Use Digital Modulation To Increase Power Output In ISM Band Radios

Ordinarily, 902- to 928-MHz industrial, scientific, and medical(ISM) radio systems operating in the United States are constrained toPart 15.249 output power levels (–1 dBm conducted) unless they exploitcode- or frequency-hopping schemes. Increasing that power level maybenefit many applications, but code- and frequency-hoppingimplementations are complicated to implement. However, engineers canalso use the “digital modulation” specification in the Part 15regulations to increase the output power for ISM band radios withoutresorting to spreading.        
Historically,for non-FHSS (frequency-hopping spread-spectrum) systems, the FederalCommunications Commission (FCC) has required a direct-sequencespread-spectrum (DSSS) approach to be applied. DSSS required the directmultiplication of binary data at the transmitter by a pseudorandom bitsequence to generate a transmitted data stream defined in terms of akilochips/second data rate and to provide a minimum of 10-dB processinggain in the receiver during the demodulation process.         
Atthe receiver, the reverse process was applied to recover the originalbinary data stream. The redundancy provided by multiplying each databit by the pseudorandom “spreading” sequence resulted in the processinggain. For example, in a low-bit-rate application, multiplying each bitby an 11-bit coding sequence could theoretically improve the linkbudget by 10 dB if the bit rate (BR) is decreased by the same factor.        
Thespreading process generated a transmitted modulation spectrum thatapproaches that of “random” (or white) noise. Since the power spectraldensity (PSD) of random noise is less than that of a coherent signal atthe same power level, this allows for a higher transmit power level tobe maintained while still complying with FCC regulations.        
Inrecent years this rule has been changed, and the FCC now permitssystems that employ digital modulation techniques that have an occupied6-dB bandwidth of at least 500 kHz to have a power spectral density ofup to +8 dBm in 3-kHz bandwidth without the need to employspread-spectrum techniques. The programmable frequency deviation andoutput power of modern ISM-band RF devices easily fit these criteria.
FCC Regulations

This carriermodulated signal has a data stream of 76.8 kbits/s and a peak frequencydeviation of 140 kHz, measured using a spectrum analyzer with afrequency span that’s wide enough to capture the entire modulationenvelope and resolution bandwidth of 100 kHz.  

        
Low-power,non-licensed devices operating in the 902- to 928-MHz ISM band areeverywhere, including simple toys, wireless security systems, wirelesstelemetry, and wireless automatic meter reading. The FCC implementsrules to limit the potential for interference to licensed operations bylow-power, non-licensed transmitters. Part 15 of Title 47 of the Codeof Federal Regulations (“47 CFR Part 15”) documents these rules.         
Operationto FCC Part 15 is subject to two conditions. First, the device may notcause harmful interference. Second, the device must accept anyinterference received, including interference that may cause undesiredoperation. Hence, there is no guaranteed quality of service whenoperating a Part 15 device.        
Foroperation in the 902- to 928-MHz band, a low-power, non-licensed devicewill generally fall within Part 15.247 (frequency-hopping and digitallymodulated intentional radiators) and Part 15.249 (general non-licensedintentional radiators). Part 15.247(a)(2) covers digitally modulatedsystems. Under the definition of systems using digital modulationtechniques, the FCC allows a device to comply with these regulationswithout necessarily implementing DSSS, provided it meets the followingrequirements:        
• The  minimum 6-dB bandwidth of the signal shall be at least 500 kHz.        
•The maximum permitted peak conducted output power is +30 dBm (1 W).However, the power spectral density conducted from the intentionalradiator to the antenna shall not be greater than +8 dBm in any 3-kHzband during any time interval of continuous transmission.        
•If the antenna used has a directional gain in excess of 6 dBi, theconducted output power described shall be reduced by the amount in dBthat the directional gain of the antenna exceeds 6 dBi.        
•In any 100-kHz bandwidth outside the frequency band of operation, thepower shall be at least 20 dB below that in the 100-kHz bandwidthwithin the band that contains the highest level of the desired power.        
•Radiated harmonic and spurious emissions that fall within therestricted bands, as defined in FCC Part 15.205, must comply with theradiated emission limits specified in FCC Part 15.209.         
Basedon these requirements, compliance with digital modulation schemesimplies wideband modulation. Wideband operation requires thetransmitter to be capable of high frequency deviation and high datarates, either with raw NRZ data or “data-whitened” spread-spectrumcommunications. In addition, the receiver requires the bandwidth to beable to correctly demodulate the transmitted data. Also, thesensitivity can’t be degraded to where it negates the benefits of beingable to transmit at significantly higher output power.        
Signal Bandwidth And Spectral Density        
Forthe purposes of this article, below is an overview of methods forcompliance with the FCC rules. Engineers should seek more thoroughinformation on implementing these rules either from the FCC Office ofEngineering Technology or from vendor data sheets. (Several resourcesare referenced at the end of this article.)         
Asan example of signal bandwidth measurement, Figure 1 illustrates acarrier modulated signal with a data stream of 76.8 kbits/s and a peakfrequency deviation of 140 kHz, measured using a spectrum analyzer witha frequency span that is wide enough to capture the entire modulationenvelope with a resolution bandwidth (RBW) of 100 kHz.

