Chapter 31: Digital Communication Modes

Text modes and image modes

The manual splits them into text modes (Morse, RTTY, packet, WSJT and others) and image modes (facsimile, SSTV and fast-scan TV). Keyboard chat and file transfer are uses, not its groups.

The transceiver’s loudspeaker or line output

Received audio comes from the loudspeaker output or a dedicated audio output and goes into the sound card’s input. The microphone input is where the sound card’s output goes.

To the radio’s microphone or modulation input

Transmit audio from the sound card’s output drives the microphone or other modulation input, and a port on the computer keys the transmitter. The loudspeaker output carries audio the other way.

The loudspeaker output level changes with the volume knob

The loudspeaker output follows the volume knob, so the level reaching the sound card would change every time it was turned. Its strength is not the problem; its changing level is.

CW

A square wave on AM switches the carrier fully on and off, which is CW (on-off keying). The same square wave on FM shifts the frequency between two values, giving FSK.

FSK

Frequency modulation by a square wave flips the carrier between two frequencies: frequency-shift keying (FSK). On AM the same square wave keys the carrier on and off, giving CW.

They now outnumber phone contacts worldwide

ClubLog uploads showed data contacts overtaking phone contacts worldwide, and FT8 alone later passed half of all contacts logged. Phone has not kept its lead.

It is integrated, so each contact is logged directly

Logging software now links to the digital-mode software, so contacts go into the log directly with no special software or hardware. Typing each one in again is no longer needed.

A standard SSB receiver feeds a sound card, which demodulates it

A standard SSB receiver recovers the audio and an ordinary sound card demodulates it, showing text, a waterfall or even speech. No special data receiver is needed.

Free or cheap software, an ordinary computer and sound card, and an SSB rig

Most digital modes run in software, much of it free, with a PC sound card and a normal SSB transceiver; this simplicity and economy explain their popularity. Specialised hardware is no longer needed.

Experts copy it by ear at up to roughly 60 words per minute

CW occupies a very narrow bandwidth, is readable in marginal conditions and needs only simple equipment; skilled operators can copy around 60 words per minute by ear. Computers can help but are optional, and CW thrives.

Morse telegraphy

Morse telegraphy, on-off keying of the carrier, is the original radio mode. RTTY was one of the first data modes, but came well after Morse; packet radio and SSTV are later still.

SSB has no place in the CW sub-bands, and it often adds hum and clicks

The manual calls this illegal, since the band plan keeps SSB out of the Morse sub-bands, and it often puts hum, clicks and pops on air. Sending slowly or at low power cures neither problem.

So that the rig’s own waveform shaping and filtering are used

Keying the transmitter directly uses its keying waveform shaping and filtering, so the CW stays clean and free of key clicks. The power available is the same either way.

It decodes all the Morse signals across a whole band at once

CW Skimmer uses a sound card and PC to read every Morse signal in a whole band at the same time, which no human can do. Speed on one signal is not its point.

Casual keyboard-to-keyboard chats and working DX

RTTY is too slow and error prone for data files, but it is still widely used for casual keyboard chats and for working DX, though the WSJT modes are taking over.

It is slow and prone to errors

RTTY runs at only 45,45 Bd with no error correction, so it is slow and error prone. Its 170 Hz shift makes it narrower, not wider, than most packet signals.

170 Hz

Amateur RTTY shifts 170 Hz between mark and space. Early FSK systems also used a wide 850 Hz shift, but 170 Hz is the amateur standard, kept by modern AFSK.

2 125 Hz for mark and 2 295 Hz for space

AFSK traditionally feeds a 2 125 Hz mark tone and a 2 295 Hz space tone into the transmitter, keeping the 170 Hz shift. Swapping the two tones would invert every bit.

Two audio tones, one for mark and one for space

AFSK feeds two audio tones (mark and space) into an SSB transmitter, which shifts its output to match. One tone keyed on and off is not AFSK; shifting the carrier with a control voltage is plain FSK.

