Software Defined Radio Handbook

1 Pentek, Inc. One Park Way, Upper Saddle River, NJ 07458 Tel: (201) 818-5900 Fax: (201) 818-5904 Email: Defined Radio HandbookEighth EditionPentek, Park Way, Upper Saddle River, New Jersey 07458 Tel: (201) 818-5900 Fax: (201) 818-5904 Email: 1998, 2001, 2003, 2006, 2008, 2009, 2010 Pentek updated: January 2010 All rights of this publication may not be reproduced in any form without written are subject to change without , GateFlow, ReadyFow and VIM are registered trademarks of Pentek, of SDRT echnologyProductsApplicationsSummaryLink sbyRodger H. HoskingVice-President & Cofounder of Pentek, Inc. Pentek, Inc. One Park Way, Upper Saddle River, NJ 07458 Tel: (201) 818-5900 Fax: (201) 818-5904 Email: Defined Radio HandbookPrefaceSDR ( Software Defined Radio ) has revolutionized electronic systems for avariety of applications including communications, data acquisition and signal Handbook shows how DDCs (Digital Downconverters) and DUCs (Digital Upconverters),the fundamental building blocks of SDR, can replace conventional analog receiver designs,offering significant benefits in performance, density and order to fully appreciate the benefits of SDR, a conventional analog receiversystem will be compared to its digital receiver counterpart, highlighting similarities and inner workings of the SDR will be explored with an in-depth description of the internalstructure and the devices used.

2 Finally, some actual board- and system-level implementations and availableoff-the-shelf SDR products for embedded systems will be , Inc. One Park Way, Upper Saddle River, NJ 07458 Tel: (201) 818-5900 Fax: (201) 818-5904 Email: Defined Radio HandbookBefore we look at SDR and its various implementa-tions in embedded systems, we ll review a theoremfundamental to sampled data systems such as thoseencountered in Software Defined s Theorem: Any signal can be represented by discretesamples if the sampling frequency is at least twicethe bandwidth of the signal. Notice that we highlighted the word bandwidthrather than frequency. In what follows, we ll attempt toshow the implications of this theorem and the correctinterpretation of sampling frequency, also known assampling Simple Technique to Visualize SamplingTo visualize what happens in sampling, imaginethat you are using transparent fan-fold computerpaper. Use the horizontal edge of the paper as thefrequency axis and scale it so that the paper folds lineup with integer multiples of one-half of the samplingfrequency s.

3 Each sheet of paper now represent what wewill call a Nyquist Zone , as shown in Figure 1 Nyquist s Theorem and Samplingfs/2fs3fs/22fs5fs/23fs7fs/20 Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Zone 7 FrequencyPentek, Inc. One Park Way, Upper Saddle River, NJ 07458 Tel: (201) 818-5900 Fax: (201) 818-5904 Email: Defined Radio HandbookSamplingFigure 3 Baseband SamplingFigure 4A baseband signal has frequency components thatstart at = 0 and extend up to some maximum prevent data destruction when sampling a basebandsignal, make sure that all the signal energy falls ONY inthe 1st Nyquist band, as shown in Figure are two ways to do this:1. Insert a lowpass filter to eliminate all signalsabove s/2, or2. Increase the sampling frequency so all signalspresent fall below that s/2 is also known as the folding frequency .Sampling Bandpass SignalsFigure 2 Use the vertical axis of the fan-fold paper for signalenergy and plot the frequency spectrum of the signal tobe sampled, as shown in Figure 2.

4 To see the effects ofsampling, collapse the transparent fan-fold paper into BasicsThe resulting spectrum can be seen by holding thetransparent stack up to a light and looking through can see that signals on all of the sheets or zones are folded or aliased on top of each other and theycan no longer be this folding or aliasing occurs during sampling,the resulting sampled data is corrupted and can never berecovered. The term aliasing is appropriate becauseafter sampling, a signal from one of the higher zonesnow appears to be at a different s consider bandpass signals like the IF frequencyof a communications receiver that might have a 70 MHzcenter frequency and 10 MHz bandwidth. In this case,the IF signal contains signal energery from 65 to 75 we follow the baseband sampling rules above, wemust sample this signal at twice the highest signalfrequency, meaning a sample rate of at least 150 , by taking advantage of a technique called undersampling , we can use a much lower sampling 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Zone 7 Energyfs/2fs3fs/22fs5fs/23fs7fs/20 Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Zone 7No Signal Energyfs/20 Folded SignalsFall On Top ofEach OtherPentek, Inc.

