Hacker News Reader

Comment by Animats | Hacker News Reader

That's an AUTOVON switching center.[1] There were at least 38 of those centers in the US. They were located in places some distance from major cities and military targets. They were hardened telephone central offices, but with many more redundant links between switches than the commercial system. So this system really was intended to survive a nuclear war.

The technology was Western Electric's 1ESS (#1 Electronic Switching System), and all 4-wire out to the handsets, so that conference calls would work clearly without feedback. 1ESS was a very bulky system. It was basically a pair of large mainframe computers running a big dumb switch fabric. The switch fabric is analog and electromechanical, using reed switches with a ferrite element so they stay in the last state to which they were set. That's why these were such big installations, even though they didn't have a huge number of lines.

[1] http://autovon.org

View on HN

The underground sites were primarily for Long Lines equipment, i.e. L-carrier Coaxial repower, regen, cross-connect and HVAC and power continuity for these. You can see in the illustration that switching at its least efficient was maybe 1/4 of the facility (lowest level), L (and TD MW) would be a bit more dense but similar floor space on first level. These multistory sites had a lot of extra room for training rooms, service bureau, and some nod toward continuity in terms of sheltering a number of people with some token supplies although if you look at enough pictures over time it doesn't appear like it was ever taken very seriously... ultimately I think these just turned out to be a way AT&T and the DoD came up with to get the US Government to more heavily subsidize Long Lines network construction.

Switching of copper end lines would often happen closer to the user, i.e. on base although some sites did have switching due to favorable proximity (i.e. Soccoro, N.M) or presumably function like a tandem (maybe this site?). You can see a little of a 1A ESS in this video (https://www.facebook.com/CheshireVolFireDept/videos/a-brief-...) and maybe some 5ESS in the background as well although it is too brief for me to tell.

Some undergrounds were dual purposed for Microwave pathing and cross connect (like this linked one), but most microwave was instead in above ground hardened facilities elsewhere for path diversity.

Some undergrounds had Echo Fox transceivers and switching http://www.coldwar-c4i.net/Echo-Fox/index.html.

Project Offices are an interesting related rabbit hole to pursue http://www.coldwar-c4i.net/ATT_Project/index.html.

Source: I own an L-3 regen bunker and have done a lot of research on them.

> and maybe some 5ESS in the background as well although it is too brief for me to tell.

The frame at 2:22 looks to me like it has a 5E in view.

They also had Number 5 Crossbar switches as well, the switching fabric wasn't huge in size, like I've seen what the frames look like, ESS was still much smaller than the crossbar that preceded it, and not that much physically larger than a comparable 5ESS

How deep did they bury the wires?

Were they run full depth from point to point?

Google "L4 transcontinental cable", but the majority of the long lines network was the famously known horn antennas on towers for FDD microwave point to point links in the 6GHz band.

It would vary by terrain and land use (e.g. agricultural), but generally just 2-3 feet deep by vibratory plow. Deeper emplacement and directional drilling were used as required to handle obstacles. For most L-carrier the entire en-route infrastructure was below ground, but it was more extensive than just the cable, with active repeaters in manholes required at 2-mile intervals for L-4. L-4 also required an "equalizing repeater" about every 50 miles, which was installed in a manhole but had a shed on top to facilitate technicians adjusting the equalization. Main stations, such as this one, were required every 150 miles.

I’m surprised they didn’t use crossbar electromechanical switches for EMP resistance.

They did - it wasn't all ESS

http://autovon.org/wp-content/uploads/2018/01/BELL-LAB-RECOR...

These AT&T facilities were hardened against EMP to the era's military design standards, and AT&T participated in validation experiments using both actual nuclear detonations and EMP simulators. I've never seen any indication that AT&T thought the electromechanical switches had an advantage in this context, but it's an interesting question if they performed any evaluation. In general, though, EMP protection was done at the "envelope" of the facility, and equipment inside of the facility did not need to itself be EMP hardened. Hardening of the facility was achieved mostly by a shielding system embedded in the outside walls, and of course AT&T had already performed considerable research into suppressing transients on the outside plant due to lightning.

Crossbar switches were indeed in use in the AUTOVON network for simple scheduling reasons, a number of AUTOVON switches were installed before the 1ESS was ready. Eventually all of the 5XBs were replaced by 1ESS. Some Automatic Electric switches were used at AUTOVON sites outside of Bell territory, these were at least semi-custom (AE just called them "the AUTOVON switch") electromechanical machines.

The entire underground structure is an engineered Faraday cage constructed from a tremendous amount of steel rebar, copper mesh, and a central grounding and bonding network.

Reply on news.ycombinator.com