Radar level measurement Radar level measurement The user's guide

Radar levelmeasurementThe user's guide

4. Radar level measurementThe benefits of radar as a level measurementtechnique are clear.Radar provides a non-contact sensorthat is virtually unaffected by changesin process temperature, pressure or thegas and vapour composition within avessel.In addition, the measurement accuracyis unaffected by changes in density,conductivity and dielectric constantof the product being measured or by airmovement above the product.The practical use of microwaveradar for tank gauging and process vessellevel measurement introduces aninteresting set of technical challengesthat have to be mastered.If we consider that the speed of lightis approximately 300,000 kilometresper second. Then the time taken fora radar signal to travel one metreand back takes 6.7 nanoseconds or0.000 000 006 7 seconds.How is it possible to measure thistransit time and produce accurate vesselcontents information?Currently there are two measurementtechniques in common use forprocess vessel contents measurement.They are frequency modulated continuouswave (FM - CW) radar and PULSEradarIn this chapter we explain FM - CWand PULSE radar level measurementand compare the two techniques. Wediscuss accuracy and frequency considerationsand explore the technicaladvances that have taken place inrecent years and in particular two wire,loop powered transmitters.47

FM-CW, frequency modulated continuous waveThe FM - CW radar measurementtechnique has been in use since the1930's in military and civil aircraftradio altimeters. In the early 1970's thismethod was developed for marine usemeasuring levels of crude oil in supertankers.Subsequently, the same techniquetransit time and hence the distance.(Examples of FM - CW radar leveltransmitters modulation frequencies are8.5 to 9.9 GHz, 9.7 to 10.3 GHz and 24to 26 GHz).The theory of FM - CW radar issimple. However, there are many prac-was used for custody transfer tical problems that need to belevel measurement of large land basedstorage vessels. More recently, FM -CW transmitters have been adapted forprocess vessel applications.FM - CW, or frequency modulatedcontinuous wave, radar is an indirectmethod of distance measurement. Thetransmitted frequency is modulatedbetween two known values, f 1 and f 2 ,and the difference between the transmittedsignal and the return echosignal, f d , is measured. This differencefrequency is directly proportional to theaddressed in process level applications.An FM - CW radar level transmitterrequires a voltage controlled oscillator,VCO, to ramp the signal between thetwo transmitted frequencies, f 1 and f 2 .It is critical that the frequency sweep iscontrolled and must be as linear as possible.A linear frequency modulation isachieved either by accurate frequencymeasurement circuitry with closed loopregulation of the output or by carefullinearisation of the VCO output includingtemperature compensation.f 2Transmitted signalfrequency∆ tf dReceivedsignalf 1t 1timeFig 4.1 The FM - CW radar technique is an indirect method of level measurement.f d is proportional to ∆t which is proportional to distance48

4. Radar level measurementVoltage Controlled Oscillator VCODirectionalCouplerDirectionalCouplerf (t + Dt)V(t)Linear sweepcontrol loopf (t + Dt)VoltageControlFrequencyMeasurementf(t)MixerFilterLinear ramp generatorIntermediatefrequencyAmplifierFront end control functionSignal samplingandFast Fourier transforms (FFT)Signal MicroprocessorFig 4.2 Typical block diagram of FM - CW radar. A very accurate linear sweep is requiredf(t)49

FM - CW block diagram (Fig 4.2)The essential component of a frequencymodulated continuous waveradar is the linear sweep control circuitry.A linear ramp generator feeds avoltage controller which in turn rampsup the frequency of the VoltageControlled Oscillator. A very accuratelinear sweep is required. The outputfrequency is measured as part of theclosed loop control.The frequency modulated signal isdirected to the radar antenna andhence towards the product in the vessel.The received echo frequencies aremixed with a part of the transmissionfrequency signal. These difference frequenciesare filtered and amplifiedbefore Fast Fourier Transform (FFT)analysis is carried out. The FFTanalysis produces a frequency spectrumon which the echo processing andecho decisions are made.Pic 2 Typical glass linedagitated processvessel. A radarmust be able tocope with variousfalse echos fromagitatior bladesand bafflesSimple storage applications usuallyhave a large surface area with very littleagitation, no significant false echoesfrom the internal structure of the tankand relatively slow product movement.These are the ideal conditions forwhich FM - CW radar was originallydeveloped.However, in process vessels there ismore going on and the problemsbecome more challenging.Low amplitude signals and falseechoes are common in chemical reactorswhere there is agitation and lowdielectric liquids.Solids applications can be troublesomebecause of the internal structureof the silos and undulating product surfaceswhich creates multiple echoes.An FM - CW radar level sensortransmits and receives signals simultaneously.50

