The PBX Scene: January, 1986

Some Recent Events

In the September-November period, Harris announced the PBX version of its 20-20 switch, Wang bought into TeleNova to expand its PBX line below the size range normally handled by InteCom, and IBM (finally) announced its token ring LAN. Not to be outdone, AT&T announced a faster version of its Starlan. But these upbeat activities were counteracted to some extent by Atlantic Richfield selling out its telecom interests in their joint venture to Ericsson, and then Ericsson closing a PBX plant in Anaheim and leaving the PC business. Both Northern Telecom and Mitel closed some plants, moving production to others; in the wake of British Telecom's buyup of Mitel, Michael Cowpland has been replaced as President and CEO by a financial man. And United Technologies planned to close its Mostek division, but later chose to sell it to Thomson CSF, and then put its telecommunications division up for sale. Presumably there will be no difficulty in finding a buyer for the former Lexar and former Stromberg Carlson digital PBXs, but it looks like the advocates of competition may be getting more than they bargained for. Even MCI is pulling out of the highly competitive paging and cellular mobile telephone fields.

To drive the point home, much space in the telecom journals was given over to objections to liberalization of the various treaties with the ex-Bell-System. The BOCs are selling Centrex with what some would like to believe are features which Centrex customers should be denied, and the unregulated segments of the BOCs are charging ahead selling CPE at prices reported to be very low. NATA indicates that BOC growth in this area will endanger the interconnect industry. AT&T, on the other hand, is now permitted to combine its marketing of both CPE and long distance service without setting up a separate subsidiary, and the FCC's Computer Inquiry III may loosen things up even more. One wonders just how much competition the country can stand.

The Rolm T-Compatible Flap

Rolm's 12 bit linear PCM coding, running at 12,000 samples per second, has been the subject of discussion for 10 years. During that time, Rolm has been able to advertise a digital PBX, and those who equate "digital" with "modern" and "office-of-the-future" have been delighted. The curmudgeons who pointed out that the digital nature of the switch was entirely internal, invisible to the customer, that there were a great many different kinds of digital PBX on the market, and that all digitals are not created equal, were looked upon with scorn. But the times they are a- changing.

The extensive deployment of T-Carrier world-wide as a transmission system has changed things. T-Carrier uses 8 bit companded PCM coding, running at 8000 samples per second. The word "companded" is this context means that small analog signals are coded differently from large ones so that one need not use as many bits for the same quality of analog-to-digital conversion as in the linear coding process. T-Carrier encodes voice into only one third the number of pulses per second used by the Rolm CBX, and is thus incompatible with Rolm's coding.

In 1975, this was of no concern. Even though PCM (Pulse Code Modulation) was used in T-Carrier, "channel banks" at each end of the transmission facility looked to switches, PBX and CO alike, as though they terminated 24 separate analog voice channels. PBX trunk circuits, even from digital PBXs, brought out individual analog voice channels which were then re-coded to digital in the transmission channel bank and multiplexed together to let 24 circuits be carried on two one-way transmission facilities such as pairs of copper wire, coaxial cable or microwave radio. In short, digital transmission facilities were tricked up to look like the analog trunks they replaced, and digital switches had their internal digital signals changed back to analog to match the trunks.

The world started to change in 1976, however, when AT&T first introduced its 4ESS, a large (to 100,000 trunks) toll switch that accepted T-Carrier span lines directly, and switched individual channels in the T-Carrier digital format, without de-multiplexing or decoding to analog. Eliminating channel banks at the 4ESS end made the "prove-in distance" for T-Carrier vs. individual copper pairs for short haul trunks go to zero miles, and T-Carrier, already heavily used for short-haul transmission, increased even more its applicability.

