How did phone companies and companies that made PBX's do switching with discrete logic or simple, non-programmable ICs?

The caller would type in a number, so the system is receiving a sequential code, and based on this code it has connect two wires together. What was the standard electronic strategy to do this?

Just to clarify here: I am not asking about electro-mechanical. I am talking about digital ttl logic, just before it was programmable. So we are talking 1980s in the United States.

  • 6
    \$\begingroup\$ Strowger switches! They're fascinating, honestly. \$\endgroup\$
    – Hearth
    Jun 13, 2021 at 2:35
  • 1
    \$\begingroup\$ I think it is instructive to watch this historical video by Dan Gelbart. The very first part of the video is exactly about what you want to know. The very first automated switching system, ever, for the phone system and the reasons why it was invented, too! \$\endgroup\$
    – jonk
    Jun 13, 2021 at 2:53
  • 2
    \$\begingroup\$ The user wouldn't type in a number early on, they would spin a rotary dial that would return via spring and send a number of pulses equal to the digit being dialed. Called pulse dialing. Only afterwards did touch-tone appear where you press buttons on a keypad and type in a number. \$\endgroup\$
    – DKNguyen
    Jun 13, 2021 at 2:54
  • 1
    \$\begingroup\$ You can build computer systems entirely from "non-programmable ICs." \$\endgroup\$ Jun 13, 2021 at 2:55
  • 1
    \$\begingroup\$ In a comment on an answer you said, "I am not asking about pre-digital technology." So what are you asking? A modern telephone switch is a computer. The bit that figures out what number was dialed and decides how to route the call is not an electronic component, it's a software component. That's true regardless of what technology is used to actually make the connection. Are you asking how it is possible to build a computer using only "digital ttl logic?" Are you asking what technology was used to connect calls in the era when computers were built entirely out of TTL? something else? \$\endgroup\$ Jun 13, 2021 at 15:05

4 Answers 4


For a while, switching was done using electromechanical relays that established interconnect networks. A common setup at one point was the use of stepping relays that would be advanced from position to position under the control of the caller with each rapid pulse in pulse dialing - each digit would plausibly step one of these relays, routing the call from switch to switch until it reached its destination line and caused the remote ringer to ring.

A further improvement on this was the so-called two-axis switch which allowed selecting both an outgoing route and multiple outgoing "banks" from each switch to reach the next, improving the capacity of the exchange. As Wikipedia describes:

These [two-axis switches] were commonly used in telephone switching with ten banks of ten contacts. The coils were typically driven by the electrical pulses derived from a rotary telephone dial. On a two-motion selector, as a digit was dialed, the wipers would step up the banks, then automatically rotate (self-step) into the selected bank until they found an "unused" outlet to the next switch stage. The last two digits dialed would operate the connector switch (final selector in Britain). The second to last digit would cause the wipers to move up and the last digit would cause them to rotate into the bank to the called customer's line outlet. If the line was idle then ringing voltage would be applied to the called line and ringing tone was sent to the calling line.

At each stage, various conditions (such as a lack of outgoing banks) could be detected and reported. An example of such a condition is the "reorder" tone (or fast-busy tone, distinct from the normal busy tone) meaning that the call could not be processed through the network for lack of exchange resources. For example, if one of the above two-axis steppers failed to find an open bank, the caller would get the tone and the call would not be routed further.

This piece of audio demonstrates a pulse-dialing code which reaches various points in an exchange and hears different tones, before ultimately reaching the destination and finding the circuit busy.

At the same time, various signalling schemes such as the use of DC voltages and tones was used to coordinate the state of scarce shared resources such as long-distance trunks.

  • \$\begingroup\$ I am asking about the post-mechanical period. \$\endgroup\$ Jun 13, 2021 at 14:31
  • \$\begingroup\$ @TylerDurden I don't see what the real issue is. If you're doing something with electromechanical relays it's certainly feasible (easier even) to just replace the relays with solid-state logic gates, electromechanical relays also being unprogrammable, power hungry, and massive in comparison. \$\endgroup\$
    – DKNguyen
    Jun 13, 2021 at 14:36
  • \$\begingroup\$ @TylerDurden As your question was written twelve hours ago, it asked for "logic", and relay logic is a fully valid form of logic. If you're asking about something like the Western 1ESS which was in service from 1976 through 2017 (!), then you had a digital CPU controlling reed relays using a stored program - nothing too interesting in itself, as a lot of the underlying topology was inherited and optimized from the electromechanical era anyway. You can probably find other details of topology... \$\endgroup\$
    – nanofarad
    Jun 13, 2021 at 14:38
  • \$\begingroup\$ ... on the Wikipedia pages of other electronic switching systems of the era, but I won't have time to rewrite this answer to address the new constraints until this evening (if I even find time then). It seems from skimming that the trend was microprocessors or distributed systems of microprocessors, overlaid on top of appropriate line hardware (anything from digital trunks to reed relays connected to the analog local loops), rather than fully non-programmable solid-state electronics. With all the investment in relays, a non-programmable IC was probably not worth it as a stepping stone. \$\endgroup\$
    – nanofarad
    Jun 13, 2021 at 14:41

