gekk.info « articles

KSUless Phones

Listen to this article This article also exists in a video form, although the video contains less specific info than this article. You can watch it below:

Listen to this article: Listen to this article
(Article is not guaranteed to be identical to audio)

Story

A couple years ago I found a phone on eBay made by a "Northwestern Bell" called the Techline 420. Without going into the whole story, I bought it because it didn't make any sense.

This is a plain copper-line POTS phone, and because the POTS phone network has not changed in nearly 100 years, there are only a few "features" you can really have on them. Mute, hold, speakerphone, that's about it.

This phone made no sense because in addition to those features, it also had INTERCOM, TRANSFER, and CONFERENCE buttons, which are impossible to implement on POTS phones; there's only one signaling line, so you have nowhere to route audio for other calls. You can do this with PBX-based phones, because the PBX has its own switching circuitry and can route calls directly between extension lines, but this phone explicitly stated on the box that it required no PBX.

I couldn't find any further info on this specific model - it's almost nonexistent online. I figured that, whatever this phone was doing, it must not be unique, so I started searching on eBay to see if there were other phones with this capability, by looking for the word "transfer" within the "Telephones" (or whatever) category, then scrolling for brands that weren't known for PBX phones. Anything by Panasonic, NEC, Toshiba or other common PBX phone brands could be ignored - and sure enough, I found results.

I found phones by RCA, Uniden, GE, and others (see below) that all claimed to work on ordinary POTS line but had these same features. Armed with this info I went to my local ecycle store and searched for compatible phones, and found some. Specifically, the RCA 25424RE1, and about three of them.

These are ugly phones, clearly made without any love or pride, but I hooked them up as the manual recommended:
Two lines (they support up to four) with all three phones using the same line one. This is critical - they won't work if line 1 isn't the same on all phones, which obviously indicates it's doing some special signaling there.

Sure enough, they worked. I was able to transfer calls between the phones, do intercom calls, all-station paging, et cetera. I had some notions at this point of how this was possible, but I've only been able to confirm a few of them because my EE skills just are not entirely up to the task.

Theory of operation

Firstly, it goes without saying that the trick here has to do with multiplexing the line - literally no other explanation is possible, this is simply a given. The question I had was about the specifics.

There are a few ways to do this that I can think of, and my first theory was that it was leveraging DSL technology, which would be something like this:

<being wrong but thinking I'm smart voice>

Using a digital encoding like Discrete Multi-tone Transmission to convert digital bits into specific analog frequencies, set up a "bus" type network, like original Ethernet, where there's a single shared wire that all network participants can see at once

Since this is a voice line, the participants only pay attention to the high frequencies, above the human range of hearing

Within that virtual network, use packets to communicate control signals between phones, assigning each phone a random unique identifier on the network.

For voice signals, encode them into digital format, like G.711, and send them over the network.

Part of why I was thinking this was likely is because the manuals for the post-2000 phones all call out that DSL will interfere with their functionality, and that DSL filters, if inserted between the phones, will interfere with their functionality. Kinda seemed like maybe a digital signal was the answer - essentially, these would be self-configuring VoIP phones forming an ad-hoc network. Unfortunately it's entirely wrong, as far as I can tell. The phones I got are almost certainly not doing most of that.

Most of the other models I found are from the 90s, which puts this kind of tomfoolery a little out of their reach, I feel. I've found some examples that might be from the 80s, and I've even been told this technology existed in the 70s, though I have unfortunately lost the info on the model number. At any rate, this would be way too big a task for any of those eras.

More importantly, I put an oscilloscope on the model I actually have, and here is my analysis:

<being less wrong voice>

At rest, there does not appear to be any unusual signal on the line, and I feel like this precludes the existence of a digital network, which would probably have a continuous "carrier" of some kind - a periodic stream of packets to maintain bit synchronization.

During an intercom call, the audio is unquestionably amplitude modulated, to around 400KHz (in this example; ignore the freq counter on the scope.) So most likely the seperate audio channels are achieved trivially, by just modulating the speakers voice to above human hearing range.

During signaling events, like when I intercommed another phone but it had not picked up yet, I saw high frequency signals but couldn't make sense of them. Again, I'm just not that great at reverse engineering, but my guess is that it was some kind of rudimentary digital signaling - not a digital packet format, probably just some set of basic byte codes sent with something like FSK.