5 One Park Way, Upper Saddle River, NJ 07458 Tel: (201) 818-5900 Fax: (201) 818-5904 Email: Defined Radio Handbookfs/20 Folded signalsstill fall on top ofeach other - butnow there isenergy inonly one sheet !UndersamplingFigure 5 SamplingFigure 6fs/2fs3fs/22fs5fs/23fs7fs/20 Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Zone 7No Signal EnergyNo Signal EnergyUndersampling allows us to use aliasing to ouradvantage, providing we follow the strict rules of theNyquist our previous IF signal example, suppose we try asampling rate of 40 5 shows a fan-fold paper plot with Fs = 40 can see that zone 4 extends from 60 MHz to 80 MHz,nicely containing the entire IF signal band of 65 to 75 when you collapse the fan fold sheets as shownin Figure 6, you can see that the IF signal is preservedafter sampling because we have no signal energy in anyother note that the odd zones fold with the lowerfrequency at the left (normal spectrum) and the evenzones fold with the lower frequency at the right (reversedspectrum).

6 In this case, the signals from zone 4 are frequencyreversed. This is usually very easy to accommodate inthe following stages of SDR sampling requires the sample frequency tobe at least twice the signal bandwidth. This is the sameas saying that all of the signals fall within the firstNyquist real life, a good rule of thumb is to use the 80%relationship:Bandwidth = x s/2 Undersampling allows a lower sample rate even thoughsignal frequencies are high, PROVIDED all of thesignal energy falls within one Nyquist repeat the Nyquist theorem: The sampling frequencymust be at least twice the signal bandwidth not thesignal major rule to follow for successful undersamplingis to make sure all of the energy falls entirely in oneNyquist two ways to do this:1. Insert a bandpass filter to eliminate all signalsoutside the one Nyquist Increase the sampling frequency so all signalsfall entirely within one Nyquist , Inc. One Park Way, Upper Saddle River, NJ 07458 Tel: (201) 818-5900 Fax: (201) 818-5904 Email: Defined Radio HandbookFigure 7 The conventional heterodyne Radio receiver shownin Figure 7, has been in use for nearly a century.

7 Let sreview the structure of the analog receiver so comparisonto a digital receiver becomes the RF signal from the antenna is amplified,typically with a tuned RF stage that amplifies a regionof the frequency band of interest. This amplified RF signal is then fed into a mixerstage. The other input to the mixer comes from the localoscillator whose frequency is determined by the tuningcontrol of the mixer translates the desired input signal to theIF (Intermediate Frequency) as shown in Figure IF stage is a bandpass amplifier that only letsone signal or Radio station through. Common centerfrequencies for IF stages are 455 kHz and MHzfor commercial AM and FM demodulator recovers the original modulatingsignal from the IF output using one of several example, AM uses an envelope detector and FMuses a frequency discriminator. In a typical home Radio ,the demodulated output is fed to an audio poweramplifier which drives a 8 Analog Radio Receiver Block DiagramAnalog Radio Receiver MixerThe mixer performs an analog multiplication of thetwo inputs and generates a difference frequency frequency of the local oscillator is set so thatthe difference between the local oscillator frequency andthe desired input signal (the Radio station you want toreceive) equals the example, if you wanted to receive an FMstation at MHz and the IF is MHz, you wouldtune the local oscillator - = 90 MHzThis is called downconversion or translation because a signal at a high frequency is shifted down to alower frequency by the IF stage acts as a narrowband filter which onlypasses a slice of the translated RF input.