Mostly it was done with relays.

Before 'touch tone' dialing, phones had rotary dials which sent the numbers as a pulse train of 1 to 10 pulses per digit. Calls were routed either by operating rotary switches with electromagnets that were stepped with the dial pulses, or relay based counters and hard-wired controllers (called 'markers') that computed the route and operated horizontal and vertical bars on 'crossbar' switches to connect the call. The markers in crossbar exchanges had hundreds of relays in them and were very complicated.

Since these systems were already in place when DTMF phones came into use, the existing exchanges were simply supplied with tone decoders that either fed the 2 of 5 codes into the same relay logic circuits that normally took the output of relay counters in crossbar exchanges, or converted them to pulses using some basic digital logic or an MCU for step-by-step and rotary exchanges. The electronic circuitry was confined to a single card per decoder, and the rest of the exchange operated the same as always.

I worked on PBXs in the 1980s and early 1990s when the transition from mechanical to electronic systems occurred in New Zealand. None of our electromechanical systems worked with touch-tone phones. To get around this we had telephones with push button keypads that generated rotary dial pulses inside the phone. These used either a small MCU or a dedicated IC to do the job.

As for 'discrete logic or simple, non-programmable ICs' for switching, except for DIY home phone systems I don't know of any PBX that used discrete solid state logic. In New Zealand we went straight to 'stored program control' using microprocessor based systems (which in some cases were less reliable than the electromechanical systems they replaced, were full of software bugs, and still used relays to connect the lines!).

  • 2
    \$\begingroup\$ In the US early touch tone phones usually had a switch to put them in pulse dialing mode. Presumably because some exchanges took longer to switch over to DTMF equipment than others. \$\endgroup\$
    – The Photon
    Jun 13, 2021 at 4:26
  • \$\begingroup\$ I am asking about the post-mechnical technology. If you read my question it is asking about digital discrete logic, just before it could be programmed with microcontrollers. So I am asking about the specific ICs and TTL logic used to do that. \$\endgroup\$ Jun 13, 2021 at 14:33
  • \$\begingroup\$ @TylerDurden As I said, I do not know of any "phone companies and companies that made PBX's" who sold PBX systems that did switching with "discrete logic or simple, non-programmable ICs". If there were any the NZPO was not interested in them. At one time we were evaluating an exchange that used reed relays, possibly a Philips EBX 8000 but AFAIK it was not put into service. This system had a CPU and stored program control, presumably made from 'discrete' logic ICs like other computers of that time. \$\endgroup\$ Jun 13, 2021 at 20:46
  • \$\begingroup\$ BTW in 1980 the Post Office sent me on a 2 week training course to learn about microprocessors, presumably because they thought we would have to service equipment that used them. I was so impressed by the little MC6800 boards we programmed on the course that soon afterwards I built my own 6800 based computer (all my own design both hardware and software, using only the datasheets for reference!) \$\endgroup\$ Jun 13, 2021 at 21:13

Before digital switching it was done with "stepping switches" or "step-by-step switches".

Here's a video which describes them:

"AT&T Archives: The Step-By-Step Switch"


  • \$\begingroup\$ In the U.S.A., those old Strowger switches started to give way to the crossbar system in the 1960s. I don't know enough about how either one of them worked though to be able to expound on the differences. \$\endgroup\$ Jun 13, 2021 at 2:47
  • \$\begingroup\$ I am not asking about pre-digital technology. Read the question. \$\endgroup\$ Jun 13, 2021 at 14:33

Telephone equipment couldn't be changed in one night due the massive amount of it. Every new piece of equipment had to work with the old ones. You want details of a non-existent era. No TTL and discrete transistor era without electromechanical switches in major roles has existed. Computers were already used as brains in some systems say 40 years ago, but still the final switching happened in electromechanical circuits.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.