4. Radar level measurementf d1 , -f d2 , -f d3 , -f d4 , -f d5f 2f 1t 1Transmitted signalReal echo signalFalse echo signalsFig 4.3a FM - CW radar level transmitters in an active process vesselIn an active process vessel, the variousechoes are received as frequencydifferences compared with the frequencyof the transmitting signal. These frequencydifference signals are receivedby the antenna at the same time. Theamplitude of the real echo signals aresmall compared with the transmittedsignal. A false echo from the end of theantenna may have a significantly higheramplitude than the real level echo.The system needs to separate and identifythese simultaneous signals beforeprocessing the echoes and making anecho decision.The separation of the variousreceived echo frequencies is achievedusing Fast Fourier Transform (FFT)analysis. This is a mathematical procedurewhich converts the jumbled arrayof difference frequencies in the timedomain into a frequency spectrum inthe frequency domain.The relative amplitude of each frequencycomponent in the frequencyspectrum is proportional to the size ofthe echo and the difference frequencyitself is proportional to the distancefrom the transmitter.The Fast Fourier Transform requiressubstantial processing power and is arelatively long procedure.It is only when the FFT calculationsare complete that echo analysis can becarried out and an echo decision can bemade between the real level echo and anumber of possible false echoes.51

Mixture of frequencies received by FM - CW radarf d1 , f d2 , f d3 , f d4 , f d5 etc combinedSignal amplitudeFig 4.3b combined echo frequencies are received simultaneouslyCombination of mixed difference frequencies received by FM - CW radarIndividual difference frequencies f d1 , f d2 , f d3 , are shownSignal amplitudeFig 4.3c The individual frequencies must be separated fromthe simultaneously received jumble of frequencies52

4. Radar level measurementFrequency spectrum echoesEach echo is within an envelope curve of frequenciesamplitudefrequencyFig 4.4 FM - CW frequency spectrum after Fast Fourier transform. The Fast Fouriertransform algorithm converts the signals from the time domain into the frequencydomain. The result is a frequency spectrum of the difference frequencies. Therelative amplitude of each frequency component in the spectrum is proportional tothe size of the echo and the difference frequency itself is proportional to thedistance from the transmitter. The echoes are not single frequencies but a spanof frequencies within an envelope curveComplex process vessels and solidsapplications can prove too difficult forsome FM - CW radar transmitters.Even a simple horizontal cylindricaltank can pose a serious problem. Thisis because a horizontal tank producesmany large multiple echoes that arecaused by the parabolic effect of thecylindrical tank roof. Sometimes theamplitudes of the multiple echoes arehigher than the real echo. The processorsthat carry out the FFT analysis areswamped by different amplitude signalsacross the dynamic range all at thesame time. As a result, the FM - CWradar cannot identify the correct echo.As we shall see, this problem doesnot affect the alternative pulse radartechnique.53

PULSE radar level transmittersPulse radar level transmitters providecedure is required to enable these shortdistance measurement based on time periods to be measured accurately.the direct measurement of the running The requirement is for a ‘slow motion’time of microwave pulses transmitted picture of the transit time of theto and reflected from the surface of the microwave pulses with an expandedproduct being measured.time axis. By slow motion we meanPulse radar operates in the time milliseconds instead of nanoseconds.domain and therefore it does not Pulse radar has a regular and periodicallyrequire the Fast Fourier transformrepeating signal with a high pulse(FFT) analysis that characterizes FM - repetition frequency (PRF). Using aCW radar.method of sequential sampling, theAs already discussed, the running extremely fast and regular transit timestime for a distance of a few metres is can be readily transformed into anmeasured in nanoseconds. For this reason,expanded time signal.a special time transformationpro-Fig 4.5 Pulse radar operates purely within the time domain. Millions of pulses aretransmitted every second and a special sampling technique is used to produce a‘time expanded’ output signal54

