I wanted to understand the audio signals used by a Textphone's modems plugged into a telephone line. I could record the audio using Audacity, which could display spectrograms. Audacity can export samples. These can be imported into a webpage that could use JavaScript to decode the modem signals and decode the Text.
* Webaudio could be used to play generated samples * HTML <CANVAS> and JavaScript could plot debug waveforms * WebApi-serial can be used to control the modem plugged in using a serial port. * These web tools do not need approval or difficult to purchase software. * The wiki and webpages can document this interesting subject that span many decades.
To explore a problem, I can use use webpages, and javascript to experiment and explore. I can make notes in the code or webpage text. This is a scrap book of notes and links.
The webpage is intended mostly for the writer. It could be that the only other readers are random AI tools and chat bots.
It is possible to do Audio in a Webpage.
See: https://developer.mozilla.org/en-US/docs/Web/API/Web_Audio_API
Web pages now support a large range of WebAPI, two of which are <CANVAS> for graphics and Audio.
https://ltu.diva-portal.org/smash/get/diva2:1030584/FULLTEXT01 - A very Interesting Document about Modems that is worth a read.
It is now possible to use a web page to generate and play audio samples and plot graphs using reasonable accessible tools available in a web browser.
http://www.dougrice.plus.com/dev/UART/xyz.htm tries out some ways to generate sine waves.
It provides a way to experiment with digital filters.
When debugging DSP that has a stereo DtoA, it is possible to use one DtoA channel to provide a debug waveform to annotate the other signal.
I can use JavaScript to do the maths and plot a graph.
A couple of lines are used as debug.
The state of the UART and sampler can be shown.
It can show when a waveform is sampled.
IIR poles and zeros - explores some ideas about IIR in the feedback.
Other Audio Experiments - Other experiments and links.
This web page started as an experiment and evolved. I wanted to understand the signalling between modems. Audacity is very good. It can record signals and show the spectrum.
The ITU V.18 modem specs for the Textphone have lots of facts to understand. Audacity cannot decode DTMF and modem squarks, so this webpage records my attempts.
Two-wire telephones summary:-
More diagrams:-
[functional blocks in a phone] has some useful dialgrams for this webpage.
Using a simple wiretap, record the telephone signal.
IMPORTANT: The two LED snub any voltage spikes when you go onhook / offhook. A cheap USB soundcard was used incase the LEDs did not snub and to provide a 3.5mm socket. This is plugged into a laptop running on batteries. We are interested in timing and frequencies.
Export the Audio samples, sampled at mono 8000 samples per second into a text file. Copy and paste them into a textarea on the webpages below.
These pages uses IIR Goetzel resonators / filters to look for the modem's FSK tones that encode the 1's and 0's. These are decoded by an adjustable 300 baud UART to decode text that Audactity cannot decode.
![[Home]](../ipstats.gif)
[samples]--[ notch filter ]--+--[ iir filter ]--[ mark energy ]--+--[ comparitor ]--[ UART ]--[code]
\--[ iir filter ]--[ space energy ]--/
http://www.dougrice.plus.com/dev/UART/UART2_20230423_ProofOfConcept.htm - Early Proof of concept to see if I could decode the bits Audacity could not decode.
http://www.dougrice.plus.com/dev/UART/UART2_V21.htm - Try and decode the capture of a V.21 modem and other modems and generate tones.
http://www.dougrice.plus.com/dev/UART/UART2_V18.htm - Try and decode the capture of a V.18 modem and other modems and generate tones for different modems used in ITU V.18.
http://www.dougrice.plus.com/dev/UART/UART2_BAUDOT.htm - This page uses a UART configured for BAUDOT that uses 5 bits and does a crude BAUDOT to ASCII
http://www.dougrice.plus.com/dev/UART/UART2_DTMF.htm can be used to decode DTMF. This uses 8 IIR filters.