So I don't have all the answers on this - if anyone else happens to figure out any of the secrets of these things, let me know, I'll add your info here. I don't even know if my conclusions are accurate - what bugs me the most about this is just how completely undocumented this is.

Market history

I know a fair bit about telephones and their history, and I know a lot of people who know a lot more than me, and I haven't met anyone at all who knew that phones like this existed. There was apparently an entire market segment based around this technology, for decades - in fact, they're still sold. The AT&T 1040, 1050 and 1070 phones are fairly advanced versions of what looks like the same thing, and you can buy them at Best Buy right now for $100-180 or so.

It rankles me that there is a technology that is intriguing and clever, solves a real problem, could even be useful to consumers, but because it was targeted entirely at businesses, nobody ever cared enough to write anything down about it.

I did extensive research on this, which unfortunately has almost all been lost on account of some boneheaded data management I did this year, but the short of it is this: Nobody ever noticed these things. I was able to dredge up a few references in magazines but by and large it's just vapor.

There's an edition of something called Kiplinger's Personal Finances from 1989 (Google Books) that references it briefly.
Note that they use the term "KSUless" - KSU means "Key Service Unit" and is a term that goes back very, very far - like, to the 40s, where AT&T would install a piece of equipment - the Key System Unit - at your business in order to support multiline phones, and the resulting product was called a Key System. A KSU-less, of course, would be a key system without the key system unit.

I found an article in Cincinnati Business Courier from 1996 that describes them as a clever solution if they fit your needs.

I found a patent awarded to Nortel in 1995 for something similar, but actually using a digital packet network like I proposed - I can't find any evidence that this was ever put into production.

And that's about it. I really couldn't find much else. If you're not a Phone Person, you have to understand that this ought to have been a revolutionary product - people HATE owning and operating PBXes. They're expensive, complicated, frequently way too feature heavy; pharmacies with 4 phones within shouting distance of each other don't want to deal with PBXes. Trust me, I've worked in telecom, they hate it.

So you'd think, in that context, that someone would have been going nuts over this; that the manufacturers would be taking out full page ads proclaiming the freedom and simplicity and savings of using their products. But they didn't, and in fact most of the companies that made these almost seem... embarrassed about the technology?

I dredged up the websites of a bunch of manufacturers who were making these in the 90s and found that they either barely mentioned these capabilities, or described them with less exuberance than the basic, meaningless back-of-the-box bullet points. Like, every phone ever made advertises "clear, crisp voice quality" - you'd think these companies would have skipped that mealy mouthed bullshit and focused on the actual novelty of their products? But no. Nobody seemed to care, either on the manufacturing or the consumer end. It's baffling.

Advanced American Telephones

I do have kind of a theory about what's going on with all this, and it has to do with a company called Advanced American Telephones.

If you look at the back of the manual for some of these devices you'll find this name listed as a license holder. This company has a messy history. They started out as "American Bell Consumer Products," a spinoff from American Bell mandated by the Modification Of Final Judgment after the Bell breakup - American Bell itself was created as part of the original breakup some months earlier but I'll admit I don't fully understand how it relates.

AT&T had once made all the phones that their customers use but had been required for some time at before the breakup to let them use phones from other companies. They kept making and selling phones, however, and the courts declared that they needed to divest that consumer hardware division in the Modified Judgment, so we got this spinoff. They retained the Western Electric and AT&T brand names, so there was no noticeable change.

Eventually this company got absorbed into Lucent but kept doing the same thing they always had - making basic phones. Then they got partially absorbed by Philips, and when that deal fell apart, popped back to Lucent. Finally Lucent sold them off to VTech, where they remain today as VTech Innovation, but do business as Advanced American Telephones - and retain a license to the AT&T trademark, so all AT&T branded phones you see sold today are made by this company.

Phew. Anyway, here's why I think AAT is at the core of all this:

The companies that make most of these phones don't, to my eyes, seem like likely candidates for Innovation. RCA, Uniden, GE, Cortelco - by the 90s these were pale shadows of former greatness, at best. None of them gave a shit about their phone products, they're just mass produced, bottom-dollar garbage to a one; always were. These are not phone companies, they're just "electronics" manufacturers with no real place in the consumer market who make stuff to fill up store shelves. None of them invented this technology, I promise you.

