Voice
Communication in Business Volume 2
Essays on telecommunications,
1981-2002
Why
should small PBXs be inferior to their larger brothers? As the cost
of processors and memory went down, it became harder and harder to
justify omitting necessary features from small PBXs, particularly
when the program for a larger brother was already written. This
appeared in the September-October, 1986, issue of Business
Communications Review.
Small PBXs - Equal
Opportunity For The Small Office?
(Business Communications Review, 1986)
Office modernization is
making giant strides. Celluloid collars with gold collar buttons are
gone, as are most quill pens, dipped into inkwells, for making
records in ledger books. We have made a good start toward getting
rid of green eye-shades and sleeve garters. We have even managed to
replace at least some messenger boys with telephones, and some
functions of the postal service with electronic messages. But there
is still a barrier. If we work in a small company, or a small branch
of a large company, many of the ivory tower types who design
telephone equipment assume our needs are fundamentally different
from those of our associates in larger organizations, and they
refuse to offer us the opportunity to use the telephone for data as
well as voice.
However, this is
something of an improvement over the past ten years. In 1975, PBX
manufacturers at trade shows hooted at the idea of putting non-modem
data through a digital PBX, large or small. Now, most of them just
hoot at the idea of putting data through a small PBX. They continue
to push small 2-wire analog space-division systems, and even to
suggest such things behind big digital PBX. And naturally they use
them with Centrex, most of which is now and will continue to be
2-wire analog space division for many years to come.
But there are some
alternatives beginning to show up. Most of the major manufacturers
have large digital PBXs up and working. They have developed
electronic telephone sets and elaborate software for these machines.
In general, they are looking for new worlds to conquer, using
weapons already available whose development is already paid for. And
PBXs for the small office are a dandy microcosm to attack. "Digital
PBXs for the small office" seems to be the battle cry.
The sensitive reader has
probably already detected a note of cynicism in my feelings about
the current batch of glossy brochures on my desk. And, indeed, I am
disappointed but not surprised. After all, I once actually worked
for a living and, as a result, have had some first-hand experience
with Research and Development (R&D). That experience, coupled with
years of observation, leads to the following thoughts that apply to
small digital PBXs here and now:
First, in any R&D
environment, we always find that whatever the project chief wants to
do can be done in no time for no cost and will be a eagerly embraced
by the clamoring multitudes. Any other approach, however, will be
too expensive, will take too long, and will most surely be something
nobody wants.
Second, market research
is always done to verify the desires of the abovementioned project
chief. You don't really think for one moment that a lot of money is
going to be spent to prove him wrong, do you? After all, he (there
aren't many shes in such positions as yet, and if there were, I
doubt if they would make some of the mistakes we see all around us)
is the expert. What does the customer know? Just take the answers
that support the thesis. What does it matter if customers get what
they don't want, don't get what they do want, and somebody else gets
the business? By then the project chief will have been promoted, on
the basis of his sterling performance, to a higher position in
another department.
Third, the current
prosperity of modem manufacturers indicates that SOMEBODY wants to
send data through the telephone network. Many of these somebodies
are the smallest of small offices: individuals. When customers vote
with their bucks, they usually have something in mind; just what it
actually is may not show up on a market research questionnaire
designed to elicit some other response. My personal feeling is that
the customer should be offered the maximum flexibility in a new
product, and the R&D types should stand back and watch what he does
with it. They might be surprised.
Finally, the major
reason from the customer's standpoint for having a digital PBX is to
handle voice and non-calls with equal ease. There is nothing that a
digital PBX can do for voice-only connections that an analog PBX
cannot do equally well. It is much harder to make a big electronic
PBX using analog rather than digital switching techniques, but small
analog PBXs are easy technically and usually less expensive than
digital.
