Connections Museum - 2020-08-03
All about the sounds and workings of the telephone network. There will be two back-to-back videos on the 1XB. This is the first. Our No. 1 Crossbar originally served Seattle out of the Kenwood/Lakeview central office in the northeast part of the city. It was brought to the museum in 1986. In 2018, it began a process of upgrades, adding 2 additional terminating senders, and 3 additional originating senders. The upgrade process is still underway. The marker video I mentioned: https://youtu.be/bQDXI7i0SGY 00:00 Introduction 03:50 Block Diagram 05:11 How a Crossbar Switch Works 09:33 Calls through a No. 1 Crossbar 11:05 Connection & Dialing 13:26 Originating Marker 18:03 Signaling & Terminating Sender 23:50 Party Test 25:39 Terminating Marker 30:00 Review
The knowledge it takes to make those videos is vast. Thank you Sarah and the Connections Museum for sharing all this. I know it takes a staggering effort to produce all this. I’m infinitely thankful for you making such in-depth content available.
Had no idea this (working) museum existed. So happy to see you all preserving such innovative and vintage telecom tech. Amazing videos!
So watching one call complete on an idle switch is one level of AMAZING. But how these systems allowed multiple simultaneous calls to progress at the same time truly boggles. What kind of limits were there, on a typical No. 1 crossbar serving 10,000 customers? How many people could pick up their phones and get dial tone simultaneously? How many completed calls could be going on at once? And, in practice, how often were those constraints ever encountered? Say there's some loud BOOM heard in town and hundreds of people pick up their phones... what do the people who DON'T get dial tone hear?
These are all excellent questions. Part of the problem with doing these videos is that they are necessarily simplified, and leave out a lot of stuff just so I can keep the duration to 30 minutes or so.
Everything in a telephone switch (including a No. 1 Crossbar) exists in multiple. There are HUNDREDS of senders, 6 or so originating markers, 6 or so terminating markers. The crossbar link frames all exist in multiple as well. It is all designed ahead of time for the expected traffic load of the area that it is serving, so a switch in a busy financial district will be outfitted with more equipment than a switch in a smaller low-traffic district. A typical No. 1 Crossbar could handle several thousand simultaneous calls. The limit of "how many calls can be in the dialing stage" is determined by the number of senders you have available. More dialing? More senders. The heaviest traffic day of the year was Mothers Day, and on that day, a certain small percentage of people just did not get dial tone. There was simply not enough senders to handle the load. If that was the case, you did not get dial tone, and had to wait a few minutes and try again.
The day that JFK was shot was also a HUGE day for traffic load. Many switches were not designed to handle that "once in a century" kind of volume.
There were many ways to deal with high traffic load. The easiest is to add more senders, more markers, and more link frames. That can get expensive though. You can also shorten the holding time of your common equipment so that more of it is available more often. You do this by engineering the circuit to eliminate millisecond delays in each operation. A few milliseconds in each of your senders or markers x 10,000 calls adds up to a lot.
I live in England and find your videos fascinating. I normally have to listen all the way through a few times and then the understanding clicks in.
I have recently acquired a crossbar switch (rare to find here) and so your crossbar videos are particularly relevant as I plan to use the single switch to link together the five other PAXs that I have. The concept of marking is important for this. My system will be very simple indeed but with your help I am slowly evolving some ideas as to how to achieve the objective.
Thank you so much!
By the way. How and where You bought such kind of telephone switch?! I suppose that not on eBay ;)
Thank you so much for explaining how the #1 Crossbar works! I'm curious on how the switch can determine who the customer is for billing purposes, either for local (message units) or for long distance. Also, I'd love to see how the #5 Crossbar works compared to the #1 crossbar. (I grew up in Lacey, WA where we had a #5 Crossbar.)
Aw man, billing is a huge concept. What we really should talk about is ANI (Automatic Number Identification), but we don't have any ANI equipment working at the moment. (It's on my list of stuff to get working!). I'll have to add that to my video queue to talk about it.