The coming of fiber optics in the early 1980s also made long-haul trunks on T-Carrier practical (microwave gives more channels for the available bandwidth if analog modulation and multiplexing techniques are used rather than digital). So today, for both local and long distance, digital transmission facilities are the preferred way to go within and between most sections of the country. Not only can channel banks be eliminated at compatible digital switches, but higher level multiplexing (combining di-groups of 24 channels) is also much less expensive. But fiber optics has so much capability that it can also be used for private networks, and AT&T and other long distance companies are finally offering T-Carrier span lines to business customers for such use. Where customers have T-Compatible PBXs, they can connect in the same way 4ESS has done for almost a decade.

Another factor in T-Carrier-related developments is the ISDN, or Integrated Services Digital Network. Suddenly emerging from the obscurity of the telephone engineering departments in various telephone companies around the world, ISDN has gone public with much hoopla and whoop-de-doo. The general idea of ISDN is to let people use T-Carrier for end to end connections via T-Carrier (at 64 Kbps) on a dial up basis, and use the channel so established for voice, data, image, or whatever. 50 Kbps data for the price of a phone call is one of the possibilities offered by ISDN.

In this context, a story in the Oct. 23 MIS Week caused a certain amount of stir. The story suggested that Rolm was going to abandon its 12 bit, 12000 sample per second (12x12) matrix coding and bring out a new PBX that is completely T-Carrier compatible. The article went on to suggest that Rolm was developing a whole new technology to do this, and that IBM was supposed to have bought Rolm for this technology. The conclusion was that Rolm's present line of CBXs would "become obsolete," and Rolm would be letting its present customer base down.

Checking with Rolm indicates that the above is something of an exaggeration. Rolm has no intention of abandoning its non-standard coding. They have spent a lot time on their T1-D3 interface, which maps voice-like signals from 12x12 coding into 8 bit companded, 8000 sample per second (8x8) coding; further, on Dec. 2, Rolm announced compatibility of their T1-D3 interface with a digital network called Cohesive Network CN-1, making use of T-Carrier span lines. On the line side, the voice path in the ROLMlink to the new ROLMphone family is T-Compatible, and is mapped into the matrix PCM in the same way. Thus, at least for voice calls, Rolm can use T-Carrier digital facilities quite well.

As for IBM's interest in Rolm's new 8-bit technology, this seems very unlikely indeed. IBM was, for some time, interested in Mitel's SX-2000 development, specifically for the new chips Mitel was developing to make the SX-2000 possible. When that development took too long, IBM dropped Mitel and bought Rolm. In the meantime, Mitel has brought its chips, and the system for which they were designed, to market, while Rolm has been busy with other things. Would IBM really drop Mitel's technology, which is on the market now, in favor of a Rolm technology which is not only not announced, but actively denied? It seems more likely that the Mitel arrangement has been, at some point, confused with that for Rolm.

Rolm obviously understands the "8x8" technology and undoubtedly has people working on it in one form or another. In spite of strong statements to the contrary, the present Rolm switching matrix is neither T-Carrier nor ISDN compatible in the sense that voice and data (and, later, image) can be handled in the same way internally, and can connect in the same way to external facilities. Rolm's voice and data switching are completely segregated, and while switched connections via T1-D3 can be made for voice, separate data channels, segregated from the voice group internal to the T-line, must be accessed separately without use of the data-switching features.

But just suppose Rolm did come out with an "8x8" PBX. What would be its impact on Rolm's existing customer base? Probably zilch. Just having a new product does not mean that old products are forgotten. There are a lot of ETS 100 sets out there, and nobody is panicking about what the ROLMphones are doing to them. Rolm has a good track record of assuring the compatibility of everything it has in the field, and that will not change.

But consider any PBX on the market today. The software is vastly more important than the hardware, both in terms of what it does and in the cost of development. It seems most unlikely that Rolm would abandon its present processors, memory and software in any hypothetical new system. Just as AT&T's System 85 uses Feature Package 8 from the earlier Dimension 2000, so Rolm would probably use the entire CBX control complex in any new machine; developing a new software-compatible 8x8 matrix with matching line and trunk cards would be easy and inexpensive. And the same control system would assure support, where it matters, forever, for systems with the older hardware.

