2026: International Year of Retro Communications

Events
1

2026 is the International Year of Retro Communications

We have lost the ability to land people on the moon and we have lost the ability for our modems to talk to each other. We can fix that. At least the modems bit, anyhow.

Communications across distances is a very important function for computers and most vintage computers have that ability through serial interfaces. The focus of the International Year of Retro Communications is to get our modems singing to each other across continents.

Modems connecting on a bench is fun. Getting them to connect within a building is interesting. Connecting any two modems located anywhere on Earth will be nearly as significant now as it was the first time around.

Ethernet and other networking technologies are equally important. They are definitely in-scope and any contributions toward increased networking of vintage computers is very welcome.

The best outcome of this focus on retro communications will be a standardised solution to restore period correct communications between any vintage device using the Internet to replace decommissioned telephone networks (eg PSTN/POTS, ISDN).

Spread the word and please report back here with responses, reports, progress, guides, howtos, solutions, blogs, vlogs, not so much “short video format” :smiling_face:

2 Likes
2

No problem, just use an acoustic coupler at each end of the tele… oh, yeah, we don’t have telephone lines anymore…

2 Likes
3

Strangely I did start pondering this a year or so ago.
the easiest bet might be to find an old Digital (but not IP) PABX somewhere with both POTs and ISDN Primary rate ports

And if that cannot be done, strangely a cunning plan/ I mean design is slowly coming together…

Unfortunately many of the ICes that would have made this easier are no longer made, so a tad tricky to get hold of, for example I was unable to identify any sources of A-law or mu-law Codecs, also I was unable to source any HDB3 encoders/decoders.

Control board has clocking (of course) and G704 interface - (primary rate) so would do Timeslot 0 detection (incoming) - technically this would be the “exchange” so generate the master clock.

Of course the Control board tracks timeslots, so I am thinking a 5 bit counter (for all 32 Time slots) feeds to the line cards and then use a comparator on the line board to identify the required timeslot.

  • my initial thought was this was to be an Analogue to E1 concentrator, but then as I was pondering tone generation, and DTMF decoding, I realised you could run two switch planes. So potentially 2 sets of counters?

For the POTs interface I’m thinking 5 channels to a card, we can still purchase AG1171S Subs Line Interfaces, these will usefully run off +5V and generate power feed, and ringing current.

A-law/mu-law Codecs are now a tad tricky to obtain, but you can use a lookup table to translate between a 13 bit ADC/DAC to 8 bit A-law/mu-law, nowadays it is probably cheaper and easier to use a microcontroller (if I can find one with 14 bit ADC) than an EPROM although they don’t usually include a DAC, but a 13 or 14 bit 8Ksps DAC shouldn’t be too hard to do.

To serialise the data, each channel has a couple of shift registers 74LS165 and 74LS595 would work here. Data is loaded between the ADC/DAC and the shift registers during Time slot 0. - of course this is assuming parallel ADC/DAC and these seem less common now?

still plenty of detail to work out, such as HDB3 encoding/decoding, Protocol engine for signalling etc…

1 Like
4

Thanks for your input. You are absolutely the person we need. My experience and knowledge ended at the wall socket.

I would like to solve the general problem of replacing no-longer-available telecom services using the internet as the interconnect. It has to be a new solution that can be built with modern parts as old parts can have supply and reliability issues. The scope of services to include also needs defining: PSTN, ISDN, ADSL, DDN, etc.

Do you mean Interface Modules Subscriber Line Interface Circuit (SLIC)?

I started to think about designing an interconnect for analogue modems but it’s too far out of my wheel house. Using modern microcontrollers and software libraries to brute force that which was previously done in silicon. You might find some ideas from my notes:

My project for this year is to get analogue telephone adapters (ATA) talking to each other over the internet so that distant analogue modems can connect once again. ATAs, like Cisco/Linksys SPA 122 ATA and Grandstream, are easy enough to find on eBay. They’re voice focused so my measure of success is to get 14.4kbps between any two modems on earth.
It’s a mix of local dial plans and ideally a centralised directory system to lookup IP address and ports.

5

The Silvertel Module does Battery (48V from a 5V supply) Overvoltage, Ringing, Signalling (loop signalling) and Hybrid functions so basically the BORSH section, I have found that the W6810ISG µ-Law A-Law CODEC is currently still in production, so just trying to work out something using them.

What I am thinking is a </= 30 Analogue channels (depending on how many line cards you plug in) with a 2,048Kbps serial bus to cross connect between them, and another 2,048kbps external port for a primary rate interface - so you could then support 56K modems.

As for connecting over IP, that is going to be a whole other source of joy, IP/Ethernet networks are asynchronous, and don’t always support quality of Service, so typically you would be using pseudowire, and derived timing (or Stratum 1 - 2.048 Mhz clocks at each end - probably more realistically though GPS derived timing, which is technically, stratum 2 (or is it 3?)) but you want a healthy chunk of buffering built in, so latency is going to be higher than you might prefer

6

One practical angle that may fit this theme is the newer SX212 drop-in Wi-Fi replacement boards coming out of the Atari community. Even though they wear an Atari badge, the original SX212 was already RS-232 and Hayes-style, so in practice the same approach can be used on other systems that expect a serial modem. From the computer side it still “feels” like dial-up, while the transport underneath is IP.

We’re also seeing similar ideas elsewhere. Projects like Zimodem, and some drop-in Wi-Fi modem replacements in the Commodore world, take the same approach: keep the original modem behavior, swap out the phone network underneath, and let vintage software keep doing what it always did.

Alongside that, devices like FujiNet show a complementary path where the network is exposed more directly, and across multiple platforms, without pretending to be a modem at all.

None of these replace rebuilding the old phone system itself, but they do show there are already working, repeatable ways to get vintage machines talking across real distances again.

1 Like