[net.periphs] Hispeed modems reposting

dya@unc-c.UUCP (04/18/84)

     The article has long since been trashed ( have  only  a
160  mb Fujitsu ) but I'll comment on the modems again, any-
way.

     Universal Data Systems 9600 A/B dial-through  modem  is
absolutely  WONDERFUL  !!!!!  You  can operate it on dial-up
phone lines without conditioning ( except where there is # 1
ESS  and line extenders, which requires type C2 conditioning
); two wire leased line conditioned or not -- up to  100,000
feet w/o load coils with no amplification and who knows with
amplification, or four wire leased service.

     The UDS 9600 A/B features automatic digital adaptive EQ
which , if given any signal at all, can compensate for fair-
ly bad envelope delay distortion.   Receive  carrier  levels
can  be set at -34 or -43 dBm; transmit can be set at -12 to
0 dBm. An optional high frequency pre-equaliser ( also digi-
tal ) can be manually selected.

     In two-wire, half duplex operations, the UDS  9600  A/B
trains  on data automatically,  2400 times per second. ( The
manual says 9600 times per second, but the  signalling  rate
is  only  2400 bps ).  Delays for turnaround can be set from
15 to 150 milliseconds depending on  the  system  configura-
tion.  In four wire operations, the UDS modem can be set for
train on data or signal quality retrain ( usually, the  car-
rier is constant in this mode ). It is completely compatible
with CCITT rule v.29 for high speed  9600  baud  modems  and
their constellation.

      The 9600 A/B is fully synchronous, meaning that if you
want  to  use  asynchronous  devices,  you  must purchase an
async-to-sync converter. Cheap async-to-syncs cost  about  $
300;  the  one  that UDS offers won't do 7200 baud ( but the
modem will ). UDS has a new one, which offers an ECC correc-
tion  scheme that supposedly lowers the error rate to one in
several years.  The MC  6850  ACIA  works  fine  with  these
modems  directly  ( use the modems tx and rx clocks to drive
the ACIA.)

      Fallback to 7200 or 4800 baud  is  possible  with  the
back  panel switch; you can also select fallback through the
DB-25 connector using a logic level. Also, it is possible to
pulse  dial the modem via your own software through the same
connector.

      Self-test facilities are extensive; with  a  piece  of
wire  jumping  DTR  to the right level; all you need are two
modems to evaluate the performance of a given phone link. In
four  wire  operations,  digital and analog loopback through
the distant modem can be selected remotely  from  the  local
modem.

      Users quickly  begin  to  correlate  the  front  panel
lights  with  line conditions ( in our application, they are
used for dial up service for medical imaging transmission ).

Experience: We have purchased about 35 of  these  modems  so
far  and have found only one place where they would not work
( yet ); suprisingly, it is a local call within the same ex-
change  (  662  ) in Buffalo, NY.  If you call long distance
they work; if you call long distance and back in, they work.
But not locally.... This is because the compandor within the
ESS mutilates the training burst sent whenever  RTS  is  as-
serted.

     Currently, I have two on my desk  which  talk  to  each
other  just  fine, a second pair is connected between my of-
fice and our PDP-11 which is over 50000 feet away (* in four
wire ). Typically,  I will see garbage about once per day on
the screen from line hits. If the lines were conditioned for
audio  this probably would never have happened ( my boss had
delusions of grandeur about using  RS  422  over  this  dis-
tance.)

     General DataComm 's 4800 bps modem  is  worthless  over
the same link , by the way.

     I got these modems to work over 100000 feet incidental-
ly.  With  conditioning ( and amplification ) they work just
as well in this room as they do from here to Los Angeles...

David ( "last of the Analog" ) Anthony
   ( decvax,akgua ! mcnc ! urp ! dya )