Questions about old-school phone tones & rings around the world

[globoy]

I'm making a gadget so that people can connect old POTS phones to things like their cellphone, or SIP-based VOIP or even to be a prop in a museum or art installation.  It'll be open-source, etc, and I might sell a board on tindie if there is interest.  One thing I want it to be able to do is generate some of the various call-progress tones and ring patterns in country-specific ways.  The problem is that I can find a lot of information online but not all.  Currently my code can support the following items (specified in North American parlance):

• Dial Tone - The tone you get when you pick up the phone (e.g. in NA it's 350 + 440 Hz)
• Re-order Tone - The tone you get when you can't make a call (aka congestion tone, fast-busy).  I use it when there's no Bluetooth connection for example.  (In NA it's 480 + 620 Hz, 250 mSec on/250 mSec off)
• Off-hook Tone - The tone you get when you forget to hang up.  Generally much louder to get your attention.  This is the main one I have trouble finding for various countries.  (In NA it is 1400 + 2060 + 2450 + 2500 Hz, 100 mSec on/100 mSec off)
• Ring Frequency/pattern - The frequency of the signal sent to ring the bell (typ 20 - 25 Hz) and any pattern (in NA it's 20 Hz, 2000 mSec/ring - in the UK it's 25 Hz, 2 times of 400 mSec on/200 mSec off per ring)

It also supports DTMF but that seems like a standard used everywhere.

EEVblog is a pretty international place with a lot of experienced - e.g. old like me  - technical people.  I'm wondering if any of you know any of this information for your country?  For example for you Aussies, I know your dial tone (at least one of the main ones since it seems like your old phone systems had several) is 425 Hz.  But what were the other tones?  What did your ring sound like?

Some pictures of my proto below if you're interested.  The project is called "weeBell" as a play on both wireless and small, and paying homage to the granddaddy of all American tech companies "Ma Bell".  The history of the phone system is fascinating to me and I think it's wild that certain parts of what I am implementing today hail back to devices that existed over 130 years ago (for example the ring frequencies/voltages and the tip and ring signals).

[Andy Watson]

Interesting project.

Not quite what you were asking for, but:
https://www.britishtelephones.com/teletone.htm
It's a bit local and mostly before my time

[boB]

Check out Evan Doorbell's  YT page for many sounds of the old phone system with explanations.

boB

https://www.youtube.com/@evandoorbell4278

[planet12]

You'll find a lot of information for this in the Asterisk PBX source code, as it uses it to generate country-specific tones.

See: https://github.com/asterisk/asterisk/blob/master/configs/samples/indications.conf.sample

An example from NZ:

ring = 400+450/400,0/200,400+450/400,0/2000

This is the "ringback" tone - combined 400Hz + 450Hz for 400ms, silence for 200ms, 400Hz + 450Hz for 400ms, silence for 2000ms.

This config file doesn't include the POTS bell ringing frequency, just the cadence - however most country sections have a link to the specific standards document which should have it (or help you find it).

[globoy]

@planet12 - thank you!  Of course!  Asterisk.  I will definitely take a look.

[I wanted a rude username]

Inferring from the photo that it'll support pulse dialling? Cool feature if so ... few ATAs support it.

[CaptDon]

On party lines here in the U.S. they had different mechanically resonant ringers each responding to a different 90vac ringer frequency to ring the correct phone on the party line. Turns out my friends ringer was resonant at 25Hz which also happened to be the frequency of the overhead catenary 11,000vac 25Hz powering the GG1 locomotives along the Susquehanna River where he lived. Whenever a group of GG1's were pulling freight or passengers along the river route his phone would rattle with an occasional 'ding' on the ringer bell! I always loved how on a foggy rainy night the arcs from the pantographs would light up the sky!! You could tell and electric powered train was coming miles before it passed by!! It was a bit strange to see arcing at the loose joiner plates (fish plates) connecting the rails when a train was within a few hundred feet of the 'Impedance Blocks' that kept continuity for the power but isolated the wayside signals from block to block. Good times long gone now.

[thermistor-guy]

Australian tones, according to a Cisco document:

https://docstore.mik.ua/univercd/cc/td/doc/product/tel_pswt/vco_prod/austrasu/aus02.pdf

That seems to be right. I remember that dial tone was modulated. Back in the day, some cheap PABXs wouldn't generate it exactly,
and you could hear the difference from "real" dial-tone.

