Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!usc!apple!portal!cup.portal.com!cec
From: cec@cup.portal.com (Cerafin E Castillo)
Newsgroups: comp.dcom.modems
Subject: Re: New Modems, Telebit Fails under impairments in PC Magazine
Message-ID: <36349@cup.portal.com>
Date: 29 Nov 90 23:39:16 GMT
References: <1715@chinacat.Unicom.COM>
  <1990Nov19.191023.11581@nstar><1990Nov21.221114.11850@unixland.uucp><1990N
 ov23.185029.2663@nstar.rn.com><1990Nov26.010656.20883@virtech.uucp><1990Nov
 26.024835.13024@ux1.cso.uiuc.edu>
  <BOB.90Nov28170251@volitans.MorningSt
Distribution: na
Organization: The Portal System (TM)
Lines: 129


OK, the inside story...

This was my personal experience in regards to the PC Magazine
review of TELEBIT's T2500 modem.  I have expressed my own personal
opinions herein.  I am not presenting this as an official TELEBIT
representative, but only as a TELEBIT modem and UNIX user.

I assisted in the Telebit side of aid to these PC Magazine technicians
conducting the review.  I should make it clear that PCMag was very
thoughtful and thorough in contacting Telebit for assistance in 
conducting a fair review.  Those of you who saw the MacUser High-Speed
Modems Review, this year, know of the bad review Telebit modems
received, as well.  Telebit was never contacted for this review and
the evidence was the satellite transmission results for V.32 modems,
PEP was never tested.  When PCMag contacted us at Telebit, we where
very surprised and happy to get a chance at a 'good' review.

Now, the reality.  In so far as those of us techies involved in
assisting the reviewers, they were conducting legitimate tests
and doing so properly.  Telebit Systems Engineers and a couple of
Tech Support Reps. were involved in assisting with the review
over the phone.  No on-site assistance was ever given to PCMag.
This may have been Telebit's biggest mistake.  The technical people,
along with Marketing Communications, at Telebit, were all disappointed
with the results, as well.

Here's what leads to this disappointment.  The V.32 tests, and only the
V.32 tests were quite accurate and administered properly, as per Telebits
review of PCMag test procedures.  Telebit has just announced the T1600
modem which addresses some of the missing criteria used in the PCMag
review.  The T1600 has a 38.4 kbps serial interface which Telebit is
promoting as "TRUE" 38.4 kbps.  I don't quite know what they mean by
this, yet.  The T1600 is using Telebit's own DSP for V.32 instead of
the Rockwell Chipset used by the T2500 and T1500.  The Rockwell chipset
proved to be a quality assurance nightmare, while I was at Telebit.
It's default settings often caused unstable connections.  While Telebit
was able to compensate for these problems in their firmware (GE6.01)
the inability of the Rockwell chipset to do V.32bis (14.4 kbps), in
its present form, caused Telebit to develope the T1600 DSP.  Cost was
obviously an issue as well.  The T1600 is listed at $795 US.  The 
T1600 can not do PEP, at the moment, and may not ever.

Speaking of PEP...

PEP's failure in the impaired line tests was due to the lack of use, by
PCMag, of the hidden registers in the modem firmware (S120/J6S36).  These
registers affect packetization for better survival of connection and
throughput over adverse line conditions ONLY.  It was not in Telebits
best interest to start talking about or publishing this kind of information.
Only recently has Telebit Tech Support introduced a document detailing
the use of these registers.  Thus, in default PEP packetizing mode, the
failure of PEP was legitimate.

PEP also manifested the drawback many of us Telebit modem users have
encountered, the "jerky" response of PEP's half-duplex (aka Adaptive
Duplex) modulation.  This is the half-screen redraw or delay in
character echo experienced over interactive sessions.  I disagree with
the gentleman who stated that PEP was only an issue in "simple typing"
sessions and not with dial-up IP connections.  I have used SLIP/CSLIP/PPP
extensively, and not even the Van Jacobson Header Compression Algorithms
have made the use of PEP any more efficient.  V.32 is still preferred.
It's no wonder that the TELEBIT NetBlazer is meant to be used with the
T2500/T1500/T1600 V.32 modems exclusively, and not necessarily with PEP.
All my testing and that of the NetBlazer project concluded that this was
the case, even in the light of IP Inverse Multiplexing used by the
NetBlazer.

RTFM, in particular TFM, is also a failure point.  I never read through
all of the over 200 pages in the TELEBIT Commands Reference Manual.  NFW!
As the UNIX Communications Specialist, my charge was to write system
specific configuration guides for UNIX systems and applications.  This was
because of the fact that the ability to configure your Telebit modem
hinged on your ability to deal with your O/S or data comm software.  Telebit
does not have a "Plug & Play" modem, yet.  I don't believe that any
modem is ever fully understood by its user and installed with such 
understanding of both modem and system.  This is part of the market that
Telebit and most modem manufacturers try to address for greater sales and
market share.  This is also the readers that PCMag is trying to educate
or, at least, bring up to speed.  "Plug & Play" means money.  The T1600,
in its planning stages, was meant to have "user macros" which would 
auto-configure the modem for the users specific system or application.
"AT&0-9" or use of the modems front panel lights would supposedly make
modem set-up fool-proof and fast, thus "Plug & Play".  Unfortunately,
my experience with this idea, at least for UNIX systems, was that Tech
Support came up with two "optimum" UNIX initialization lines to be used
as macros.  No one in Telebit Tech Support is a Sys Admin level UNIX user.
Thus, the macro proposal I saw was nowhere near accurate.  I tried to 
convince the developers to use a VERY generic, semi-"fool-proof" init
macro for UNIX systems and SLIP/CSLIP/PPP applications.  The objective
should be to get the modem to accomodate the O/S functions of the system,
as opposed to doing "optimum" performance, which is different from
system to system.  Just remember, of the over 75 Telebit modem registers,
only a dozen or so are necessary for basic RS-232 and PSTN configuration.

And Tech Support...

Great bunch of guys!  Highly professional and knowledgeable, except for
UNIX!  This is my biggest beef.  There is no one at the Tech Support level
that I can qualify as being UNIX experienced enough to handle the complex
O/S issues involved with UUCP or dial-up IP configuration.  There is
basic knowledge and documentation, but any "stickies" might be met with
a shoulder shrug.  E-mail support, from Telebit, is all but abandoned.
Tech Support handles this, now.  If they can't get an automatic return
path from their mail system, you will not get an answer to your e-mail.
Forget postings or responses to netnews, this part of the net support
is not even being monitored.  I am biased here because of the fact that
I was laid-off in September '90 by Telebit due to their lack of "making
their numbers" in that quarter and bad forecasts for the remainder of the
year.  I was the guy in charge of e-mail and netnews support.  Luckily,
my new employer, a TELEBIT distributor, allows me to continue my netnews
activity in support of TELEBIT modem and NetBlazer users.  This is
fortunate for me because I consider my self part of this community, as
opposed to just an industry rep..


I hope that this info is of help to understanding the issue at hand.

===============================================================================
Cerafin E. Castillo                       ||      //\\  ||\\  ||
Network Consultant                        ||     //__\\ || \\ ||  Los Altos
Los Altos Networks                        ||    // ---\\||  \\||  Networks
340 Second St. #6                         ||___//      \ |   \ |
Los Altos, CA  94022
(415) 941-8031      UUCP:     {apple,sun,uunet}!portal!cup.portal.com!cec
                INTERNET:     cec@cup.portal.com

                      "...No hay mal que por bien no venga..."
===============================================================================