Audacity was used to capture audio of a modem training up on a telephone line.
https://www.audacityteam.org/ has been useful to analyse audio on phone lines
It has found problems, that a scope would not find. Screen shots provide a useful way to summarise protocols that may take a few seconds to transact.
The picture below show Audacity displaying the spectrogram of the capture when a V.21 modem answers an incoming call.
The ringing, 2100Hz ANS, V.21(high) 1650Hz Carrier , V.21(low) 970Hz carrier and burst of data are seen.
Audacity cannot decode DTMF or the Modem FSK signals used by V.21. I cannot work out how to import a WAV file into these webpages and get the samples, but it is possible to export the samples to a text file.
Select all and copy and past the samples into one of the web pages below. They use JavaScript to analysis it. The CANVAS can be used to plot waveforms and debug waveforms.
Audacity's Analyse spectrum and Spectrograms can establish what frequencies are being used, and sections of interest.
The decoded text can be used to add Labels to annotate the Audacity capture.
Typically the Modem takes a serial signal, using voltages encoding the "1"'s and "0" using MARK and SPACE and uses FSK frequency Shift keying.
[computer: [computer program]--[UART]--]--[RS232 cable]--[ DB25 connector ]--[modem]--[telephone system]--[modem]--[UART]--[computer program]
or:-
[computer program]--[USB modem: --[UART]--[modem]--]--[telephone system]--[modem]--[UART]--[computer program]
These are very useful diagrams from https://docs.rs-online.com/a02c/0900766b800af005.pdf - data sheet for V21 modem.
It is quite simple to analyse the samples exported from Audacity using JavaScript, using code run for each sample. The pages can be updated as required for the task in hand.
The samples are passed into IIR filters to detect the V.21 FSK signals.
Some logic is needed to decide if the FSK is encoding a "1" or "0".
Logic is used to find the "1"'s MARK and "0"'s SPACE. These are fed into the UART. It looks for the start bit and then clocks 5 bits for BAUDOT or 8 bits for V.21.
Countdown State machines are used for the FSK logic, UART and DEBUG.
The UART emits characters. Parity may need to be stripped, and the decoded text can be displayed.
You can view and study the source code on this early proof of concept page:
http://www.dougrice.plus.com/dev/UART/UART2_20230423_ProofOfConcept.htm - Early Proof of concept to see if I could decode the bits Audacity could not decode.
These pages were expanded to add more features:-
http://www.dougrice.plus.com/dev/UART/UART2_V21.htm - Try and decode the capture of a V.21 modem, and now other modems, and generate modem signals and play them using Web Audio.
http://www.dougrice.plus.com/dev/UART/UART2_V18.htm - Try and decode the capture of a V.18 modem and other modems and generate tones for different modems used in ITU V.18.
http://www.dougrice.plus.com/dev/UART/UART2_BAUDOT.htm - This page uses a UART configured for BAUDOT that uses 5 bits and does a crude BAUDOT to ASCII
http://www.dougrice.plus.com/dev/UART/UART2_DTMF.htm can be used to decode DTMF.
ITU V.21 used FSK with 1070 +/- 100Hz and 1750 +/-100Hz, or 970,1170, 1650, 1850 Hz. Some modems use bandpass filters.
It is possible to filter out the V.21(high) 1650Hz or V.21(l) 970Hz using Audacity.
This Filter designer: http://jaggedplanet.com/iir/iir-explorer.asp was used to design a Butterworth notch filter
A modem using the ITU V.8 spec, has other interesting bursts of characters to decode as seen in the picture above. Other pages can decode DTMF and BAUDOT captured using Audacity.
Modems use V.21 as part of the ITU V.8 protocols and it is unlikely that the USB modem can output these.
However by copying the recorded samples into my webpage, it is possible to decode the ITU V.8 CI messages and TXP sent by some modems.
ITU V.8 CI signal captured and decoded. A A A A A 0x41 0xff 0x0 0x41 0xff 0x0 0x41 0xff 0x0 0x41 0xff 0x0 0x41 0xff 0x0
Where the V.21 signal had both carriers, Audacity could be used to filter out one carrier.
