[net.dcom] 2400 bps modems can be non-standard

bob@islenet.UUCP (Robert P. Cunningham) (05/09/85)

[These are my notes on some problems with U.S.-made 2400 bps modems,
using information from a variety of different articles and discussions
with various vendors.  Clarifications and corrections welcome.]

2400 bps modems.  Twice the throughput of 1200 bps modems for less than
twice the price.  It sounds good, but many of the new 2400 bps modems
now on the market in the U.S. are not completely compatible with similar
models from other manufacturers.

There is no U.S. asynchronous dial-up 2400 bps standard in the same sense
that the Bell 212 modem set the standard for 1200 asynchronous modems.

There are two European standards: "CCITT V.22 bis" and "CCIT V.26 ter".

These are written standards, while the Bell 212 was a complete
working product, with very well known operating characteristics.  A
minor difference in principle, but a tremendous difference in practice.

Not only are there loopholes in the CCITT standards that give each
manufacturer considerable room to be creatively different, but there are
some modifications that U.S. manufactures tend to make in
order to maintain some compatiblity with existing U.S. equipment.

The result is that many of the 2400 async dial-up modems are
incompatible with each other in various ways.

Most of the new U.S. made 2400 bps async dial-up modems
follow the V.22 bis standard.  They transmit and receive
simultaneously by splitting the available bandwidth in half, using half
to receive and the other to transmit, with a 16 point Quadrature
Amplitude Modulation (QAM) technique at 600 baud [precisely speaking,
the baud rate is the rate of change of the signal ... QAM and most other
techniques provide a way of encoding several bits into each change of
the signal].

A few use an alternate CCITT specification -- V.26 ter.  
Signal cancelling (the receiver cancelling out the echo of its own
transmitted signal) allows the whole bandwith of a phone line to be
used.  V.26 ter uses a Differential Phase Shift Keying technique to
handle 2400 bps at 1200 baud.

V.22 2400 bps is probably less reliable than 212-type 1200 bps over
long-distance lines, V.26 ter is probably better than 212.

Of course V.22 bis and V.26 ter are completely incompatible.

While V.22 bis seems to be the preferred standard now, there is a good
chance it may eventually be superseeded by V.26 ter.  [Then again, maybe not;
the Vadic 1200 bps technique is -- in some ways -- more effective than 212,
but it's never really caught on.]

Now, about those loopholes in V.22 bis ...

The standard designates a fall-back speed, if the originate and answer
modems can't handle a 2400 bps connection.  However, the standard
doesn't specify how the connected DTE equipment (computer or terminal)
is to be notified of the fall-back.  The RS232C standard doesn't cover
it.

With V.22 bis, each U.S. manufacturer seems to have chosen a DIFFERENT
way of indicating a speed change when the modem falls back, using
various of the seldom-used secondary control pins on the RS232C
connector.

Chances are that the typical DTE device you hook up your 2400 bps modem
to will ignore the speed change signal.  Then, when you obtain a dial-up
connection that's a bit noisy, the modem falls back.  It sets up and
maintains the connection nicely, but not at the baud rate your computer
or terminal expects.  This can tie up the equipment at each end
indefinitely.

V.22 bis specifies the CCITT V.22 format for 1200 bps fallback.
Unfortunately, that's incompatible with Bell 212.  To allow V.22 bis
modems to be used together with regular 212 modems, many (but not all)
of the U.S.  manufacturers have chosen to make 212 rather than V.22
the fallback.

As a convenience, some (but, again not all) of the U.S. makers who
provide 212-type 1200 bps fallback also provide a further 103-type 300
bps fallback from 1200 bps.  Nice feature, but definitely not in V.22 bis.

There's still another common "Americanization" that U.S. manufacturers have
adopted.  V.22 bis assumes that the European standard 2,100 Hz answer
tone be sent by the answering modem during initial connection
handshaking.  Many U.S. manufacturers have instead adopted the regular
U.S. 2,225 Hz answer tone -- again for 212-type compatibility.

Unfortunately, this means that many U.S.-made V.22 bis modems
won't handshake at all with a European V.22 bis modems.

V.22 bis specifies V.25 (or V.25 bis) autodialing.  U.S. makers
prefer their own variation of the Hayes autodialing commands (or the
Concord technique, or the AT&T technique, or Cermatek ... there's
definitely no effective U.S. standard for autodialing commands).


Summary & recommendations:

If you want a 2400 modem that will talk to European-made modems
make sure it uses the 2.1 kHz answering tone, and has V.22 (not 212)
fallback.  Find out whether the other end uses V.22 bis or V.26 ter.

If you need point-to-point 2400 bps dialup in the U.S., choose your
favorite manufacturer, but you'll have more consistent results of you
have the same model from the same company at the other end.  Otherwise,
don't be surprised when your modem "hangs".  In any case, may expect
to see more "phone line hits" -- especially over long-distance lines --
than you get with your 1200 bps modem.

-- 
Bob Cunningham   ..{dual,ihnp4,vortex}!islenet!bob
Honolulu, Hawaii