Friday, March 12, 2010

101 ESS

The 101 ESS was a model of telephone switch which played a very important part in Ma Bell's emerging technology, but it only appears in fleeting references in historical documents, eclipsed by the 1 ESS and others.

The 101 was Ma Bell's first! Yes, predating the 1E!

Some will tell you that the 101 ESS is essentially the red-headed stepchild of the Bell System. It was quirky, had a short operational life, and had a bad reputation in the field.

For the time, the 101 was ahead of its time, literally. It had many innovations which are reflected in the mature technology of today. It was the first 'stored program control' switching system used in commercial service. (More about that term below.) It was also the first switch to use TDM (Time Division Multiplexing) to switch voice paths. It was Ma Bell's first commercially-deployed switch which supported Touch-Tone.

The 101 began life in the 1950s as the EPBX project, Ma Bell's initiative to create a state-of-the-art all-electronic small telephone switch which could be deployed at a customer's premises in lieu of the very bulky, noisy, and space-consuming Step equipment, and the somewhat bulky Crossbar PBX systems.

It was renamed the 101 ESS after the term ESS was settled upon to be the service mark of Ma Bell's electronic switches. The number 101 was chosen since full-scale CO switches were normally assigned single-digit numbers, and PBX systems were normally assigned three-digit numbers. 101 was, of course, the lowest number available under those criteria.

The numbers vary as far as the line capacity of the 101. Some literature will cite limits in the hundreds of lines, others refer to low thousands of lines being available from later 101 systems.

The strategy was to place a large central control unit in the CO, acting as the equivalent of the line link section of an otherwise normal 5 Crossbar office, and a 'smart' remote unit at each customer's premises, connected by an appropriate number of signal and control trunks to the CO. To be commercially viable, each CO supporting 101 service had to serve a certain number of remote customers, 10 or so, depending on size and density.

The 101 service was sold under the service mark of 'Centrex-CU', the CU meaning that (some of) the switching equipment resided at the customer's premises. Centrex-CU included normal features such as direct inward dialing to four-digit extensions, direct outward 'dial 9' service with ANI (automatic calling number identification) with detailed reports and variable billing options, and such 'Custom Calling Features' such as speed dialing, call transfer, camp-on, remote pick-up of a ringing extension, call forwarding, and user-initiated 3-way and 4-way conference calling.

The 101 was also used for some stand-alone, non-DID PBX applications, but these were few and far between.

The technology used at the time for the 101 was discrete-component diode-transistor logic (DTL), making up the bulk of the electronics.

The switch fabric used an innovative but unusual form of analog time division multiplexing. Instead of multiple physical audio paths through the switch matrix, there was one path which was time-shared by all active calls, each being switched in and out at the rate of 12,000 samples per second. The actual 'switching' of audio paths involved assigning the appropriate timeslot to a particular call.

The audio signals were smoothed and integrated by capacitors, giving full response in the telephony band of roughly 300 to 3000 Hz. The 12kHz sampling signal was well above the (intended) bandwidth of the system, but some claimed to hear it, or its effects. Most users, however, heard only the usual crystal-clear analog telephone signals which were expected and typical of telephone systems of that age.

The first production roll-out (that I can find) was at Mayo Clinic in 1963.

It would seem to some that the 101 might be well-suited to CDO (rural Community Dial Office) operation, in a similar architecture to the host-remote scheme commonly seen today. This, unfortunately, was not practical with the 101. If it were, I'm sure that the 101 would have seen a longer and far more visible life. The design of the analog section of the 101 precluded very long loops between the switch and the set. Although off-premises extensions from 101 systems did exist and were somewhat common, the loop length was limited to 10,000 feet or so, and in rural offices, a good percentage of customer loops exceed this threshold. (There's also no evidence that the 101 ever supported party lines of any type, which were still required by the lack of adequate outside plant during the life of the 101.)

There's not that much literature in the public that talks about the nuts and bolts of the 101. The items that are widely available are full of arcane terms such as Stored Program Control, Program Store, Call Store, etc., terms that are not commonly used even by today's hardware-savvy technoids.

Let's talk about some of these.

Stored Program Control means computer control. The system is controlled by a 'stored program engine' which most of us would call a CPU. This CPU was built of discrete components, no microprocessor or VLSI, not even any SSI in the original 101. Computer architects might even call this a very early RISC processor, since the instruction set was small and primitive. However, it had a few very specialized instructions which were well-adapted to the control of the switch.

Program Store is what we would call the executable code, stored in a 1960s era version of what we might call ROM, or Read Only Memory. The ROM memory cards were aluminum-base cards with each bit represented by a magnetized or non-magnetized spot on the card which could be read by the CPU.

Likewise we can think of Call Store, used to store the call information and other items making up the state of the switch, as RAM.

The RAM at the time was ferrite core memory, with each bit being represented by a ferrite bead. The beads were magnetized by a 'write' wire, and the state of each bead was read by a matrix of 'scan' and 'sense' wires.

The 101 could not be re-programmed on the fly in the field. New code sets ('generics') had to be delivered on mag card sets. There were no on-line mass-storage units such as disk drives. These did not come to switching technology until the advent of the 1A ESS many years later.

Many of the features of the 'CPU' of the 101 were adapted to the 1 ESS, but the 1E did not use the TDM switch fabric, instead employing 'fereed' crosspoint switches in crossbar-like configuration.

In Omaha, the 101 Centrex-CU service was promoted heavily beginning in the late 1960s and (at least) two #5 Crossbar offices were adapted to support 101 ESS customers.

The Omaha 101 installations which I can find evidence of are:


Internorth, 348
Clarkson Hospital, 348
Omaha Police, 348
Omaha National Bank, 348
Norwest Bank (now Wells Fargo), 536
Lutheran Hospital, 536
Creighton University, 536
OPPD, 536
UNMC, 536, expanded and cut to 541
Lozier Corporation, 541

This, of course, was the critical mass necessary to make the service practical and profitable to Ma Bell.

Such was not the case in west Omaha.

90th and Western (originally hosted at 135th):

Bergan Mercy Hospital, 398
IBM Corporation, internal PBX, not Centrex, no DID.

Even with the vast westward expansion of Omaha's business community between the 1970s and 1990s, the 101 ESS never saw widespread deployment in the 'burbs', the service being mostly pushed out of the way by 'Centrex CO' services hosted out of 1E and 1A switches.

One of my former employers was 'served' by one of the above 101 installations. My desk phone, for the most part, worked perfectly well as far as regular POTS features, but the custom features such as forwarding, transfer, etc., were sporadic at best.

I know that it worked for 103 type 300bps data calls, not so sure about 1200 and above.

The one quirk which I remember most about the switch was the cadence of the ringdown and ringback. The 101 tone generator used the newer 'precise' sounds for such things as dial tone, ring tone, and busy tone. However, when a call was placed, the first ring was often several times the normal length, then often stuttering before settling in to the normal 1-on, 2-off cadence typical of PBX installations.

I have never received any explanation of why this happened.

I do know that in a later 101 installation in NYC, this quirk was not present, and that 101 had the normal ringback cadence found in other systems.

All 101 installations appeared to have vanished by the mid 1990s or so, mostly replaced by Centrex-CO and Centrex-CU on various Dimension, Definity, and systems of other manufacturers.

An interesting sidebar to the history of the 101 is a legend on the etymology of the term 'octothorpe' to refer to the pound/hash/sharp/whatever sign. It involves the tale of the name being coined on the fly during the commissioning of the Mayo Clinic's Centrex. Researching this will be an exercise for the student. ;-) Hint: Check the archives of Telecom Digest. ;-)

No comments:

Post a Comment