4. Radar level measurementTo illustrate this principle, considerthe sine wave signal in Fig 4.6. It is aregular repeating signal with a periodof T1. If the amplitude (voltage value)of the output of the sine wave is sampledinto a memory at a time period T2T1which is slightly longer than T1, then atime expanded version of the originalsine wave is produced as an output.The time scale of the expanded outputdepends on the difference between thetwo time periods T1 and T2.PeriodicSignal(sine wave)SamplingsignalT2Expandedtime signalFig 4.6 The principle of sequential sampling with a sine wave as an example.The sampling period, T2, is very slightly longer than the signal period, T1. Theoutput is a time expanded image of the original signalA common example of this principleis the use of a stroboscope to slowdown the fast periodic movements ofrotating or reciprocating machinery.Fig 4.7 shows how the principle ofsequential sampling is applied topulse radar level measurement. Theexample shown is a VEGAPULS transmitterwith a microwave frequency of5.8 GHz.PeriodicSignal(radar echoes)T1EmissionpulseEchopulseT2SamplingsignalFig 4.7 Sequential sampling of a pulse radar echo curve. Millions of pulses per secondproduce a periodically repeating signal. A sampling signal with a slightly longerperiodic time produces a time expanded image of the entire echo curve55

This periodically repeating signalconsists of the regular emission pulseand one or more received echo pulses.These are the level surface and anyfalse echoes or multiple echoes. Thetransmitted pulses and therefore thereceived pulses have a sine wave formdepending upon the pulse duration. A5.8 GHz pulse of 0.8 nanosecond durationis shown in Fig 4.8.The period of the pulse repetition isshown as T1 in Fig 4.7. Period T1 isthe same for the emission pulse repetitionas for any echo pulse repetition asshown.However, the sampling signalrepeats at period of T2 which is slightlylonger in duration than T1. This isthe same time expansion procedure bysequential sampling that has alreadybeen described for a sine wave. Thefactor of the time expansion is determinedby T1 / (T2-T1).Fig 4.8 Emission pulse (packet).The wave form of the 5.8GHz pulse with a pulseduration of 0.8nanosecondsExampleThe 5.8 GHz, VEGAPULS radar level transmitter has the following pulse repetitionrates.Transmit pulse 3.58 MHzReference pulse 3.58 MHz - 43.7 HzTherefore the time expansion factoris 81920 giving a time expanded pulserepetition period of 22.88 milliseconds.There is a practical problem in samplingthe emission / echo pulse signalsof a short (0.8 nanosecond) pulse at 5.8GHz. An electronic switch would needto open and close within a few picosecondsif a sufficiently short value of the5.8 GHz sine wave is to be sampled.These would have to be very specialand expensive components.T1 = 279.32961 nanosecondsT2 = 279.33302 nanosecondsThe solution is to combine sequentialsampling with a ‘cross correlation’procedure.Instead of very rapid switch sampling,a sample signal of exactly thesame profile is generated but with aslightly longer time period between thepulses.Fig 4.9 compares sequential samplingby rapid switching with sequentialsampling by cross correlation witha sample pulse.56