This webpage http://www.dougrice.plus.com/dev/UART/UART2_20230423_ProofOfConcept.htm is an early Proof of concept from 23/04/2023 to see if I could decode the bits Audacity could not decode.
The page uses IIR Goetzel resonators / filters to look for the modem's FSK tones that encode the 1's and 0's. These are decoded by an adjustable 300 baud UART to decode text that Audactity cannot decode.
[samples]--+--[ iir filter ]--[ mark energy ]--+--[ comparitor ]--[ UART ]--[code]
\--[ iir filter ]--[ space energy ]--/
ITU V.18 3.3 says "A CI sequence consists of 10 ONEs followed by 10 synchronization bits and the call function octet."
Table 1/V.8 – Preamble - "Ten ONEs preceding each information sequence , Synchronization for CI sequences 0x00,
The call function octet 0x41 is found in:-
Table 2/V.8 – Information categories
Table 3/V.8 – The call function category
My first pages included the samples in a textarea or between DIV tags but the files got very big to compare using notepad++ compare so I wrote a cscript program to split the files.
https://homepages.plus.net/dougrice/dev/UART/jsUARTsplit.js.txt
My pages now generate test signals to decode to make the download smaller. These can also be played using WebAudio over the sound card , speaker or headset and played into the modem.
V.18 Annex B DTMF is a way of sending text using DTMF characters.
This link tried to document the names of the signals on the RS-232 cables. The names date back to the early 1960's when computers and modems were in seperate boxes.
http://ccgi.dougrice.plus.com/cgi-bin/wiki.pl?Modems
This link has some interesting early history, which explains the confusing names used for signals and abreviations.
https://www.auvik.com/franklyit/blog/history-of-the-modem/
Early modems for V.21 had a RS232 interface and loads of signals defines in the RS232 and V.24 specs. ITU V.25 and ITU V.8 describe calling and answering tones used to establish connections.
ITU V.25 is worth a read as it specifies how a modem sends calling tone and answer tone. Modems work over a variety of telephone circuits and private wires with many ways of signalling. A modem needs a simple calling and answering tone signalling.
More modern modems may have a USB plug and use the AT commands.
WebAPI allows a webpage to be used to interface to the modems
http://ccgi.dougrice.plus.com/cgi-bin/wiki.pl?AT_Commands_-_Serial_Using_Web_Browser
ITU V.24 "List of definitions for interchange circuits between data terminal equipment (DTE) and data circuit-terminating equipment (DCE)"
The C in DCE is "Circuit-terminating" not "computer" - easy to be confused. -
https://www.analog.com/en/resources/technical-articles/fundamentals-of-rs232-serial-communications.html is well worth a read.
https://pinoutdiagrams.com/cable/rs232-null-modem-cable-pinout
When I had my first PC about the the time of the IBM XT PC, we used serial ports and RS232 leads. I did not have a dial-up modem. The RS232 leads could be tricky to get to work. I had a Ball Point Pen test, short pin 2 and pin 3 and see if you got an echo when using Putty or HyperTerm.
At the time, I never really understood DTR , DSR, DTR, RTS, CTS in the modem context and without the modem, it is easy to get confused. DCE does not mean "Data Computer equipment", it is short for "data circuit-terminating equipment" (DCE).
[ computer running Terminal software ]--[ RS232 lead ]---[ modem ]--[ Telephone line ]--[ modem ]--[ computer ]
DTE ->|-> DCE
We used:-
[ computer running Terminal software ]--[ RS232 lead wired as null modem ]---[ computer ]
DTE ->|-> DCE
People used to have a box to fake the signals to get RS-232 leads to work, and there are various null-modem wiring diagrams.
I suppose the computer program had to wait until these signals indicated the computer at the other end was connected.
RS232 is very much alive, but more at the API interface than the DB25 cable.
ITU V.24 defines the circuits. The diagrams below are useful summaries.
This document has a very useful summary.