Voices From The Past
Glancing momentarily
backward, we discover that the two most successful electronic PBXs,
AT&T's Dimension and Mitel's SX-200 and -100, sold a total of over
60,000 units; they made their fortunes by not being digital. Why
spend a lot of money to code a signal into a digital format to move
it 18 inches from one matrix port to another? Next, consider the two
biggest selling digital PBXs from 1975 to about 1981: Rolm and
Harris's DTS 1200. Neither one of them was or is directly compatible
with T-carrier, the most common digital transmission medium
available from one PBX to another. Why digital? Indeed, Rolm's
current advertising and PR releases are making a considerable effort
to assure the customer that CBX digital techniques are completely
invisible, and their incompatibility with T-carrier and ISDN need be
of no concern.
There were a few
companies, back in 1975, notably Northern Telecom, General
Telephone, and Stromberg Carlson (now Telexicom), who recognized
that internal and external digital capabilities might one day fit
together, and started off in the right direction. And almost
everybody, when they got a second chance, went digital with
T-carrier compatibility: AT&T went from Dimension to Systems 75 and
85, Mitel finally brought the SX-2000 to market and used some of its
capabilities to expand and augment the SX-200, Danray people founded
InteCom, Harris now has the 20-20 to replace the 1200, and the NEAX
12, a 2-wire analog switch, as well as the NEAX 22, a digital switch
with 2-wire analog space-division line concentrators, have been
replaced with the alldigital NEAX 2400. Although the Ericsson
Prodigy and the Telexicom Lexar, as delta-mod systems, are still
with us, both companies have T-compatible digital systems available
as well. And of course, all the new systems coming on the market
today (InteCom, Telenova, Tadiran, SRX, Redcom, Solid State Systems,
ITT 3100, etc.) are T-compatible Digital. Only Rolm/IBM continues
its stately way with its own private CBX technology.
Three Small-Office Digital
PBXs
But what about the small
office? Are small business customers doomed forever to have only the
2-wire analog phones of which the ivory tower designers and their
pet market researches feel them worthy? Let's take a quick look at
three new, small digital PBXs coming to the market from major
manufacturers: Rolm's Redwood, AT&T's System 25, and NEC's Impulse.
They will show us the best and the worst of the ivory tower thinking
of their designers. Then we'll take a glance at some additional
small systems currently available and see what others are doing.
First, let me summarize
the results. Redwood is all-digital, including its multi-button
sets, but can't handle data. System 25 is a digital switch which
only handles analog phones, but CAN handle data as long as it is not
integrated with the analog sets. Finally, Impulse seems to have
somehow gotten almost everything right and offers digital
multi-button sets with optional data interfaces that permit handling
voice and data alike. In all three instances, standard sets from
larger PBXs are taken over directly for the small systems.
Redwood. The big
news about Redwood is, of course, its use of "8x8" coding (8000
samples per second, coded into 8 bit bytes) rather than the "12x12"
coding of the CBX series. Another improvement is the attention given
to human factors at the communication manager interface for moves,
changes, and other administration. Finally, because Redwood
emphasizes the use of ROLMphones only, 1A2 behavior is emulated
rather well; flash and feature code operation, basic to the CBX and
CBX II product line, is vigorously derided.
Exclusive use of
ROLMphones as instruments on Redwood (along with ROLMlink as the
set-PBX connection) is a great step forward: any ROLMphone from the
smallest to the one with the most buttons and displays can be
plugged into the same jack, use the same ROLMlink, and the same
line-card in the PBX cabinet. Thus everything involved with moves
and changes can be handled by software and simply plugging the
particular phone into a standard jack (no line-card or MDF jumper
changes). Further, ROLMphones need only a single pair, so presumably
existing small PBXs with single line sets can be replaced fairly
easily.
ROLMphones and ROLMlink
were designed to work on standard 8x8 coding, with rather intricate
mapping into the CBX and CBX II 12x12 format on each line card. In
Redwood, where the switching matrix uses 8x8 coding, this "digital
signal processing" can be omitted to permit considerable line-card
simplification. With the codec located at the ROLMphone itself,
voice and data can hit the ROLMlink as digital bit-streams,
multiplexed together. Indeed, ROLMlink is designed to handle 64 Kb/s
for voice, 64 Kb/s for data, and have 128 Kb/s left over for
signaling, supervision, and other things which may be required in
the future.