4. Radar level measurementEmission / Echo pulseSample signalSampling with picosecond switchingSampling by cross correlation with asample pulseFig 4.9 Comparison of switch sampling with ‘cross correlation’ sampling. The pulseradar uses cross correlation with a sample pulse. This means that rapid ‘picosecond’switching is not requiredInstead of taking a short voltagesample, cross correlation involves multiplyinga point on the emission or echosignal by the corresponding point onthe sample pulse. The multiplicationleads to a point on the resultant signal.All of these multiplication results, oneafter the other, lead to the formation ofthe complete multiplication signal.Fig 4.10 shows a short sequence ofmultiplications between the receivedsignal (E) and the sampling pulsesignal (M). The resultant E x M curvesare shown on page 58.Then the E x M curve is integratedand represented on the expanded curveas a dot. The sign and amplitude of thesignal on the time expanded curvedepends on the sum of the area of theE x M curve above and below the zeroline. The final integrated value correspondsdirectly to the time position ofthe received pulse E relative to thesample pulse M.The received signal E and samplesignal M in Fig 4.10 are equivalent tothe periodic signal (sine wave) andsample signal in Fig 4.6. The result ofthe integration of E x M in Fig 4.10 isdirectly analogous to the expandedtime signal in Fig 4.6.57

EME x MmaxIntegralE x M0minFig 4.10 Cross correlation of the received signal E and the sampling M.The product E x M is then integrated to produce the expanded time curve. Thetechnique builds a complete picture of the echo curveThe pulse radar sampling procedureis mathematically complicated but atechnically simple transformation toachieve. Generating a reference signalwith a slightly different periodic time,multiplying it by the echo signal andintegration of the resultant product areall operations that can be handled easilywithin analogue circuits. Simple, butgood quality components such as diodemixers for multiplication and capacitorsfor integration are used.This method transforms the highfrequency received signal into an accuratepicture with a considerablyexpanded time axis. The raw valueoutput from the microwave module isan intermediate frequency that is similarto an ultrasonic signal. For examplethe 5.8 GHz microwave pulse becomesan intermediate frequency of 70 kHz.The pulse repetition frequency (PRF)of 3.58 GHz becomes about 44 Hz.58

4. Radar level measurementPulse echoes in a process vessel are separated in timeamplitudet 1 t 2 t 3 t 4 t 5timetransmit pulseFig 4.12 With a pulse radar, all echoes (real and false) are separated in time. This allowsmultiple echoes caused by reflections from a parabolic tank roof to be easilyseparated and analysedPulse radar operates entirely withinthe time domain and does not need thefast and expensive processors thatenable the FM - CW radar to function.There are no Fast Fourier Transform(FFT) algorithms to calculate. All ofthe pulse radar processing is dedicatedto echo analysis only.Part of the pulse radar transmissionpulse is used as a reference pulse thatprovides automatic temperature compensationwithin the microwave modulecircuits.The echoes derived from a pulseradar are discrete and separated in time.This means that pulse radar is betterequipped to handle multiple echoes andfalse echoes that are common inprocess vessels and solids silos.Pulse radar takes literally millions of‘shots’ every second. The return echoesfrom the product surface are sampledusing the method described above. Thistechnique provides the pulse radar withexcellent averaging which is particularlyimportant in difficult applicationswhere small amounts of energy arebeing received from low dielectric andagitated product surfaces.The averaging of the pulse techniquereduces the noise curve to allow smallerechoes to be detected. If the pulseradar is manufactured with welldesigned circuits containing good qualityelectronic components they candetect echoes over a wide dynamicrange of about 80 dB. This can makethe difference between reliable andunreliable measurement.59

60Fig 4.11 Block diagram of PULSE radar microwave module

4. Radar level measurementPulse block diagram - (Fig 4.11)The raw pulse output signal (intermediatefrequency) from the pulseradar microwave module is similar, infrequency and repetition rate, to anultrasonic signal.This pulse radar signal is derived inhardware. Unlike FM - CW radar,PULSE does not use FFT analysis.Therefore, pulse radar does not needexpensive and power consumingprocessors.The pulse radar microwave modulegenerates two sets of identical pulseswith very slightly different periodictimes. A fixed oscillator and pulse formergenerates pulses with a frequencyof 3.58 MHz. A second variable oscillatorand pulse former is tuned to afrequency of 3.58 MHz minus 43.7 Hzand hence a slightly longer periodictime. GaAs FET oscillators are used toproduce the microwave carrier frequencyof the two sets of pulses.The first set of pulses are directedto the antenna and the product beingmeasured. The second set of pulses arethe sample pulses as discussed in thepreceding text.The echoes that return to the antennaare amplified and mixed with thesample pulses to produce the raw, timeexpanded, intermediate frequency.Part of the measurement pulse signalis used as a reference pulse thatprovides automatic temperature compensationof the microwave moduleelectronics.Pic 3 Two wire pulse radar level transmitter mounted in a process reactor vessel61