However, since AAT was operating in the mid 80s with access to all of AT&Ts R&D, and it's been hinted to me that AT&T had one of the first phones that could do this, I suspect this is what happened:

In 1981-82, AT&T's hardware division came up with a brilliant solution to obviate the need to sell people complex, expensive PBXes. Then the breakup came and shook everything up. Instead of receiving the support of the parent company, suddenly they were on their own. AT&T had bigger problems than what telephone sets people were buying and wasn't interested in a huge marketing push

Left holding a fascinating bag, AAT presumably had no marketing division of their own, and simply integrated this technology into the ordinary, ho-hum plastic telephones that were their bread and butter, not particularly calling any attention to the fact that they were revolutionary; not taking out full-page ads; not getting reviews in consumer electronics magazines. Just making the product available and hoping for the best.

Now, this is pure speculation - I can't find any info, so what else can I do? But what sort of contradicts it, I think, is that they don't appear to have ever stopped making them. I can't find much info on the 80s models, but I think they've been selling these phones from the early 90s to present without any breaks. Maybe they were doing direct sales, or maybe I've just managed to miss all their advertising? Who knows.

At any rate, I do think that to eke some more profit out of the technology they started licensing it out. I base this on the fact that most of the phones I've found have very similar features, extension limits, etc. - they just look very much like the same thing with a different plastic case. In some cases I found the exact same model sold under three different brands, none of which are known for their telephone products.

I'll list everything I found about those models below.

Detailed information

All of the models in this list are devices I was able to find info about online, and usually at least one specimen on eBay, so if you want to investigate this further I recommend starting here, but make sure you read further for the PDF with the full list.

MakeModels
AT&T1040, 1070, 1080, 954, 854, 964, 874
Panasonic KX-TS4300, KX-TS4200, KX-TS4100
Bell SonecorBE-5300BLF*
Northwestern BellIntegra Digital 28, Varicom 28, Techline 420
RCA 25414RE3-A, 25424RE1 (also sold as ViSys)
Siemens Gigaset 8825
Uniden/SBC/ITC Intellitouch 420, 410
GE 29487 GE2-A
VerizonHAC NSQ412
(also sold as Qwest, inexplicably sold under an unknown Landmark brand; actually made by International Resources Inc)
Cortelco2740, 2742, 2743
TT SystemsSimtel 210, 410

* This page, from a past member of the W3C, says the Sonecor phones were actually made by Phonemate - not sure where that info comes from, but if we look up "Phonemate 4-line" we find that it's a Casio brand, and sure enough, it's *another* KSUless phone. I then found another page listing this as an "ITT Cortelco" Phonemate, and when I open the brochure it's listed as a TT Systems PhoneMate. The actual Sonecor, by the way, was later acquired by SBC. It's all a god damn mess.

The Skutch adapter

Hey: Skutch was a remarkable company (after a fashion) that is now out of business. Consider reading about their other, ridiculous products.

The most useful resource I found in all this was a doc called "Master Compatibility List, a cross reference that gave me info on even more models. This document lists supported hardware for a music on hold device made by the concerningly named "Skutch Electronics." It lists a ton of phones including several brands I hadn't heard of. There's a lot to address here.

First, this confirms that this industry was substantial - there are hundreds of models here, from 27 manufacturers, making it even weirder that nobody has ever heard of any of them.

Second, this strongly supports my theory that these are all using the same tech - probably a single chip, made by AAT and sold by the tube - because the same Skutch MOH adapter works on wildly different models. Almost teh entire AT&T lineup prior to the 1000 series works with the AS-702-B - maybe it speaks different languages, but more likely all these phones are using the exact same signaling.

What's peculiar about this is that, per the documentation, that compatibility doesn't exist at the phone level. AT&T warns in their docs that the 800, 600, and 900 series (I think) are all incompatible. But what I think is more likely than protocol incompatibility is just that the protocol contains an identifier - "Hey, I'm a compatible AT&T 900 phone" - and they chose to change it between series', while Skutch's MOH box shrugs and ignores that field completely.

Third, this supports my theory that a lot of these are just rebrands. I tried looking up a bunch, and in a number of cases (sorry, I didn't write them down) the phones are identical with different names on them.