The argument that a new Rolm 8x8 PBX would make the older CBX "obsolete" is obviously silly, no matter what your definition of obsolete may be. In a certain sense, all other forms of digital transmission were made obsolete in 1962 when AT&T first introduced T-Carrier. Whether T-Carrier is good, bad or indifferent, it is there in vast quantities, and systems not compatible with it will not be as effective in making external connections as those that are. You can "transcode" from Rolm's 12x12, delta mod in its variety of forms, PAM or whatever, but it is not the same as maintaining bit and byte integrity end to end. And maybe, with luck, the use of digital technology between PBXs may soon be as available as its use internally, and for data and image as well as voice. That's what ISDN can do for us, if we're lucky.

Adventures with the ISDN

Bell Atlantic had an ISDN demonstration for the consultants in the Philadelphia area on Dec. 4. Two of us were present, so we had the full attention of the Bell Atlantic people, who outnumbered us about 5 to 1. And we found out many interesting things, most of which are of considerable interest to PBX customers.

First, 2B+D for Centrex Extensions is the main thing that interests Bell Atlantic. The 23+D channel (of most interest to customers with digital PBXs) was hardly mentioned. Second, the 2B+D channel, using time compression multiplexing (ping-pong), will be able to work over a 12,000 foot loop of 26 gauge wire. Checking through this manual, it is hard to find any electronic sets with a range half that great. Somebody must have made a break-through somewhere. Third, Bell Atlantic has no intention of using AT&T's "proprietary" CCIS for CO to Toll Office signaling. They will wait for CCITT No. 7 Common Channel Signaling, which may require several more years in the standards stage in Geneva. In the meantime, they are using digital trunks from their local offices to AT&T's toll offices; presumably the signaling employed is E&M, using the bit-robbing technique in each voice channel, reducing it to 56 Kbps from 64 Kbps.

Two things become apparent from the above. The BOCs, if Bell Atlantic is an example, are far more interested in using ISDN to sell Centrex for local communication than they are in developing long-distance digital communication through AT&T or any other long distance carrier. And second, T-Compatible PBX customers may find the long-haul carriers much more interested in direct digital connections than are the BOCs. However, getting there may or not be half the fun. The whole point with Common Channel Signaling is that it gives you 64 Kbps "clear" channels. Presumably, within AT&T's toll network, where CCIS is widely used, digital channels are all clear. But without 23B+D to get there, T1-D3 channels with bit-robbing supervision will be necessary. This will limit all data communication to 56 Kbps, to avoid troubles with the bit robbed for supervision. And 23B+D is supposed to use CCITT No. 7, which will be available to the customer about the same time it will be available to the BOCs.

In the meantime, we have Northern Telecom's CPI, or Computer to PBX interface, which uses 24 bit-robbed 56 Kbps channels on a T-span, while AT&T is pushing DMI, or Digital Multiplex Interface, which uses 23 64 Kbps clear channels with a separate signaling channel. These are both, presumably, for going into a computer from a PBX. But maybe they could go into a long-distance network as well. And for those who want to find out how things are really going to be done, there is always AT&T's new PUB 41459, Integrated Services Digital Network (ISDN) Primary Rate Interface (June, 1985), which spells out the whole thing. Maybe tomorrow got here yesterday.

What's In This Mailing.

This mailing includes one new system, the UTX 1001, which used to be and may yet again be the Lexar. We also have a more complete version of the Hitachi, and an update on Rolm. The principal news there is the fact that Rolm now considers ROLMbus 295 an available option, and is shipping them regularly. There is also an upgrade on the VSCBX, and miscellaneous information such as the extension of ROLMlink to more distance ROLMphones.

[ Top ] [ Next ] [ Table of Contents ]

Copyright 2005 Lee Goeller. All Rights Reserved.