See the Cable and Adapters diagrams. The Minimum Straigt Cable diagram explains how the signals are grouped.
Normal timing sequences during establishment of communications are shown below.
On half duplex circuits, RTS is dropped as soon as the data is sent. This is to signal a turn around of the circuit.
DTR > _ _ - - - - - - - - - - - - - - DSR < _ _ _ - - - - - - - - - - - - - RTS > _ _ _ _ - - - - - - - - _ _ _ _ CTS < _ _ _ _ _ - - - - - - - - _ _ _ TD > _ _ _ _ _ _ - - - - _ _ _ _ _ _
CD < from modem indicating Carrier is being received.
So this should be used to hold back buffered user typing.
If characters are received before CD is asserted, It could be the V.18 CI.
Check if there is a frame error or parity error. Check if they are spaced correctly.
Microchip 16F877 stop bits can be used for 9th bit Address.
Some UARTS lock if there is an overrun error and need to be reset. e.g. Microchip 16F877
See the Cable and Adapters diagrams.
The API used still hang onto the RS232 signals.
Some API documents for windows and Linux.
Using Web_serial to do RS232 to a Modem.
https://developer.mozilla.org/en-US/docs/Web/API/Web_Serial_API
http://ccgi.dougrice.plus.com/cgi-bin/wiki.pl?AT_Commands_-_Serial_Using_Web_Browser
Typically the Modem takes a serial signal, using voltages encoding the "1"'s and "0" using MARK and SPACE and uses FSK frequency Shift keying.
[computer program]--[UART]--[modem]--[telephone system]--[modem]--[UART]--[computer c program]
The UART interface to the computer program is summarised as a TX and RX register and two flags TxEmpty and RxFull.
CPU - UART:-
A write to TX resets TxEmpty. A read from RX resets RxFull.
UART - Host:-
A read from TX sets TxEmpty. A write to RX sets RxFull.
If you have a FIFO then there need to be changes.
TxEmpty may need to be TxSpaceAvailable RxFull becomes RxAvailable
I have a JavaScript page that explores this: http://www.dougrice.plus.com/dev/DigiSpark/text0.html
Between the UART and the C program is a lot of API.
https://developer.mozilla.org/en-US/docs/Web/API/Web_Serial_API
http://www.dougrice.plus.com/dev/serial/rpi_ser_tcc.c
And these dabbles to try driving an old modem over a serial port. The Webpage does not always see the USB modems, so I have to use a serial lead plugged into a Modem.
Not all modems recognise the same AT commands and these do not work.
The web pages automate driving the modem just enough and can always be copied and extended.
https://homepages.plus.net/dougrice./dev/DigiSpark/simpleterm/simpletermModem.html
https://homepages.plus.net/dougrice./dev/DigiSpark/simpleterm/simpletermModemOne.html
V.21 modems are FULL DUPLEX and uses these frequencies
ITU-T v.21
Originate
1080 = Carrier
1180 = Space (Carrier + 100Hz)
980 = Mark (Carrier - 100Hz)
Answer
1750 = Carrier
1850 = Space (Carrier + 100Hz)
1650 = Mark (Carrier - 100Hz)
By using four frequencies , it is possible to use the 2-wire Telephony. so both diagrams (a) and (c) can be used.
In the diagram (c) ( two wire - FULL DUPLEX ) , It would be more difficult to detect the CARRIER if the TX and RX used the same carrier.
In the diagram (a) ( four wire - FULL DUPLEX ) , consider only transmitting if the carrier from the other end is present.
BAUDOT only sends when a key is pressed and the modem listens and RX is disabled when it is sending.
This is 2 wire - Half Duplex
BAUDOT: - 5 bit * carrier (for 150ms) + 0 + 11000 + 1 (for 40ms) * * 0 - 1800Hz - SPACE * 1 - 1400Hz - MARK * START 0 * STOP 1 *
It has an added complication of needing to shift between letters and numbers which must be tracked for both directions.
https://learn.adafruit.com/clue-teletype-transmitter/tty-fundamentals
There is a EDT mode that is also 2 wire - Half Duplex and uses the V.21 frequencies , but 110 BAUD.