Regarding the Skutch unit itself - spparently the purpose of this instrument was to provide music on hold, so presumably the way it works is that it watches the line for the "i just put line 1 on hold" signal, and then connects itself and starts playing MOH until it sees the "i'm taking line 1 off hold" message. Assuming the signal protocol really is mostly the same between models, that would make this a pretty simple device as things go, so I can see why it would support so many different phones.

The Cortelco phones

The Cortelco entries are particularly interesting. Cortelco was originally a company called Kellogg Switchboard & Supply Company that dates back to 1897, they've variously been Lucent and ITT over the years, but are now Cortelco, manufacturers of some of the cheapest looking worse-than-Radio Shack-grade phones I've ever seen. Really, their phones look awful. Nonetheless, they produce some *fascinating* adjuncts to this ecosystem.

The Cortelco 2742 is what in VoIP terms we'd call an ATA - it allows you to connect ordinary POTS telephones, which is very strange to contemplate. Ordinary POTS phones could just be plugged into the same phone lines the other phones use and just work, but this adapter considerably upgrades them.

When you plug a POTS phone into this and pick up, you get - presumably - a synthesized dialtone, not the tone from the line. At this point you can dial another station, and it'll make the call over the internal modulated channel - so you can pick up your cheap radio shack phone and just dial 11, and make a call straight to a Cortelco KSUless phone.

To connect to a real line, you can just dial 9 for the first available, or 81, 82, 83, 84 to access specific lines, so I guess you could use that to barge in on active calls, maybe? You can also do in-network call holds with a hookflash, and follow it by dialing another station to do a transfer, among many other things.

Essentially this thing is a cheap little PBX, but instead of using T1 or PRI trunk lines, it uses the proprietary internal network generated by this wacky KSUless technology. Wild!

The Cortelco 2743 is a multifunction device which can provide hold music and overhead paging, which is pretty self explanatory, but what's intriguing is that it really rounds out the system, making it a fully fledged PBX-style telephone network, yet none of the other systems that I've seen offer anything like that.

I'm sure this is all based on the same technology used in all these other phones, but I had assumed that it was all very rigid - that Advanced American Telephones had produced some kind of ASIC, or at least a simple ROM mask for a microprocessor or something along those lines, and was just selling the completed system to all these companies to build a plastic chassis around. But the addition of these adjuncts makes me wonder if there isn't more to it - maybe these devices are fully programmable, the digital network is a totally general-purpose system and any functionality can be implemented in the firmware? Maybe Cortelco isn't using the commodity chips, but licensed the tech and then designed their own silicon to implement it? There's no way to know without spending thousands of dollars to buy and disassemble these things for comparisons.

The AT&T 1000 series

The AT&T 1080 has what I think is the most interesting functionality of all. It offers voicemail and an auto attendant. The former seems dull - $30 cordless phone sets have voicemail - except that I think that it has integration to the other phones, e.g. the other phones are smart enough to call whichever 1080 is acting as the "master" set in order to get their voicemail. Don't quote me on that part though, the manual is... dense.

The auto attendant is what's more interesting to me. When you call in, a 1080 with an AA will immediately pick up and offer a menu (with a custom greeting, of course), at which point the caller can enter an extension to be transferred to or, I believe, be transferred to common or extension-specific voicemail.

I'd guess that the transfer mechanism works just like a normal transfer - it tells another extension "pick up line 2" and when that phone picks up, the first extension releases it. But the limitations are kind of curious - the AA can't function if any line on the phone is in use, or if an intercom call is in progress. The former I guess I get - the phone doesn't have the processing power to handle a phone call at the same time as an AA call? And the latter, I suppose, suggests that control signals can't be sent when an intercom call is in progress. I wish I knew more about the protocols underpinning these things.

Anyway, as a result, the system supports multiple redundant AAs - like, if you have 10 phones, you put the same AA on each, set them all up with incrementally higher answer delays, and then when someone calls in, the first phone that's not in use grabs the call. This is, I admit, where the cleverness of the system gives way a bit to the realization that there are drawbacks inherent in this approach, but it probably works well enough for a small business.

Alright, that's all I know! Thanks for reading! Once again, if you learn more or know more, email me and I'd be happy to add it here.


List of Articles