I greatly appreciate the content you’re publishing as someone responsible for implementing modern to bleeding edge infrastructure. Understanding how analog systems were implemented and run fills in many previously unknowns.
Ive been astonished seeing sharp young people understanding and embracing the most legacy of technologies.
Fantastic! I’d love to see a step system description video in the future. Keep up the good work!
This is so detailed and amazing that I can’t believe it. Thank you so much for these!
I find all of this to be the most fascinating stuff ever! Thanks for what you do!!! I have been interested in this telephone switching equipment since I was a kid living in NZ, I used to drive around & look through the windows of telephone exchanges with the utmost awe.
Interesting. I worked in a UK crossbar system, put in during the 1970s. There were two major designs and four (I think) types in use. There was nothing simple about crossbar systems!
This is a great series. I look forward to the next episode!
I've always wanted to know the guts of the mechanical switching systems but I grew in the era before the Internet. As a kid in the 70s I toured every CO in Yakima but was never able to get into the derail that you do. Thank you so very much! I hope that someday I can shake your hand and thank you in person. I've been to the museum a few times.
This was fantastic, thank you so much! Looking forward to the next segments!
Can you do a video on how multi-line hunting works in older telephone exchanges work (eg. business customers who have multiple phone lines, the first line is busy and the call gets routed the second, third, fourth line and so on)? How about call setup for calls terminating at private branch exchanges?
The level of design, engineer, and manufacturing represented by this technology is simply mind-boggling. And not a single transistor, tube, or integrated circuit involved. So yeah, when the internet and semiconductor industry dies, this technology will still be available to handle phone calls. 26:55 "... now the action of these frames can get rather complex ... " holy dial call switch girl ... EVERYTHING on this electro-mechanical furball is complex!!!
Awesome video. I especially liked the part at 10:40 - 11:05. That was very nicely put together and very pleasurable to watch (and to listen to!). I'm looking forward to future episodes :)
I have one question: What is the purpose/function of the thing that looks like a large key with a thick wire attached to it on the line link frame?
Oh, that! That is actually a plug, with a telephone at the other end of it. It's the phone you see @ 10:27
I like how the automatic subtitles come up with "[Music]" when it hears the sounds of the relays in the audio
I would love to learn more about the decoder and how it and the sender talk to one another. Or, indeed, more information on the sender itself, or as you point out, how that state machine deals with failure and frustration: can't find a circuit, subscriber stops dialing, subscriber dials 0 for operator, subscriber wants to direct dial long distance. I have no idea if panel switches ever got the smarts to deal with ten-digit dialing, or for that matter 1 + 7-digit as we had where I grew up, when we dialed to exotic parts of our own area code. I love the complexity of these devices, and how you have laid out your explanations to make it all comprehensible. So of course I want more, and I thank you sincerely for what you've already done.
THANK YOU. This really helped peel back the curtain to understand how the offices and frames pass signalling info in-band by temporarily switching out the voice components to send pulses then switch things back. The backwards compatibility with the panel switch was ingenious yet clearly necessary.
There's a lot of undefined jargon/terminlology which is still confusing as I'm unfamiliar with them. e.g. the abbreviations/acronyms terms used by the markers to indicate routing, "incming/final brush", etc.
Is there a reference PDF/doc explaining the signalling or terminology?
If you watch the previous video on the Panel switch those routing terms will make more sense. They all refer to things that were physical items on the Panel, but in the #1 Crossbar are reduced to conceptual values. Incoming and final were the two sets of frames used to terminate the call in Panel, and brush refers to selecting the set of contacts to be used on the selector rod
Been waiting for such detailed description for a long time, thank you
Under what circumstances does the system connect a single horizontal to more than one vertical, or vice versa, in the actual switching elements (not the crossbars used as memory)? I realized that it would be possible for a marker to test a called line for busy by just connecting itself to that line on a spare crosspoint vertical (or horizontal, whichever one it is) and then checking if it's in parallel with someone talking :) but you say there's a number group connector and a block relay frame, which sounds like extra redundancy.