The important thing to
note is that the 256 Kb/s ROLMlink bit stream goes simultaneously in
both directions on a single pair, with directionality separated by
hybrid circuits at each end, and voice and data look exactly alike
at the line card. But Redwood can switch voice and cannot, as yet,
switch data. A line card supports 8 sets, and 8 time slots in the
switching matrix reserved for it. Apparently there are not enough
time slots to handle 8 more inputs for data. (Note that in the Mitel
SX-2000 technology, which IBM rejected in favor of Rolm's, each line
card also supports 8 sets, but has 32 time-slots available to allow
for just such situations).
Without data capability,
the rest of Rolm's desktop product line (Cypress, Cedar, Juniper,
etc.) cannot be used. And, just to put the cherry on top, the
digital ROLMlinks will not support standard analog modems. Install
Redwood, and abandon data completely. But that's ok. Rolm press
releases cite Dataquest as authority for the idea that less than
0.1% of the users of small systems need data capability. My guess is
that, in an existing 1A2 environment where intra-system channels
(intercom links) are few or non-existent, people of necessity walk
down the hall to talk in person. If you don't have something, you
can't use it, and market researches can easily draw the wrong
conclusions. Maybe with a PBX, which implies intra-system paths,
behavior might change. And when desktop computers replace quill pens
and foolscap, the ease with which written material can be commented
upon, edited, and put into finished form may demand intra-group data
communication. If a small PBX can't do the job, a LAN will be
required to duplicate the functions so conveniently omitted.
In all fairness, Redwood
supports 2500 sets as OPS instruments, via analog cards plugged into
analog card slots (which are different from the ROLMphone digital
card slots) intended normally for CO and tie-trunks. So there is a
way to handle some data. And Rolm people say they will be a able to
handle data soon. But how? Like voice? Apparently not with the
present line cards. Perhaps they will handle data completely
differently as in CBX and CBX II. We'll have to wait and see.
System 25. As
already pointed out, System 25 can handle only some of AT&T's new
electronic sets. Specifically, it cannot support the 74xxD digital
sets used on Systems 75 and 85, the only sets that offer an
alphanumeric display for directory, messaging, etc. But, more
important in the present context, only the 74xx sets support the
work-station interface for data.
The strategy seems to be
to support sets and wiring already in place to facilitate
replacement of the discontinued Horizon and smaller Merlin with
System 25. System 25 can support 2500 sets, 71xxA single line sets,
and 72xxH and 73xxS "hybrid" sets, similar to Horizon MET sets and
Merlin sets, respectively. Because these same sets and their wiring
are also used on Systems 75 and 85, outgrowing the System 25 again
means that investment in existing station equipment and wiring is
preserved.
Unfortunately, 2500 and
71xx sets take one kind of line card, 72xx sets take a different
card, and 73xx take yet another type. This will lead to
administrative costs in connection with inventory, moves, changes,
and upgrades in the future, to be somewhat compounded if and when
another line card is made available to support the 74xx digital
sets.
Without the 74xx sets,
how does System 25 handle data? It uses separate units from System
75 and 85 that provide data-only interfaces where telephones are not
needed (as at mainframe computer ports, etc.). Needless to say,
these ADUs (Asynchronous Data Units) require yet another type of
line-card. With voice handled via analog phones and data handled via
ADUs that are related to specific telephones by program only if at
all, AT&T's advertising experts have demonstrated considerable
creativity in claiming "voice and data integration...You can talk
and send or receive data at the same time with both voice and data
signals carried on the same wiring."
