Why, 120 years after the pioneering demonstrations of radio transmission by Marconi, Bose, Popov, and probably others, is software-defined radio (SDR) so exciting?

To understand this, and what SDR has to offer all of us, you need to know how a radio receiver works. At its most basic – and there are still plenty of extremely simple radios in use – it consists of:

• an antenna, which collects the electro-magnetic (radio) waves, and converts them into a very small oscillating electric voltage;
• a tuner, which selects only a narrow band of frequencies in the radio frequency (RF, from 3 kHz to 3000 GHz) signal from the antenna;
• a detector/converter, which converts the RF to audio frequencies (AF) within our hearing range of 20 to 20,000 Hz;
• a speaker or similar output system which enables us to hear the AF.

One implementation of such an elegantly simple radio receiver is a crystal radio, still widely used around the world for receiving traditional amplitude modulated (AM) radio stations.

However AM radio has many disadvantages, and has in many uses been replaced by frequency modulation (FM) for regular analogue radio transmissions. Decoding FM requires a separate demodulation step to decode the FM to AF, and a more complex receiver design, such as the single superhet shown.

These AM and FM designs work well when there are strong signals from nearby broadcasting stations, pumping out kilowatts of RF, and there is little noise or interference. Following the early breakthroughs, a whole industry of analogue RF design developed, and produced very potent and sophisticated transmitters and receivers.

Since the Second World War, there have been huge advances in squeezing more information into relatively narrow bandwidths, first to enable the broadcast of TV, and most recently for high-quality stereo sound and digital modes. The latter have been important to support radio transmission of still images such as weather charts, as well as telemetry and computer data, and encryption for privacy.

These new modes, and the huge advances in digital signal processing (DSP) which have come with computers, have enabled a fundamentally different approach to receiver design, as shown in the lower two block diagrams for modern SDR systems. Instead of working with oscillating signals throughout the circuits in a receiver, SDR converts the signals to digital form, then processes those instead.

The most basic form of SDR is illustrated by the RTL device (item 3 in the diagram), which has an R820T chip to tune (select) the RF signal from the antenna and convert it to an intermediate frequency (IF), and the RTL2832U chip to convert that from analogue to 8-bit digital, and stream the digital data over USB to a computer. Software running on the computer then processes the signal, generating audio for output.

When the radio industry comes along with a new mode, such as digital audio broadcasting (DAB) or digital radio mondiale (DRM), all you need to do is develop host computer software to decode that new mode, and your SDR will support it. You should not need to replace your radio receiver.

This also allows you to use your radio receiver to do much more adventurous things, which previously required the development of expensive dedicated systems. The same SDR can be used for radio-astronomy, sniffing WiFi, analysing RF interference, and as a radar receiver: all it needs is the development of software, which is relatively quick and cheap.

The fourth block diagram shows a different approach to SDR which is becoming popular in more sophisticated receivers and transmitters, such as those from the Italian specialist ELAD. There is no tuner after the antenna, merely some broad RF filters (which can be switched out anyway) before the RF is converted to a digital signal: this is direct sampling, or direct digital conversion, which takes in the whole received spectrum from 9 kHz up to 52 MHz, and turns it into 16-bit digital samples at a rate of nearly 2 gigabits per second (Gbps).

With such a huge pipe of data, the only way to process it is with a field-programmable gate array (FPGA) digital signal processor (DSP), in this case a Xilinx Spartan. Spartan FPGAs are widely used in set-top boxes, WiFi base stations, and similar products. The Spartan is used here to generate 16-bit in-phase and quadrature (IQ) samples which are streamed over a USB2 connection to a computer, where they are further processed in software.

Thus the more advanced, and costly, ELAD digitises to 16 rather than 8 bits, and streams those samples at a rate of 384 ksamples per second (for each of two channels) rather than the 2.4 to 3.2 Msamples per second (single channel only) for the RTL-SDR. Coupled with more sophisticated signal processing software on the host computer, the ELAD can decode Morse (CW), sideband (USB, LSB, DSB), AM, FM (narrow and wide), RTTY (radio teletype, an early digital mode), DRM, and RDS (radio data system, a modern digital mode).

Now that I have both an RTL-SDR and an ELAD FDM-S2, future articles will describe my explorations with them.

Important Abbreviations

AF – audio frequency, typically 20 to 20,000 Hz.
AM – amplitude modulation, older form of encoding radio signals.
DAB – digital audio broadcasting, usually over VHF and higher bands using MP2 or AAC+ encoding.
DRM – digital radio mondiale, digital audio broadcast over HF bands using MPEG-4 encoding.
DSP – digital signal processing, using microprocessors and related chips.
FM – frequency modulation, newer form of encoding radio signals.
IF – intermediate frequency, common to any given design of receiver, and typically 20-30 MHz.
IQ – in-phase and quadrature, a highly efficient means of representing a signal.
RF – radio frequency, typically 3 kHz to 3000 GHz.
SDR – software-defined radio.