45,45 Bd

Amateur RTTY runs at 45,45 Bd, a 22 ms pulse. 50 Bd (20 ms pulses) was the usual rate on European and South African teleprinter landlines.

13,3 ms

Pulse length = 1 ÷ signalling rate = 1 ÷ 75 = 0,013 3 s = 13,3 ms, just as 45,45 Bd gives 22 ms. Writing 75 ms only reuses the rate as a time.

125 Bd

Signalling rate = 1 ÷ pulse length = 1 ÷ 0,008 s = 125 Bd. Reading 8 ms as 8 Bd confuses the time of one pulse with the number of pulses each second.

32

n bits give 2ⁿ codes, so five bits give 2⁵ = 32. Multiplying 2 × 5 = 10 is the slip; 128 is 2⁷, the count for seven-bit ASCII.

Its 32 codes were too few for every letter, figure and punctuation sign

Five bits give only 2⁵ = 32 codes, too few for 26 letters, 10 figures and punctuation, so each code means a letter or a figure according to the shift. Baudot has no small letters.

128

Seven bits give 2⁷ = 128 symbols. 2 × 7 = 14 is the slip of multiplying, and 2⁸ = 256 needs an eighth bit, which can extend the character set.

It is the usual character code in computers

Short for American Standard Code for Information Interchange, ASCII is the 7-bit code of computers and data equipment. Amateur RTTY uses the 5-bit Baudot code, and a nibble per decimal digit is BCD.

American Standard Code for Information Interchange

ASCII stands for American Standard Code for Information Interchange; it is the 7-bit code of computers and communication systems.

A parity bit, for error checking

An eighth bit may be a parity bit for error checking or, with other error checks, it can extend the set to 256 symbols. Letters and figures shifts belong to Baudot.

A start bit first, then one or two stop bits at the end

Asynchronous data frames every character with 1 start bit and 1 or 2 stop bits, each as long as a data bit. Sending the bare code on its own is synchronous transmission.

Space

The idle line sits at mark, so the start pulse drops to space to show that a character is beginning; five data bits follow, then a stop pulse at mark.

1 100

Each character takes 1 + 7 + 1 = 9 bits, so 9 600 ÷ 9 = 1 067 characters per second, nearest 1 100. Leaving out the start and stop bits gives 9 600 ÷ 7 = 1 371.

The least significant bit

Serial standards send the least significant bit first, so the code 1001101 for M goes out as 1, 0, 1, 1, 0, 0, 1. Most significant first is the slip.

1100001

The least significant bit, the right-hand 1, goes first, so 1000011 is sent as 1100001, reversed. Sending 1000011 as written would be most significant bit first.

One signalling event, or symbol, per second

A baud is one discrete signalling condition or event per second. It equals one bit per second only when each event carries one bit, as in two-tone RTTY; four-phase PSK carries two.

1 200 Bd

Each of the four phases carries two bits (a dibit), so the baud rate is half the bit rate: 2 400 ÷ 2 = 1 200 Bd. Taking baud as bits per second gives 2 400 Bd.

Other stations along the way relay the packets onward

Packet frames hop through digipeaters and network nodes to stations out of direct range, much as the Internet passes data on. High power and big antennas are not how packet works.

Sends it on again, adding no information of its own

A digipeater just re-transmits any frame carrying its callsign in the digipeat field, adding nothing. Holding messages until they are collected is the job of a bulletin board (BBS).

Holds messages until the addressee, or anyone if public, fetches them

A BBS stores messages for later collection by the addressee or, if posted publicly, by anyone, such as newsletters. Re-sending frames at once without storing them is digipeating.

Several at once, all sharing it

Packet stations send short frames in turn, so a single channel serves many communications at once. A phone-style circuit that ties up the channel for one contact is what packet switching avoids.

The TNC

The TNC (terminal node controller) builds the computer’s ASCII data into frames and hands them to a modem, which turns them into audio tones; on receive the steps reverse.

Terminal Node Controller

TNC stands for Terminal Node Controller, the heart of a packet station: it assembles the computer’s data into frames for the modem and radio.