Choice of frequencyProcess radar level transmittersoperate at microwave frequenciesbetween 5.8 GHz and about 26 GHz.Manufacturers have chosen frequenciesfor different reasons ranging fromlicensing considerations, availability ofmicrowave components and perceivedtechnical advantages.There are arguments extolling thevirtues of high frequency radar, lowfrequency radar and every frequencyradar in between.In reality, no single frequency isideally suited for every radar levelmeasurement application. If we compare5.8 GHz radar with 26 GHz radar,we can see the relevant benefits of highfrequency and low frequency radar.2.6 GHz5.8 GHzFig 4.14 Comparison of 5.8 GHz and 26 GHz radar antenna sizes. These instrumentshave almost identical beam angles. However this is not the full picture when itcomes to choosing radar frequencies62

4. Radar level measurementAntenna size - beam angleThe higher the frequency of a radarlevel transmitter, the more focused thebeam angle for the equivalent sizeantenna.With horn antennas, this allowssmaller nozzles to be used with a morefocused beam angle.For example, a 1½" (40 mm) hornantenna radar at 26 GHz has approximatelythe same beam angle as a 6"(150 mm) horn antenna at 5.8 GHz.However, this is not the completepicture. Antenna gain is dependent onthe square of the diameter of the antennaas well as being inversely proportionalto the square of the wavelength.Antenna gain is proportional to:-Antenna gain also depends on the apertureefficiency of the antenna.Therefore the beam angle of a smallantenna at a high frequency is notnecessarily as efficient as the equivalentbeam angle of a larger, lower frequencyradar. A 4" horn antenna radarat 6 GHz gives excellent beam focusing.A full explanation of antenna gainand beam angles at different frequenciesis given in Chapter 5 on radarantennas.2diameterwavelength2Focusing at different frequencies5 GHz 10 GHz 15 GHz 20 GHz 25 GHzFig 4.13 For a given size of antenna, a higher frequency gives a more focused beam63

Antenna focusing and false echoesA 26 GHz beam angle is more small objects that may be effectivelyfocused but, in some ways, it has to be. ignored by the 5.8 GHz radar. WithoutThe wavelength of a 26 GHz radar is the focusing of the beam, the high frequencyradar would have to cope withonly 1.15 centimetres compared with awavelength of 5.2 centimetres for a more false echoes than an equivalent5.8 GHz radar.lower frequency radar.The short wavelength of the 26 GHzradar means that it will reflect off manyFig 4.15 a Low frequency radar has a wider beamangle and therefore, if the installationis not optimum, it will see more falseechoes. Low frequencies such as5.8 GHz or 6.3 GHz tend to be moreforgiving when it come to false echoesfrom the internal structure of a vesselor siloFig 4.15 b High frequency radar has a muchnarrower beam angle for a givenantenna size. The narrower beam angleis important because the shortwavelength of the higher frequencies,such as 26 GHz, reflect more readilyfrom the internal structures such aswelds, baffles, and agitators.The sharper focusing avoids thisproblem64