8 The band-width of the IF stage is equal to the bandwidth of thesignal (or the Radio station ) that you are trying toreceive. For commercial FM, the bandwidth is about100 kHz and for AM it is about 5 kHz. This is consis-tent with channel spacings of 200 kHz and 10 kHz, of SDRANALOGLOCALOSCILLATORIF AMP(FILTER)SPEAKERANTENNADEMODULATOR(Det ector)ANALOGMIXERAUDIOAMPRFAMP0RF INPUT SIGNALFROM ANTENNAMIXER TRANSLATESINPUT SIGNAL BANDto IF FREQUENCYANALOG LOCALOSCILLATORFRFFIFS ignalPentek, Inc. One Park Way, Upper Saddle River, NJ 07458 Tel: (201) 818-5900 Fax: (201) 818-5904 Email: Defined Radio Handbook0 FSIGMIXER TRANSLATESINPUT SIGNALBAND to DCDIGITAL LOCALOSCILLATORFLO=FSIGCHANNELBANDWIDTHI F BWSignalDIGITALMIXERDIGITALLOCALOSCDSPDD CD igital DownconverterRFTUNERA nalogIF SignalAnalogRF SignalA/DCONVD igital IFSamplesLOWPASSFILTERD igitalBasebandSamplesSDR Receiver Block DiagramFigure 9 Principles of SDRF igure 10 SDR Receiver MixerFigure 9 shows a block diagram of a softwaredefined Radio receiver.

9 The RF tuner converts analog RFsignals to analog IF frequencies, the same as the first threestages of the analog A/D converter that follows digitizes the IF signalthereby converting it into digital samples. These samplesare fed to the next stage which is the digital downconverter(DDC) shown within the dotted digital downconverter is typically a singlemonolithic chip or FPGA IP, and it is a key part of theSDR conventional DDC has three major sections: A digital mixer A digital local oscillator An FIR lowpass filterThe digital mixer and local oscillator translate thedigital IF samples down to baseband. The FIR lowpassfilter limits the signal bandwidth and acts as a decimat-ing lowpass filter. The digital downconverter includes alot of hardware multipliers, adders and shift registermemories to get the job digital baseband samples are then fed to a blocklabeled DSP which performs tasks such as demodulation,decoding and other processing , these needs have been handled withdedicated application specific ICs (ASICs), and program-mable the output of the mixer, the high frequencywideband signals from the A/D input (shown in Figure10 above) have been translated down to DC as complex Iand Q components with a frequency shift equal to thelocal oscillator is similar to the analog receiver mixer exceptthere, the mixing was done down to an IF , the complex representation of the signal allows usto go right down to tuning the local oscillator over its range, anyportion of the RF input signal can be mixed down to effect, the wideband RF signal spectrum can be slid around 0 Hz, left and right, simply by tuning thelocal oscillator.

10 Note that upper and lower sidebands , Inc. One Park Way, Upper Saddle River, NJ 07458 Tel: (201) 818-5900 Fax: (201) 818-5904 Email: Defined Radio HandbookDIGITALMIXERDIGITALLOCALOSCA/DCO NVD igital IFSamplesLOWPASSFILTERD igitalBasebandSamplesTranslationFilterin gTuning FreqDecimationDDC Signal ProcessingFigure 12 Principles of SDRF igure 11A Local Oscillator Frequency SwitchingDDC Local Oscillator and DecimationF1F2F390OA/D Sample Rate(before decimation)Sample Rate: FsDecimatedFilter OutputSample Rate: Fs/NFigure 11B FIR Filter DecimationBecause the local oscillator uses a digital phaseaccumulator, it has some very nice features. It switchesbetween frequencies with phase continuity, so you cangenerate FSK signals or sweeps very precisely with notransients as shown in Figure frequency accuracy and stability are determinedentirely by the A/D clock so it s inherently synchronousto the sampling frequency. There is no aging, drift orcalibration since it s implemented entirely with digital the output of the FIR filter is band limited, theNyquist theorem allows us to lower the sample rate.