The address field

An AX-25 frame runs flag, address (destination, source, digipeaters), control, information, frame check sequence, flag; real frames also carry a one-byte protocol identifier after the control field. AX-25 is usually written AX.25.

Detecting errors in the received data

The 2-byte frame check sequence (FCS) lets the receiver detect a corrupted frame. The callsigns sit in the address field near the start, and the message in the information field.

A keyboard session with another station

TCP/IP, the protocol set that runs the Internet, also runs over packet radio: Telnet gives a live keyboard session (the manual’s chat session), while FTP (File Transfer Protocol) moves files between stations.

Real-time tracking of mobile stations by GPS

APRS (Automatic Position Reporting System) sends position reports, usually from a GPS receiver, so stations can be followed on a map in real time. It does not report propagation.

It sends UI frames, which need no link to each station

Packet’s connected mode would need a permanent link between all stations; APRS sends unnumbered information (UI) frames that anyone can receive, so any number of stations can join.

Control can follow every team and vehicle on a map

With GPS receivers on rescuers and vehicles, control stations see every position on a computer map in real time. APRS reports positions; it is not a voice system.

Readings from a digital weather station

Some APRS stations add data from digital weather stations to their position reports. APRS carries short data, not video or sound, and it does not forecast propagation.

A suite of weak-signal modes, each optimised for its own use

WSJT packs many weak-signal modes into one free package, each optimised for a use such as EME, meteor scatter or HF DX. It is a suite of modes, not one mode.

Very slow signalling, with a great deal of redundancy

Most WSJT modes use FSK or PSK variations with very slow baud rates and heavy redundancy; fast meteor-scatter modes such as FSK441 and MSK144 are the exception. They do not depend on repeat requests.

Meteor scatter

FSK441 is a fast WSJT mode for the brief trails of meteor scatter; MSK144 has since largely replaced it. EME needs slow modes such as JT65.

EME on 2 m

JT65B was optimised for EME at 144 MHz and long served as the standard there; newer WSJT-X modes such as Q65 now share that work. Meteor scatter needs fast modes.

FT8

FT8, from the WSJT stable, is now the de facto mode for HF DX. RTTY was long the DX data mode but has been declining fast in favour of the WSJT modes.

It is almost obsolete, displaced even by plain RTTY

AMTOR, an error-correcting development of RTTY from maritime use, has fallen out of amateur use: better protocols and RTTY itself displaced it. It was never a fast mode.

It has exactly four mark bits and three space bits

Every 7-bit AMTOR character has four marks and three spaces, so a received character with any other ratio must be corrupt and can be rejected.

Mode A asks for a repeat of bad characters; Mode B sends each character twice

Mode A is ARQ (automatic repeat request): the receiver checks each character and asks for a repeat. Mode B is forward error correction, sending every character twice, so it needs no return path.

Its codes vary in length, the commonest characters having the shortest

PSK31 (PSK at 31,25 Bd, by G3PLX) uses a variable-length code in which the commonest characters have the shortest codes, much as in Morse, so text goes faster. A fixed-length code, such as Baudot, gains nothing from this.

Very low error rates at moderate speed in an HF channel

PACTOR joins packet-style blocks to AMTOR-like coding, with a 16-bit CRC and memory ARQ, giving very low error rates at 100 to 200 Bd in 500 Hz of HF. It is not a fast VHF mode.

An HF data system that adapts itself to propagation conditions

CLOVER, by W7GHM, is an adaptive HF data system: it uses four tones in 500 Hz at about 250 Bd and changes its modulation to suit the conditions. It does not keep one fixed speed.

Sending detailed still pictures, such as weather charts

Facsimile, the oldest image method, sends very high-resolution still pictures over voice-bandwidth channels: weather charts, newspaper photos and weather-satellite images. Moving pictures are television.

By sending slowly, so each picture takes several minutes

Fax trades time for detail: a slow data rate lets 800 to 1 600 lines through a voice channel, at 4 to 10 minutes per picture. A higher data rate would need more bandwidth.