Also, I would like to see more about how this is actually implemented in relay logic, probably not every detail, but more detailed than "the marker sets the crosspoints". That marker has a whole ton of relays in it - can we maybe have an overview of how it works?
I also wonder whether the 0-1-2-4-7 counting system was used because it was particularly convenient to implement.
Excellent accurate description, thank you.
I find it strange that the communication between the originating and terminating systems consists of the terminating system essentially saying "I'll count from 1 to 10 (or perhaps from 10 down to 1), stop me when I get to the right number" rather than having the originating side just pulse the right count. This is probably also a holdover for compatibility.
It is 100% a protocol compatibility thing. The 1XB was designed to work in large cities that had panel offices, so they emulated the existing protocol with relays. They did this instead of upgrading every panel switch to speak a new language, which would have been expensive, and time consuming.
Interesting video. Can you goes into more detail about the tone plant/ring generator? I think the electromechanical generation of all the calling tones would be an interesting video all by itself.
That will definitely be a future video, but our tone plant is solid state and is not very interesting. We do have a non-running mechanical ringing generator that I hope to get spinning someday though.
@Connections Museum In some countries such as France (and perhaps India), in the days of electromechanical switching, the caller would hear a "routing tone" (in France it would sound like a series of fast beeping sounds) while the call is being setup to a distant location (this presumably mutes all the switching sounds to the caller, unlike in North America you might get to hear sound sounds in the background). The "routing tone" would then stop once the ringing or busy signal starts playing from the distant exchange. Would this type of setup have been possible for those North American cross-bar/panel/step by step etc switches?
@yvr2002rtw It would have been possible, yes, but it was never developed for use here in the US. I'm not sure why the US phone companies never decided to go with this option.
Connections Museum I guess the US phone companies not doing this resulted in a generation of phone phreaks! In Europe they just stuck to boring and generic single tone dial, ringing, busy tones at 425 Hz (440Hz for France).
It would be awesome to hear how the old ring from the mechanical tone plant sounds. Thumbs up if you also think new one sounds like an angry woodpecker in a can. No offense to solid state technology is meant here.
Fascinating video. Keep em coming!!
What happens if more people request service at one time than you have register senders? I know someone wouldn't get dial tone but would they provide service to the extra line once they becomes available or does the calling party have to start over? I'm also curious about the effects that might have.
Yes, eventually the subscriber would get service from the next available sender, but in practice, they probably wouldn't wait more than a few seconds.
If hundreds of lines in excess of what the machine can handle go off hook at once though, it can lock up, and the only resolution for that is to wait for call volume to subside.
I apologize for appearing to hijack the video, but I have an urgent doctors appointment so I need to run before even watching the whole thing.
There is a webpage about the crossbar switch that has always made me drool (due to the engineering and mathematics of a non-blocking switch, and it's ability to reroute mid-call). I find it so fascinating that I keep it permanently bookmarked, as a phone phreak. Here it is:
https://en.wikipedia.org/wiki/Nonblocking_minimal_spanning_switch
Great video! Where are the compensating resistance values for each subscriber line stored?
The compensating resistance values are actually for the trunks between central offices, not for customer lines. Those values are stored in the cross-connect field on the originating marker, which is shown around 14:27 :)
My first job after high school was as a switchman in a #1 Crossbar office for New York Telephone in 1964
Not even watched it yet but I know I'm gonna like it! Thanks :)
One of the first examples of a wire protocol being maintained for backwards compatibility? That's super cool.
Thanks for this series and this instance, and I love the animations.