Notice: they didn't say
wires. Both AT&T and IBM have decided that the user is to install
four pairs to each telephone location, although neither has a
telephone set that needs all four. Thus with the right adapters at
the set and intricate cross-connects at the MDF, voice and its
control can use some of the pairs to System 25, while data can use
other pairs among the standard four.
What is likely to
happen, of course, is that the customer will use some of the extra
pairs at each work station to put in his own data system, buying it
from a vendor other than AT&T or IBM, probably at much smaller cost.
By all means, pull extra pairs to each possible telephone set
location; that gives you the freedom to use your new PBX for voice
and data or, if you like, to use a completely separate system
without tearing up the place to pull new wire.
Impulse. The NEC
Impulse is built to support the digital D-term sets developed for
the NEAX 2400. The same snap-in adaptor for data can replace the
Dterm base, and the line-card can interface both voice and data bit
streams to the matrix in exactly the same way, just as in big
brother. The idea is so logical one wonders why Rolm and AT&T didn't
think of it, too. Giving the customer the chance to add data if he
wants cannot have increased the cost of line card in the Impulse
very much, since voice and data are handled alike, and NEC will not
have to redesign their line card later to add data capability. Maybe
small offices actually do NOT want to switch data through their
telephone systems, but what they will actually do when the chips are
down depends on what is possible. Making small PBXs that either do
not handle data at all or handle it awkwardly is a sure way to make
a self-fulfilling prophecy about how great a LAN can be.
By and large, the NEC
Impulse comes off quite well in that it can handle voice and data
today, as can System 25, but it can integrate voice and data through
its Dterm sets and their uniform wiring. Unlike Redwood, it does not
leave 50% of its set capability unavailable to the customer. It will
support 2500 type sets, but does not have to support a variety of
other analog sets to protect its existing users at the expense of
future administrative costs.
About the only flaw that
shows up in the Impulse glossy brochure is the following paragraph:
"We're also a world leader in integrating computers and
communications (C&Ctm). We were the first to point out the
significance of this concept, and the first to bring together the
separate technologies that comprise it." A very careful definition
of "C&C" would be necessary to turn this hype into reality.
Data between PBXs.
But now, how about moving non-voice signals BETWEEN PBXs? Not one of
these machines offers a direct access circuit to a T-span. Such
cards are available for Systems 75 and 85, NEAX 2400 and Rolm CBX
II, but not for their little brothers. No CPI, DMI, or T1-D3. Just
loop start, ground start and E&M, all dating back to the days when
Edison was a boy. Maybe intra-switch data isn't needed in small
systems, but data connections to other PBXs may be one of the things
making Hayes rich. Both System 25 and Impulse talk about modem
pooling. Perhaps that is enough for now, but something better than
103/212 modems should be considered. One of the hang-ups on IBM's
laptop computers appears to be the built-in modem which can only do
103/212, and cannot provide the flexibility customers associate with
Hayes modems.
Maybe Rolm, AT&T and NEC
all plan to wait for the ISDN 23B+D standards, after they've sold
lots and lots of existing electronic sets that are not quite 2B+D.
Then they can do the PBX-to-network connection right.
Some Other Small Digital
Systems
Is intra-PBX data
transmission really needed in a small location? Well, there are some
who suggest that small LANs are going to be highly useful to tie
together a boss, secretary, and a particular work-group who have to
interchange text for comment, updating, etc. Although such a group
might speak among themselves directly far more frequently than they
would talk on the phone, they could pass documents to each other
electronically as easily as by hand, and modify them electronically
a lot easier than on paper. If this turns out to be the case, a
small PBX with more "intra" paths than a 1A2 key system could do the
job as well as a LAN, assuming favorable life-cycle costs.
Northern Telecom seems
to be following such a strategy. The SL-1S, the smallest of their
Meridian PBXs, will not support the packet transport equipment (PTE)
which the larger Meridians will use for LAN functions, but in the
100 line and down size range we find the DV-1. The DV-1 is sort of a
stand-alone PTE that provides packet switching for data and circuit
switching for voice among Northern's 4020 terminals, along with
access to applications processors, main frame computers, etc.