My first student job was in a telephone exchange, bottom floor had step-by-step equipment, top floor had crossbar equipment.
Whenever there was a controversial topic on talk-back radio, the exchange would be loud with the noise of switching equipment
- people trying to dial in, getting a busy signal, and trying again.

[globoy]

Thanks @thermistor-guy.  That is helpful.  The modulation makes for an interesting sound (which is part of the fun I think...all the different sounds people grew up with).  I'm curious: was Dial-A used much more than Dial-B?  Or at a different time in history?

@CaptDon - I didn't know they used different resonant frequencies for party lines back in the day.  Thanks for the story!

@I want a rude username - yeah, supporting rotary phones is definitely a part of this. Actually it's not hard as the AG1171 SLIC I'm using provides the off-hook signal and it's just a matter of a state machine that times it at the two levels.  Interesting note is that some rotary phones around the works swapped positions of the 1-9 digits on the dial ('9' was in the upper right position).

@boB and Andy Watson - both fascinating websites.  Thank you. 

[boB]

Also,  try

phonetrips.com

boB

[thermistor-guy]

... I'm curious: was Dial-A used much more than Dial-B?...

Yes, Dial-A was the standard for Australian telephone exchanges. The national carrier was strict about it,

But standards may be more relaxed now. Have a look at NBN document:
    https://www.nbnco.com.au/content/dam/nbnco/documents/sfaa-wba-uni-v-electrical-specification-r5-%2020130702-to-20130828.pdf

which refers to AS/CA S002:2010:
    https://www.commsalliance.com.au/__data/assets/pdf_file/0006/1122/S002_2010.pdf

Personally I would prefer the modulated tones as per the Cisco document above.

[Neomys Sapiens]

The German standard was called 1TR110-1.

Basically
ready to dial  425Hz constant tone
calling free subscriber 425Hz 1s tone 4s pause
occupied subscriber  425Hz 480ms tone 480ms pause
no free channel 425Hz 240ms tone 240ms pause

[I wanted a rude username]

I remember that dial tone was modulated. Back in the day, some cheap PABXs wouldn't generate it exactly,
and you could hear the difference from "real" dial-tone.

Asterisk's trick of combining a 413 Hz and a 438 Hz tone to generate an approximation of the 425 Hz tone modulated at 25 Hz was a good hack and almost sounded right ... the genuine dial tone had a softer quality to it, perhaps in part due to the various distortions of the POTS. The "warbling" Australian ready and progress tones are beautiful.

If the PBX couldn't generate two tones then it would have sounded glaringly wrong. The first time I called an overseas number and heard the flat, unmodulated progress tone, I thought the line was busy ...

[globoy]

Thanks for the German info @Neomys Sapiens.

And thanks again to everyone.  This is been very interesting and makes me realize that I will - at least eventually - need to create a much more sophisticated way to specify and implement tone generation.  And that ring frequency is important because the phones will probably have that as a resonant frequency.

Initially I'm going to cheat, and as IWARU said, create a hopefully familiar enough but not right Australian tone alternating two frequencies.

I'm now sure I'll never get it 100% right.  Never going to be able to duplicate the huge work done by all the telephone companies...

[elecdonia]

On party lines in U.S. they had different mechanically resonant ringers each responding to a different 90vac ringer frequency to ring the correct phone on the party line. My friends ringer was resonant at 25Hz, the frequency of the overhead catenary 11,000vac 25Hz powering GG1 locomotives along Susquehanna River where he lived.

Too bad there isn't a way to run a GG1 locomotive these days! Several GG1 locomotives still exist, but none can move under their own power. In their heyday more than 100 GG1 locomotives were in operation.

25Hz overhead catenary is still in use by some electric rail systems located in the northeast United States:
     https://en.wikipedia.org/wiki/Amtrak%27s_25_Hz_traction_power_system
But the trend is to replace them with systems operating at 50/60Hz.

All of the GG1 locomotives were retired from service between 1950 and 1983:

https://www.historictrains.org/collection/hank-27nex#:~:text=Pennsylvania%20Railroad%20GG%2D1%20Electric%20Locomotive&text=After%20more%20than%2050%20years,16%20remaining%20in%20museums%20today.

