround for this
problem found.  A typical failed call would have the Spirit
connecting at 14400 no EC, and the other modem also getting
a non-EC connect, but garbage being spewed out both ends.
Others who forced an EC-or-bust connect, would get a "NO
CARRIER" instead.  It made no difference to the caller whether
they set MNP4, or V42bis, or no-EC.

7. It is undocumented what the "V42 with/without phase detection"
actually means (they use the words in the manual, but don't
define it in the glossary).

8. Sometimes after a failed call, the Spirit doesn't realise that
the other guy has hung up.  It sits there spewing random garbage
at the computer.  I needed to set my software to drop DTR after
a few minutes to hang up the modem.  At one point this random
garbage was so voluminous that it crashed my beta version of
OS/2, and/or would beep the speaker due to buffer overflow (I
think).  I actually put sticky-tape all over the speaker to stop
getting woken up by inconsiderate callers.

9. When sending a file to a Microcom Deskporte Fast 28.8 modem,
the V42bis appears to be corrupting the data, causing the
file transfer protocol to ask for resends (forever).  Doing a
V42 connect works fine.  Initially I thought it was a very
wierd problem with the comms software, but replacing the Spirit
with a USR Courier made the problem go away (and the V42 vs
V42bis proved it wasn't software too).  Other modems (ie non
Microcom) did not have this problem.

10. There were a couple of other problems, e.g. the defaults and
documentation of S9 + S10, which are documented in Trev
Roydhouse's "Spirit Fact Sheet" document, which I won't repeat
here, mainly because I didn't discover them first-hand or they
never concerned me etc.
@EOT:

---