How is the party line test done? How can you tell apart one party from another, if they share the same tip and ring connections? Also, it just blows my mind that this is all done just with relays. Whenever you say "then it does suchandso" that "does" isn't some program instruction in a computer, it's physical wires from one relay to another with some kind of other discrete components comparing voltages or whatever, and then applying voltage to some other relay. It can't possibly ever work. Think of the minds that DESIGNED it all! (And of course the fact that its complexity absolutely pales in comparison to all our modern technology, by, like, tens of orders of magnitude. Gosh I love technology.)
Thanks for making these wonderful videos!
Relay logic is completely wild. You are correct in that there are no programs written in a conventional sense. It's all wired into the fabric of the machine. When the marker "looks" at the office frame, its attaching a series of relays through a connector circuit which will tie it in to the crosspoints of the frame. If a crosspoint is closed, a relay will operate, removing ground from another relay, which indicates "this choice isn't available". In other places, chain circuits were devised to limit access to common equipment on a first-come first served basis. Checking circuits were employed everywhere to ensure data integrity. If you want to know more about relay circuit engineering, the best book out there is "The Design of Switching Circuits", by Keister, Ritchie, and Washburn. Ritchie, is actually Alistair Ritchie, Dennis Ritchie's dad. https://archive.org/details/TheDesignOfSwitchingCircuits
A party line is testing for which side of the subscriber's telephone has the ringer connected to ground. The ringer coil (the magnet that makes the bells ring) is either on the tip side of the line, or the ring side of the line, and the far side of that coil is connected to ground. So the sender will throw a polar relay in-circuit with the subscriber's loop, and see which side of the customer's phone is connected to ground. That will indicate tip or ring party.
@Connections Museum I assume that only works for two parties per line, then. Nice, though, that with that system you wouldn't hear a ring at all for the other party. I've often wondered how billing worked with the rural farmer MULTI party lines that were common in rural Minnesota, unless those only date back to the days of operator switchboards.
@Connections Museum Thank you so much!
@Connections Museum Knowing that Phone Phreaks began sooner than my own obsession, I wonder what would happen upon "odd" conditions:
- Ringer (and it's series capacitor to ground) disconnect, leaving no identifiable party
- Ringer connection moved from tip to ring (or ring to tip) [wrong party billed], but still to ground
- Ringer moved from ground to it's normal single-party wiring to tip and ring (as if the phone was replaced with a standard single party phone)
- Two ringers, both to ground, each wired separately to tip and ring (not necessarily a hacker, but perhaps a homeowner accidentally severing the connection to a second phone in the house, and sloppily reconnecting sufficiently to get dial tone (but tip and ring swapped).
Obviously I'm not asking for hacking reasons (no longer used), but to understand how smart those "relay computers" were, understanding that Bell Labs (or Western Electric) had the best engineers.
Very good. Thank you.
Great Video thank you.
What exactly is the party line test?
Why is some of the equipment painted red? Did this mean retired or remove? Thanks.
When the machine was wired up at the museum in the 1990s, only a small part of the equipment was used. The red paint indicated which frames, which crosspoints, and which multicontact relays were actually active at the museum. It's not meaningful anymore, since I've brought up a lot more equipment since then. Just haven't gotten around to removing the paint!
@Connections Museum Thanks for the reply. When I was at Bell in old c.o.'s where equipment was retired weco would come in remove power and mark the equipment in red as 'RIP' for Retired In Place where only a couple frames were involved. Like repeater bays CARW. as examples. Thanks for the museum.
Hicken65 - 2020-08-05
This is beautiful work! You should be very proud of these productions. I'm surprised by how many people are keenly interested in electromechanical telephone switching equipment. When I put my SXS videos on YouTube, I was expecting a few hundred views over several years with only a few questions or comments. I have been overwhelmed with views (170K) and lots of questions. Now I'll be able to refer viewers with panel or XB questions, to watch your videos. I look forward to your upcoming videos.