Although hardly a PBX, lacking such amenities as Automatic Route
Selection, the DV-1 supports 2500 type telephone sets for regular
voice users, and analog loop start and ground start trunks for use
with COs or behind PBXs. It does not, however, as yet support the
Meridian electronic sets used by the SL-1 and SL-100, and has no
digital trunk interface. Advertised as an office automation
controller, its most interesting feature is a voice conferencing
system using the displays on 4020 terminals to show who is present,
who is absent, and who has a plain old telephone vs. who has the
full display capability of the 4020 terminal including Northern's
"Share," a data feature that lets everybody see the same screen
picture and have equal opportunity to modify it in front of all in
real time.
But some companies such
as Telenova, SRC, CXC, Ztel, Redcom, Tadiran and ITT are charging
into the small end of the PBX market at a great rate. Wang, owner of
Telenova, has put out a complete office automation system using the
Telenova to hold it together, and Telenova is as likely to have a
digital CO interface as soon as the larger companies.
SRX can go down to 6
extensions economically, can emulate 1A2, and operate as a regular
PBX or some combination of PBX and key. Universal card slots are
used for analog or digital ports. Digital ports use ping-pong
transmission to and from the set on a single pair, and the same
digital line card supports three digital telephone sets as well as
several data interfaces. Data interfaces, like digital telephones,
use their own pair to their own port on a digital line card, so
two-pair wiring is required where a station user has both voice and
data requirements. Data is circuit-switched through the system on a
transparent basis. '
Tadiran can go to quite
small sizes with voice and data capability; Tadiran circuit-switches
synchronous data, but for asynchronous data uses packet switching
(or stat-mux) techniques to pack many connections into a single 64
Kb/s channel. The packet assembler and disassembler (PAD) used also
makes speed conversions between connected data ports. Tadiran, too,
is working on a digital CO interface.
The ITT 3100 uses
standard 3-pair wiring between line card and telephone and, as one
method of handling data, has interfaces to map RS-232 signals
directly onto the 6 wires available. At the PBX, the RS-232 signals
can then be switched directly as in a data PBX, and for about the
same incremental cost per input.
Both CXC and Ztel have
emphasized their ability to go to large sizes, but single modules
should be economical down to sizes well under 100 lines. Both
emphasize the ability to handle voice and data in their advertising.
Redcom has just introduced a digital telephone to provide voice and
data switching, a sort of 1B+D arrangement on a single pair, and may
offer other improvements in the near future.
Conclusions
These and other
approaches suggest the small office has a chance to tie its PCs
together via the telephone system even yet. And, with luck, maybe
digital connections to other offices can be switched through one or
another of the ever more digital public or private networks. But
should we rush out and buy? The main reason for caution is the
imminence of ISDN. Nobody today has a 2B+D digital telephone set,
although some present designs may turn out to be quite close to the
final standards when available. And 23B+D standards, complete with
CCITT # 7 Common Channel Signaling on the D channel, are also just
around the corner.
The main question whose
answer we all need is this: Can the internal data communication,
presently offered between units of proprietary station equipment, be
mapped in a standard way into B channels to other PBXs and COs in
the future ISDN pattern? If the answer is yes, then existing systems
can interface each other through ISDN with no more than a change to
23B+D equipment. If the answer is no, then internal digital
communication is about all we can expect, proprietary sets will do
fine forever, and modem pools will take on greater importance in the
future than they have today.
Somehow, it seems silly
to take a digital data stream and convert it into an
analog-appearing whistle to send it through an interface into a
digital network where it is recoded into digital once again for
transmission, and then, at the far end, have the reverse process
take place. But stranger things have happened. After all, we are
still driving around in Model T Fords, cleverly disguised by decades
of cosmetic manipulation and marketing hype.
[
Top ] [ Next ] [
Table of Contents
] |