In the diagram (b) ( two wire - HALF DUPLEX ) , consider only transmitting using BAUDOT or EDT based on the method used from the other end.
The CD output should ensure that the carrier has been present for long enough and at a strong enough level.
It is okay to transmit a "ping" test character sequence if the CD has not been detected a carrier from the other end.
See ITU V.8 for examples.
A TDM telephone system can be sumarised by this diagram, where the codecs are clocked at multiples of 8kHz sample rate. Jitter buffers were used as well.
http://www.dougrice.plus.com/dougnapTheory/index.htm has some more theory about the two wire telephony. Rice simplified Webb's equations for a telephone exchange in PO603 , which is available from the BT Heritage site.
Computation of the Characteristics of Telephone Connections
So it is possible to write down a list of 2x2 matrices and for each frequency use a computer to work out the the Microphone to earphone sensitivities. In about 1984, I was using a NASCOM 2 , Z80 based computer using BLS pascal. At some stage I purchased Turbo Pascal 3.0 for CP/M on an Amstrad 664, and also for the IBM PC. In 2009, the code was converted to JavaScript to run on the web browser.
The model for the four wire exchange model becomes:
and matrices for each element:
Hybrid using a transformer, similar to that used in GPO dialphone:-
Textphone, used keyboards and UARTS and modems to send the typing over the 2-wire Telephony using audio instead of voice:-
and Textphone, used modems UARTS and displays to display the typing over the 2-wire Telephony :-
Summary of a 2-wire Phone:-
- X - Switch controlled by microcontroller in phone.
phone line -- [ Hookswitch ] - X - [ bell / alerter ]
X - Caller ID -- [ display ]
X - To Hybrid
[ dtmf keypad ] - [ dtmf IC ] - X --|
[ microphone ] - [ audio ] - X --|
[ keyboard ] - [ modem ] - X --|
|-- to Hybrid
from Hybrid --|
| -- [ audio ] - X - [ loudspeaker / handset ]
| -- [ modem ] - X - [ display ]
http://www.dougrice.plus.com/dougnapTheory/index.htm has notes about 2-wire phones and Hybrids
https://www.britishtelephones.com/diagrams/n846.pdf has the circuit for phone and explains the hookswitch.
The hook switch used to connect the Bell circuit across the line until the User picked up the handset.
The bell was AC coupled to the phone line. When the user picks up the handset, The phone draws current.
With the handset off hook the hybrid circuit was connected to the phone line.
The line current powered the phone and signalled that the phone was off hook.
Modems that plug into a phone line have a similar functional block diagram.
simplified Hybrid used in Post office 706 and 746 from the 1960's
Consider the coil, assume the windings are the same, so the voltage across each winding are the same.
If ZL == the Balance Z components, and the Earphone is high impedance, then when the voltage V3+V2 == V1+V across Zl, the Voltage across The ear phone will be minimal when the person speaks.
When the person does not speak, the AC Voltage across the MIC will be zero so V Earphone will be -V across ZL.
Circuit for Post Office 706 Telephone
Loop Dis dialling for Post Office 706 Telephone
Telephones use DTMF to speed up the signalling.
http://ccgi.dougrice.plus.com/cgi-bin/wiki.pl?DTMF_Dabbles
Once established , a phone call is between KNOWN and KNOWN end points and allow either end to talk next.
The Client Server model, only allows the Client talk to the Server, which can respond. A PEER to PEER protocol needs to be engineered on top, with keep alives
Client Server is from an UNKNOWN to a KNOWN, but
IP4 addresses makes the endpoints UNKNOWN
There are only four things certain in life
1) Taxes 2) Death 3) Our company's Early release Schemes and 4) Re-Engineering IP protocols to have a keep alive!
The Web Browser is a useful tool!