Peak power spectraldensity can be measured. Since the spectral line spacing is greaterthan 3 kHz, no correction factor is required.

        
TheFCC requires the method of measuring the power spectral density to besimilar to that used to measure the conducted output power. For thepurposes of this article, we measure both peak conducted output powerand power spectral density.         
Tomeasure peak spectral density, center the spectrum analyzer on to theemission peak(s) within the signal passband. Set the instrument RBW to3 kHz and video bandwidth (VBW) to greater than the RBW. The sweep timeshould be set to the frequency span/3 kHz (i.e., for a 1.5-MHz span,the sweep time should be 500 seconds). The peak measured signal levelshould not exceed + 8 dBm. Next, designers should account for severalfactors correct the measured results.         
Ifthe measured spectral line spacing is greater than 3 kHz, no correctionfactor is required. If the measured spectral line spacing is equal to,or less than, 3 kHz, reduce the RBW until the individual spectral linesare resolved. The measured results must be normalized to 3 kHz bysumming the power of all the individual spectral lines within a 3-kHzband (in linear power units) to determine compliance.         
Ifthe spectrum line spacing cannot be resolved on the available spectrumanalyzer, the noise density function on most modern conventionalspectrum analyzers will directly measure the noise power densitynormalized to a 1-Hz noise power bandwidth. Add 35 dB for correction to3 kHz.        
Figure2 shows how the peak power spectral density can be measured. Since thespectral line spacing is greater than 3 kHz, no correction factor isrequired. There is no longer a requirement for the addition ofprocessing gain in the signal path, which was typically achieved by theimplementation of an encoding/decoding algorithm or “chip spreading.”         
Itis possible to transmit at a higher carrier power level while stillcomplying with FCC regulations. Figures 2 and 3 illustrate theindicated power spectral density and peak output power when measuredfollowing the procedures recommended by the FCC for a carrier modulatedwith a 76.8-kbit/s preamble data stream at a peak frequency deviationof 140 kHz.        
Whilea preamble data stream of alternating “1” and “0” bits is notparticularly representative of a normal data transmission, this exampledoes illustrate how implementing a digital modulation transmissionscheme in compliance with FCC 15.247 enables a much higher peak outputpower than the power spectral density to be transmitted.

It’s possible totransmit at a higher carrier power level while still complying with FCCregulations. Following the procedures recommended by the FCC for acarrier modulated with a 76.8-kbit/s preamble data stream at a peakfrequency deviation of 140 kHz yields this power spectral density andpeak output power.

        
Alternatively,the average power spectral density can be measured. First, center thespectrum analyzer on the emission peak(s) within the signal passband.Set the RBW to 3 kHz and VBW to greater than 9 kHz. The sweep timeshould be set to automatic. Next, the spectrum analyzer’s peak detectormode should be used.         
Asample detector may be employed providing that; (i) the bin width(i.e., frequency span/number of points in the spectrum display) is lessthan 0.5 RBW; (ii) the transmission pulse or sequence of pulses remainsat maximum transmit power throughout each of the 100 sweeps ofaveraging and the interval between pulses is not included in any of thesweeps (e.g., 100 sweeps should occur during one transmission, or eachsweep gated to occur during a transmission). If this condition cannotbe met, a peak detector set to max hold must be used.        
Selectvideo triggering and ensure that the trigger level only triggers ontransmitted pulses at the maximum power level. The transmitter mustoperate at maximum power for the entire sweep of every sweep. If thedevice transmits continuously, with no off intervals or reduced powerintervals, the trigger may be set to “free run.” Then, average morethan 100 sweeps and determine the peak from the resulting traceaverage. Ensure that the spectrum analyzer does not default to sampledetector mode in averaging mode.  
      