Tom Storey - 2020-11-04
@Hicken65 Ive watched all of your videos and they are incredibly informative, the deepest dive into how they work that one could ever hope for. Answered a million questions that I had, but I still have millions more. :-D
Still amazed by the fact that all of this can be achieved with relays.
Hicken65 - 2020-11-04
@Tom Storey Thank you for your kind comments on my videos. Electromechanical relay circuits, for those who are interested in such equipment, are fascinating to watch, listen to and to be able to study their operation. Sarah and her colleagues have the equipment, knowledge and skills to be able to do this for us and are doing a wonderful job.
Beginning in the 1960's, relays proved to be too slow and too expensive to survive the onslaught of new telephony features being developed. In the Bell System, the #5 Crossbar was the last hurrah. The slowness of the common control equipment necessitated a very expensive expansion from 8 to 12 completing markers and from 4 to 6 dial tone markers. The provision of Phase III Centrex was a nightmare of massive new hardware add-ons that included a parallel switching and trunk network just to provide dial transfer capability. The time had definitely come for digital logic and software. The fact that relay switching circuitry survived as long as it did is a great testament to the involved research and development teams all around the world.
Kuba Sunderland-Ober - 2022-12-01
Those open frame relays - though beasts - were relatively slow. I wish someone made a small-scale homebrew reimplementation using reed relays. Instant x10 speed up just for using those, without even trying hard. So there could be much more parallelism just by having some multiplexed subsystems take less time to do their job. But, alas, solid state tech made it all unnecessary. If you, the reader, have a solid plan in place, I can hook you up with relays for a project like that :)
Doug Dingus - 2023-06-24
I was surprised too so I asked a few people, and this isnwhat I got:
We can see it do stuff
We can hear it do stuff
We wonder how the phones worked as kids
It is like a computer
Phones have been a thing for a really long time
My own keen interest is driven by all those human things along with how fast this seemingly big, clunky old tech actually works!
When we think about digital gates, they too take time to change States, and it all adds up, loads, fan out, sequences, timing. So do these mechanical devices, relays, rotating clicking, clacking, and it is all so big!
My intuition constantly suggests there is simply not enough time, but there is!
When I see one of these really active, and I know the museum is still ramping up nowhere near what active could and did once mean, I get the goosebumps being able to perceive some of what is happening, my minds eye filling the gaps with images of connections everywhere, the people oblivious to the massive coordinated activity enabling what they experience as ordinary life, in a sense a sort of matrix behind the scenes, literal rooms filled with this tech making amazing things happen.
I grew up rural and we had a simpler step by step servicing us. But being out in the sticks also meant being around a lot of old tech, relays, tubes (valves, which I find a charming name), motors, gears, chains and all manner of gadgets one could see, hear, feel, interact with.
Perhaps the best part all this pre transistor tech has in common is it all being at more or less at human scale and operating for the most part within human perception. Sure, we can't see the signal in a tube, but we can give it a tap and often hear that in the output.
And those fast clicks! An interested human can differentiate them, which is very cool, and yet and despite it, I suppose, there is ample time for the machine to do the job at a bit more than human speed, leaving a curious one just enough of a whiff to wonder at all that happened while they spin a dial, or mash on some buttons.
That tech is robust too. I used to place calls by manually delivering dialing pulses by slapping the on hook switch for fun! One can vary the rate of digit signaling as well as the time between digits and so long as the time between is not shorter than the pulse cadence it all seemed able to, and for me did, work just fine!
To me, that remains one of the more impressive aspects. People, when presented with a need or opportunity to input, do that and how they perform the input varies extremely widely. One can expect to see almost anything!
Despite that, these old machines just worked. Super impressive!
Thank you for a well produced video that answered old questions bouncing around in my head from childhood on!
Sadly, or maybe not depending on what gets produced in the future, I am left with more questions, a primary one being how the "map" of decisions is stored. Is there a part of this system that is a ROM, perhaps filled with physical connections that deliver information when addressed?
I would like to see that and it's role in all of this electro-mech magic!