25Hz overhead power worked well with conventional locomotive traction motors with brushes and commutators. Most of them operated on 600V DC or on low-frequency (25Hz) AC. The GG1 locomotive didn't have rectifiers. Their traction motors ran on 25Hz power. The mercury vapor rectifier tubes available 90 years ago were large, delicate, and incapable of supplying the high currents required by a heavy locomotive. With that said, mercury vapor rectifiers were used for light-rail applications. But they were located inside stationary trackside AC/DC substations which supplied 600V DC to the overhead trolley.

Large DC traction motors cannot operate efficiently on 60 Hz AC because their field impedance at 60Hz is too high to pass the required field currents. Also the eddy current losses are huge at 60Hz. But 25Hz was a sufficiently low frequency for existing traction motors to be adapted to handle 25Hz without difficulty.

Today's electric locomotives have highly efficient solid-state rectifiers. Their DC output is fed to standard brush/commutator DC traction motors or to variable frequency solid-state AC drives feeding AC traction motors.

[thermistor-guy]

quote author=globoy link=topic=376715.msg4861214#msg4861214 date=1683994587]
...This is been very interesting and makes me realize that I will - at least eventually - need to create a much more sophisticated way to specify and implement tone generation ...
[/quote]

Well, looking at your h/w block diagram, you have everything you need right now:

* store the progress tone PCM samples in arrays, in EPROM;
* output a sample to the codec every 125 us;
* use pointers to track the current sample, start and end of the selected progress tone array.

When I did this for a professional product (long ago), for use in a primary rate multiplexer, the microcontrollers that were cheap enough
for this purpose weren't fast enough. So I designed a multi-channel DMA engine using small cheap PLDs.

Every time slot, the engine would fetch a sample of the desired progress tone from EPROM, and output it into its associated PCM channel.
The multiplexer would route that progress tone channel to the desired customer channels as needed.

There was one special "progress tone" that was a Recorded Voice Announcement, which customers received when the mux was only
accepting emergency calls. So the mechanism was quite flexible.

You can do this with software. EPROM space permitting, and with today's processors, you can implement a very flexible, single-channel
tone generation with your existing architecture.

[Njk]

I'm now sure I'll never get it 100% right.  Never going to be able to duplicate the huge work done by all the telephone companies...

The wired stuff is not in big fashion at the moment but that's an advantage, because you don't have to participate in fierce competition and can afford a relaxed schedule. In the past I'd designed a series of similar adapters for various applications. Let me say the two things.

First, you don't have to duplicate the huge work. By now, all you need to know is how many tones and cadences you're going to support. To allocate enough memory for the description table. You can update this table from computer any time later. As for generation, a good SLIC device must have programmable generators for exactly that purpose. With appropriate HW support, it's not a rocket science to generate a tone. All my designs were based on the parts from Silabs. At the time, it was the best vendor in the field, and some senior executives were previously our employee, so we were able to negotiate a good price preference. Now it's more of a grab-and-hold type corporation by back at that time they just moved to the nice new corporate building in Austin at the Colorado river embarkment. No idea if they're still making a SLIC parts (Si3210, Si3216, etc.). The other vendor at that time was Legerity. Again, I'm not aware of the current state of the art, just a suggestion.

Second, the tones are not the only headache. Nothing wrong will happen if only one country standard will be supported. But if you'd like to build a true thing, don't forget about the caller ID protocols. Many telephone devices had a screen for displaying that information. Similar to the tones, a number of different signaling methods existed. They're standardized, but it's the worst standardization example. Instead of fighting against all but one well-established interest groups, IEC just rubber-stamped all existing solutions (similar to what is done to the electric vehicle charging connectors). So you will have to implement all of them (5 or 6, IIRC), because rare specification for telephone device mentions the caller ID protocol type.

[globoy]

Thanks @Njk and @thermistor-guy.  Good input and appreciated.

I happened across this incredible open-source C library called spandsp (https://github.com/svn2github/oslec) written in the first part of this century (I think mostly used by Asterisk) that provides all kinds of incredible functionality like line echo cancellation, DDS (sine-wave) tone generation, DTMF tone generation and decoding and even the modem functionality required by caller ID (which I'd like to add later).  Right now the tone system uses the DDS engine so it can be configured with a table of frequencies/cadences.  But I'll probably do what you suggest and add another mechanism to index sound samples in flash for those tones like the Australian and UK tones that are modulated.

CPUs were slower when spandsp was written and the code has big sections optimized for both a Blackfin DSP and x86 MMX assembly instructions.  Today a 240 MHz LX6 core in the ESP32 is able to easily handle the audio tasks just running the straight C. 