Practical ISM Band Device Measurements        
Itcan be noted that with a minimum RBW of 100 kHz and a VBW greater orequal to the RBW, the indicated peak-to-peak signal bandwidth is muchgreater than the twice the dynamic single sideband bandwidth thatdefines the minimum receive filter bandwidth (BBW). The dynamicsingle-sideband bandwidth (BBWSSB) is defined as (Equation 1):
        
Forthe example illustrated above, the dynamic BBWSSB of a 76.8-kbit/s datastream modulated at a peak frequency deviation of 140 kHz is 178.4 kHz.Thus, the double-sideband bandwidth is twice this value, or 356.8 kHz.        
Soto obtain a measured 6-dB signal bandwidth of 500 kHz, it is not alwaysnecessary to transmit at a frequency deviation in excess of 200 kHz. Atransceiver with a wide range of programmable receive filter bandwidthsis ideally suited to optimizing receiver bandwidth to transmittedspectrum, maximizing link budget.        
Notethat in the examples below, the VBW of the spectrum analyzer is setequal to the RBW for illustrative purposes only to highlight that theconfigured frequency deviation is the maximum required to ensurecompliance with FCC regulations.        
Table1 shows the results obtained for the transmitted 6-dB bandwidth, powerspectral density, and receiver sensitivity of the Semtech SX1231transceiver with a two-level frequency-shift keyed (FSK) transmissionagainst the requirements for both FCC Part 15.249 and in compliancewith the digital modulation system requirements of FCC Part 15.247. Ashas been noted, the example of power spectral density measurementsusing a preamble data stream is not representative of typical datatransmissions. For the purposes of the analysis below, a PN15pseudorandom data stream has been used.        
Athigher data rates, the mode of operation will typically comply with therequirements of FCC Part 15.247. Table 2 provides an example of linkbudgets obtained with the SX1231. Note that the actual power spectraldensity will depend upon the content or bit-pattern of the data streamtransmitted.
Conclusions        
ModernISM band transceiver radios can be used to increase the output powerabove the classical FCC Part 15.249 conducted output power limitationof typically –1 dBm without using complex spread-spectrum techniquessuch as FHSS. Through the use of digital modulation, the higher the bitrate, the higher the advantage of using such radios in terms of linkbudget.
In addition, the use of data whitening can increase theoutput power limit still further (while maintaining the spectral powerdensity limits outlined above). The data whitening process distributes1 and 0 FSK patterns equally throughout the transmission spectrum.         
Additionally,coding gain could drastically improve the budget link. For example, asa low-BR application, 1024-bit coding could improve the budget link by30 dB if BR is decreased by the same factor.
Further Information        
Thelatest published version of Part 15 of Title 47 of the Code of FederalRegulations can be obtained from the National Archives and RecordsAdministration Code of Federal Regulations online at www.access.gpo.gov/nara/cfr/cfr-table-search.html#page1.        
FCCOffice of Engineering and Technology, “Measurement of DigitalTransmission Systems Operating under Section 15.247,” March 23, 2005:
  http://gullfoss2.fcc.gov/prod/oet/forms/blobs/IDBretrieve.cgi?attachment_id=20422        
Semtech  Application Note AN1200.04, “FCC Regulations for ISM Band Devices: 902 - 928  MHz”:
  www.semtech.com/pc/downloadDocument.do?navId=H0,C1,P3593&id=1525        
FCCOffice of Engineering and Technology, “Understanding the FCCRegulations for Low-Power, Non-Licensed Transmitters (OET Bulletin63),” February 1996:
  www.fcc.gov/Bureaus/Engineering_Technology/Documents/bulletins/oet63/oet63rev.pdf        
The  Federal Communications Commission Web site can be found at www.fcc.gov/.        
The  Office of Engineering and Technology of the FCC home page is available at www.fcc.gov/oet/.

Steve Jillings is asenior RF applications engineer at Semtech. He has more than 20 yearsexperience, the past 10 of which have been specialized in the field oflow-power RF design and applications. He can be reached atsjillings@semtech.com.