SSTV is slower and lower in definition than fast-scan TV

SSTV squeezes TV-like pictures into a speech channel under 3 kHz, at the cost of seconds to minutes per frame and lower definition. Fast-scan TV, not SSTV, gives the sharper pictures and sends many each second.

Its signal needs several MHz of bandwidth

Fast-scan TV follows broadcast scan rates and needs several MHz: one channel would fill every HF band up to 50 MHz. Bandwidth, not propagation, keeps it on UHF and above.

120 s

Frame time = lines × line time = 256 × 0,45 = 115,2 s, nearest 120 s, about 2 minutes. Taking the 120-line monochrome frame instead gives 54 s.

24 s

Three colour frames of 8 s each take 3 × 8 = 24 s, the manual’s figure. Forgetting that each colour needs a whole frame of its own gives 8 s.

1 200 Hz sync, 1 500 Hz black, 2 300 Hz white

Pixel brightness runs from 1 500 Hz (black) to 2 300 Hz (peak white), and sync pulses sit below black at 1 200 Hz, all inside an SSB speech channel.

Interference during one frame could spoil the whole picture

Sending whole red, green and blue frames in turn meant one burst of interference could ruin a colour frame and so the picture; sending each line three times, still through the three filters, limits the damage.

Luminance and chrominance spoilt sharp, high-contrast edges

Luminance-chrominance halved the frame time and stayed compatible with monochrome sets, but sharp, high-contrast edges came out badly, so by the manual newer modes returned to RGB. Popular PD modes still use luminance-chrominance.

A 5 ms sync pulse starts a line; a longer one at the same tone starts a frame

A short 5 ms pulse at the 1 200 Hz sync tone begins every line; a longer pulse on that tone begins each new frame. Swapped, the receiver could not keep line timing.

The detector need only tell a zero from a one, and coding fixes errors

A digital detector need only decide between 0 and 1, and error-detecting and correcting codes repair many mistakes, so the link stays robust in poor propagation. Wider filters let in more noise.

Waveform coding and source model coding

Waveform coders quantise the speech wave itself, giving high quality at a high bit rate; source model coders (vocoders such as AMBE) send the settings of a speech model at far lower rates.

96 kb/s

Bit rate = bits per sample × samples per second = 12 × 8 000 = 96 000 b/s = 96 kb/s. Doubling it again for Nyquist is the slip: the sampling rate already allows for that.

24 to 32 kb/s

ADPCM predicts each sample and codes only the difference, cutting 64 kb/s PCM to 24 to 32 kb/s; the manual loosely calls these vocoders. The 2 to 5 kb/s range belongs to model-coding AMBE vocoders.

2 to 5 kb/s

AMBE (Advanced Multiband Excitation) model-coding vocoders give good speech at 2 to 5 kb/s, low enough for a normal 2,5 to 3 kHz HF channel. 24 to 32 kb/s is ADPCM.

The settings of a speech model: pitch, spectrum and voicing

A model-coding vocoder describes each 10 to 40 ms slice by its fundamental frequency (pitch), spectral coefficients and voicing, needing only 2 to 5 kb/s. Dropping samples would only distort the speech.

None: it uses the radio’s normal audio connections

The digital voice unit connects to the microphone and loudspeaker audio, so any SSB transceiver, even an old valve set, works unmodified; the signal fits the normal 2,5 to 3 kHz channel.

It still struggles when many signals share one channel

Digital speech is robust and efficient, but its biggest obstacle is many signals in one channel, where the human ear still copes better with analogue signals. It is no cure-all.

To link repeaters and users over the Internet

VoIP routes voice over the Internet or other IP networks; amateur systems such as EchoLink and IRLP use it to link repeaters and users worldwide. It is not a radio modulation.

Reach a repeater through the Internet from a computer

EchoLink, built on Speak Freely VoIP software, lets users reach repeaters from a computer over the Internet instead of by radio. Moon echoes are what an EME echo mode measures.