This project is being done for the fun of it and I might make a dollar or two selling boards (but they are never a profitable enterprise - they're just a way to share - and for me as a contract designer, a form of marketing).  So I don't feel terrible if I can't 100% match the capability of the original exchange but it's a fun goal - and educational - to try to match as much as possible.  Even today it feels exotic to hear the dial tones and rings that I first encountered on a trip abroad as a kid. 

[globoy]

Second, the tones are not the only headache. Nothing wrong will happen if only one country standard will be supported. But if you'd like to build a true thing, don't forget about the caller ID protocols. Many telephone devices had a screen for displaying that information. Similar to the tones, a number of different signaling methods existed. They're standardized, but it's the worst standardization example. Instead of fighting against all but one well-established interest groups, IEC just rubber-stamped all existing solutions (similar to what is done to the electric vehicle charging connectors). So you will have to implement all of them (5 or 6, IIRC), because rare specification for telephone device mentions the caller ID protocol type.

LOL, Yes(!) I saw that document.  Before the first ring, after the first ring, DTMF, seize the line, qualification tones, etc, etc.  I think some countries even had multiple implementations.  Telephony was the wild west.  Some shared underlying ideas but then everyone went and did their own thing.

[globoy]

Dave's post about paid promotions made me think about how badly I self-promote so here's a bit of that.  For anyone who wants to follow my progress on this gadget: https://hackaday.io/project/191002-weebell-personal-central-office-for-pots-phones

I have added the ability to playback tones as well as generate them using DDS so I have implemented both the Australian modulated dial tone and the UK off-hook howler.

The basic functionality is done in code and I am - sadly - waiting for Rev 3 of the PCB (I2C fun...).  I am still researching some countries to add and in a future release I think I can add support for caller ID.

[TimFox]

On party lines in U.S. they had different mechanically resonant ringers each responding to a different 90vac ringer frequency to ring the correct phone on the party line. My friends ringer was resonant at 25Hz, the frequency of the overhead catenary 11,000vac 25Hz powering GG1 locomotives along Susquehanna River where he lived.

Too bad there isn't a way to run a GG1 locomotive these days! Several GG1 locomotives still exist, but none can move under their own power. In their heyday more than 100 GG1 locomotives were in operation.

25Hz overhead catenary is still in use by some electric rail systems located in the northeast United States:
     https://en.wikipedia.org/wiki/Amtrak%27s_25_Hz_traction_power_system
But the trend is to replace them with systems operating at 50/60Hz.

All of the GG1 locomotives were retired from service between 1950 and 1983:

https://www.historictrains.org/collection/hank-27nex#:~:text=Pennsylvania%20Railroad%20GG%2D1%20Electric%20Locomotive&text=After%20more%20than%2050%20years,16%20remaining%20in%20museums%20today.

25Hz overhead power worked well with conventional locomotive traction motors with brushes and commutators. Most of them operated on 600V DC or on low-frequency (25Hz) AC. The GG1 locomotive didn't have rectifiers. Their traction motors ran on 25Hz power. The mercury vapor rectifier tubes available 90 years ago were large, delicate, and incapable of supplying the high currents required by a heavy locomotive. With that said, mercury vapor rectifiers were used for light-rail applications. But they were located inside stationary trackside AC/DC substations which supplied 600V DC to the overhead trolley.

Large DC traction motors cannot operate efficiently on 60 Hz AC because their field impedance at 60Hz is too high to pass the required field currents. Also the eddy current losses are huge at 60Hz. But 25Hz was a sufficiently low frequency for existing traction motors to be adapted to handle 25Hz without difficulty.

Today's electric locomotives have highly efficient solid-state rectifiers. Their DC output is fed to standard brush/commutator DC traction motors or to variable frequency solid-state AC drives feeding AC traction motors.

The GG1s would have been retired earlier due to their age and worn castings, but the replacements didn't work well until 1983.  I was in one at a museum in Roanoke, Virginia around that time and was surprised at how cramped the operator's cabin was for such a huge vehicle.  In Europe, a similar compromise between transformer size and use with DC commutator motors was done with 16-2/3 Hz (1/3 commercial line frequency).

[MathWizard]

I'm trying to remember the sound of the ringer on the old rotary phone in my house as a kid. I can't tho, but I remember the rotary sound pretty good.