ghg@pur-ee.UUCP (George Goble) (12/24/87)
I bought a pair of Telebit Trailblazer Plus modems on my VISA card
back in October for home use. THEY WORK GREAT!
Other useful tidbits, etc..
The "Plus" model, has hardware support for Data compression and
is currently being shipped with V3.00 firmware. The V3.00
firmware is on two 27512 EPROMS and they are almost full according
to Telebit folks. The firmware is around 120,000 lines of "C".
The PROM sockets will accept bigger EPROMS.
The "plus" model has a very much improved analog front end over
the "plain" model (pre-October).
Earlier this fall I tested a pair of the "plain" Trailblazers
(remarketed as GTE Trailblazers). They (V2.00 fimrware I think)
worked very well, but did not have all that fancy protocol support
that they have now. These modems had a phone cord which had
an RF choke (coil) about 1.5" diamater and .5" thick and a warning
(from GTE) about interference and not complying with FCC
regs if this choke was not used. They weren't kidding, even with
the choke inplace, the modem emitted wideband RF noise pretty much
all across the shortwave spectrum (2-30Mhz) and pegged out the
S-meter on a shortwave receiver about 15 feet away (ICOM-R71A).
Although not bothersome to me, these early modems had a "noisey"
fan.
Indiana Universities maintain a voice phone network (SUVON),
State Universities VOice Network, comprised of mostly poor
quality throw away grade phone circuits to communicate with
other state universities. The level loss is so great and the
circuit so noisy, that voice communication is often almost
impossible at best. We are also *GTE* territory, although
recently converted to ESS.
Tests of the early GTE trailblazer were still impressive.
Local (both on campus and city-to-campus) connections ran
at 16-18Kbaud. A double SUVON connection (Purdue->INDY->Purdue)
would not connect, although the remote modem could be heard faintly
in the speaker. Regular 1200 baud Vadic modems sometimes have
problems with only a single SUVON (Purdue <-> Indy, 60 MI)
connection, but usually work well enough for a USENET feed.
For interactive use, the 240 millisec round trip echo time
was a little bothersome, but no worse than running on
a loaded network, but STILL PREFERABLE to 1200 baud anytime!
Enter the Trailblazer Pluses. The noisey fan is gone.
The choke in the phone cord is gone. The same shortwave
receiver now can barely tell the modem is turned on at some
frequenices, and there is no S-meter deflection from it.
Visual inspection shows all leads bypassed with caps and
ferrite beads, good job.
The Pluses connect EVERY TIME on SUVON, at 15-17Kbaud, shows
what the new analog front end did.
Took one Plus down to Fort Lauderdale right after their October
hurricane, phone lines were still wet and noisey down there.
From a Hotel room, I got 13Kbaud connections back to West
Lafayette, Indiana. All of this was error free of course.
My boss tested the plus modems from a small outlying town
where the phone company is worse than GTE (old crossbar still).
They worked (15-17Kbaud). Various cheepie 2400baud modems are
unusable from his house.
All the baud rates for the Trailblazer Plus modems are with
compression turned off, so line quality could be compared.
With compression enabled, any connection better than 10Kbaud
got turned into 19.2KBaud (plain text) .. solid. For average
"text" the compression seems to be good for about 50%
I was able to get connections as low as 1000-2000 baud by
blowing in the microphone during the protocol initialization.
Although jerky, these modems still worked under that abuse.
As another (not very useful) test of the compression,
I sent a file of the same character repeated over a connection
established at 2000 baud (abused the initialization), and it
came out at a solid 19.2K (roughly 10X compression)
The GTE ESS switch runs diags at 3-4AM. These appear to cause
short dropouts, which would disconnect my Vadic 3451 (1200
baud error-prone) modem. Telebit is not affected. One can also
pick up the phone and Touch Tone dial some numbers and
the Telebit only slows down a little. Worse abuse like holding
down a key on the phone will force a modem retrain (around
an 5-8 sec pause), but will not glitch or drop the connection
I talked to somebody with a USR 9600 modem and merely picking
up the phone made the connection drop instantly.
I am a BETA test site for V4.00 Firmware. This introduces
"micro-packets". The "mushiness" for interactive use
almost disappears. The round-trip interactive echo time
was reduced from 240 millisec to around 100-120 millisec,
now almost not noticable! Telebit will sell this firmware
during the later part of Q1 I understand.
The V3.00 and later modems can set the "PEP" tones to come up
last during a connection sequence, so other non-Telebit modems
can still connect at 300/1200/2400 baud and up to MNP level 3.
The above baud rates (other than 19.2Kb) were obtained by reading
S-registers S-70 and S-72 from the modem. The "solid 19.2Kb"
was from the visual inspection of a Lear ADM-3a terminal doing
full speed output will no observable pauses. There were no
accurate and detailed measurements made. The Telebit manual
states that S70 and S72 registers show the effective
bit rate of the transmission, but the actual thruput is lower
due to protocol overhead, etc. The compression appears to
make up for this in my use of the modem. I bet this modem could
do 25-26Kbaud on a good line with compression, if only the
interface could run 38.4Kbaud.....
I plan to attempt to interface the Telebit to a cellular phone
to operate a battery powered terminal in a briefcase. (Zenith Z-181).
George Goble
Electrical Engineering
Purdue Univ
W. Lafayette, IN 47907
ghg@purdue.edu
{ucbcax,pur-ee}!ghg
--ghg