• Introduction

Software Radio takes from, and has impact on, many of the concepts and technologies mentioned in previous app notes in the Engineering Focus Series.  One of the most fundamental and important concepts behind the Software Radio is the analog to digital (A/D) converter technology.

The idea of Software Radio is to move as much of the processing as close to the antenna as possible and to perform the majority of processing in software from then on.  This necessitates the use of extremely fast (samples/second) and wide (bits/sample) A/D converters that operate over a wide frequency range at low noise.   If the target platform is handheld or mobile based, then the additional design constraints of low cost, low power and small size come into play.  These targets are hard (if not currently impossible) to achieve simultaneously given current technology.

The job of the A/D converter is to digitize the input signal and to produce a quantized version of the input signal at the output for further digital processing.  The A/D converter can digitize the signal after the I/Q demodulator, after the RF downconverter (IF) or in a more advanced design, the A/D converter can digitize the signal at the antenna terminals (or prior to the downconverter).

• A/D Converter Specification Metrics

Since the output of an A/D is a digitally quantized value, it will not represent the value of the input signal exactly.  The samples will fall in between quantization levels (the higher the resolution the more accurate the output).   The difference between the energy represented by the approximate quantization level and the true analog level contributes to "Quantization Noise".  The resolution of the A/D is a function of several factors, including the input voltage range (dynamic range), the required sampling speed of the A/D and the power drain.

The Signal to Noise Ratio (SNR) is the ratio between the signal power and the noise power.  Here the noise power usually refers to thermal noise.

The Signal to Quantization Noise Ratio (SQNR) is also the ratio between the signal power and the noise power.  However, in the case of SQNR, the noise power refers specifically to the quantization noise.

The Signal to Noise and Distortion Ratio (SINAD) is the ratio between the signal power and all undesired energy, including thermal noise, quantization noise and Intermodulation distortion.  The SINAD measurement is obviously the worst case measurement as in includes both SNR noise contributions and SQNR noise contributions and Intermodulation Distortion.

The Effective Number of Bits (ENOB) can be approximated from the SINAD measurement as follows:

ENOB ~ (SINAD(dB) for a given Number of Bits - 1.76 ) / 6.02

Total Harmonic Distortion (THD) is a metric indicative of the noise that is related only to harmonics (like Intermodulation Distortion).  This measurement abstracts out the noise components due to thermal and quantization noise.   In essence then, SINAD is the ratio between the signal power and (thermal noise + quantization noise + THD), where Intermodulation Distortion is a subset of THD.

The Spurious Free Dynamic Range (SFDR) is the difference (usually in dB) between the desired output signal and any undesired harmonics in the output spectrum.  This metric is important for wideband A/Ds as the wider frequency range opens up the possibility to generate undesired harmonics.

• This is a cross-section of the current state of the art in A/D converter technology:

Component (all by itself)

From Analog Devices (part number = AD6644)
14 bit sampling @ 65 MHz.

Board

This is what is commonly available from Board vendors. The performance here is somewhere between "Component" Performance and "System" Performance.

Spectrum Signal Processing
Part Number = PMC - MAI.

12 bit sampling @ 65 MHz.


System - These are Real Life A/D implementation examples.

This is what industry and Academia are prototyping as complete systems.
http://www.mmitsforum.org/MTGS/mtg_23_apr01/kohno_tokyo_05_01_01.pdf

10 bit sampling @ 60 MHz.
12 bit sampling @ 40 MHz.
14 bit sampling @ 5 MHz.


It is one thing to achieve a given spec or benchmark at the component level, but quite another to achieve the same results when accounting for all the other "system issues".

• Benefits of Software Radio

There are some potential benefits to using Software Radios that might also be realized via the use of more classical software and hardware implementations.  Here the idea is to find the optimal path in terms of minimizing cost, complexity etc.  Software Radio is one possible way to achieve these types of goals.

• Multimode / Multiband operation

These abilities are already being implemented with existing technology.  Software Radio might be a cleaner solution although it may not be more economical at the present time.

There are many potential benefits that might be brought about through the implementation of Software Radios that might not be as easy to achieve otherwise, including:

• Varying Class / Quality of Service

• Pay as you Go Features

• Bit rate guarantees

Future app notes in this series will explore additional aspects of Software Radio.