4. Radar level measurementAgitated liquids and solid measurementHigh frequency radar transmitters uid surface is agitated. The lowerare susceptible to signal scatter from frequency transmitters are less affectedagitated surfaces. This is due to the signalwavelength in comparison to the It is important that, whatever the fre-by agitated surfaces.size of the surface disturbance.quency, the radar electronics and echoThe high frequency radar will processing software can cope with veryreceive considerably less signal than an small amplitude echo signals. As discussed,pulse radar has an advantage inequivalent 5.8 GHz radar when the liq-this area no matter what the frequency.Fig 4.16 High frequency radar transmitters aresusceptible to signal scatter fromagitated surfaces. This is due to thesignal wavelength in comparison to thesize of the surface disturbance. It isimportant that radar electronics andecho processing software can cope withvery small amplitude echo signals.By comparison, 5.8 GHz radar is not asadversely affected by agitated liquidsurfaces. Lower frequency radar isgenerally better suited to solid levelapplicationsCondensation and build upHigh frequency radar level transmittersare more susceptible to condensationand product build up on the antenna.There is more signal attenuation atthe higher frequencies, such as 26 GHz.Also, the same level of coating or condensationon a smaller antenna naturallyhas a greater effect on the performance.A 6" horn antenna with 5.8 GHz frequencyis virtually unaffected by condensation.Also, it is more forgiving ofproduct build up.Steam and dustLower frequencies such as 5.8 GHzand 6.3 GHz are not adversely affectedby high levels of dust or steam. Thesefrequencies have been very successfulin applications ranging from cement,flyash and blast furnace levels to steamboiler level measurement.In steamy and dusty environments,higher frequency radar will suffer fromincreased signal attenuation.65

FoamThe effect of foam on radar signalsis a grey area. It depends a great dealon the type of foam including the foamdensity, dielectric constant and conductivity.However, low frequencies suchas 5.8 GHz and 6.3 GHz cope with lowdensity foam better than higher frequenciessuch as 26 GHz.For example, a 26 GHz radar signalwill be totally attenuated by a very thindetergent foam on a water surface. A5.8 GHz radar signal will see throughthis type of foam and continue to seethe liquid surface as the foam thicknessincreases to 150 mm or even 250 mm.However, the thickness of foam willcause a small measurement errorbecause the microwaves slow downslightly as they pass through the foam.When foam is present, it is importantto provide the radar manufacturerwith as much information as possibleon the application.Minimum distanceHigher frequency radar sensors havea reduced minimum distance whencompared with the lower frequencies.This can be an additional benefit whenmeasuring in small vessels and stillingtubes.Summary of the effects of radar frequencyBetter focusing at higher emitting frequency means:. higher antenna gain (directivity). less false echoes. reduced antenna sizefocusing5 GHz 10 GHz 15 GHz 20 GHz 25 GHzFig 4.17 Focusing and radar frequencyfrequency range66

4. Radar level measurementreduced signal caused by dampingReduced signal strength caused bydamping at higher emitting frequencycaused by:. condensation. build - up. steam and dust5 GHz 10 GHz 15 GHz 20 GHz 25 GHzfrequency rangeFig 4.18 Signal damping and radar frequencyHigher damping caused by agitatedproduct surfacereflection from medium. wave movement. material cones with solids. signal scattered5 GHz 10 GHz 15 GHz 20 GHz 25 GHzfrequency rangeFig 4.19 Signal strength from agitated and undulating surfaces and radar frequency67

AccuracyThere is no inherent difference inaccuracy between the FM - CW andPULSE radar level measurement techniques.In this book, we are concernedspecifically with process level measurementwhere ‘process accurate’ andcost effective solutions are required.The achievable accuracy of aprocess radar depends heavily on thetype of application, the antenna design,the quality of the electronics and echoprocessing software employed.The niche market for custody transferlevel measurement applications isoutside the scope of this book. Thesecustody transfer radar ‘systems’ areused in bulk petrochemical storagetanks. Large parabolic or planar arrayantennas are used to create a finelyfocused signal. A lot of processingpower and on site calibration time isused to achieve the high accuracy.Temperature and pressure compensationare also used.Range resolution andbandwidthIn process level applications, bothFM - CW and PULSE radar work withan ‘envelope curve’. The length of thisenvelope curve depends on the bandwidthof the radar transmitter. A widerbandwidth leads to a shorter envelopecurve and therefore improved rangeresolution. Range resolution is one of anumber of factors that influence theaccuracy of process radar level transmitters.Pulse radar bandwidthThe carrier frequency of a pulseradar varies from 5.8 GHz to about26 GHz.The pulse duration is importantwhen it comes to resolving two adjacentechoes. For example, a onenanosecond pulse has a length of about300 mm. Therefore, it would be difficultfor the radar to distinguish betweentwo echoes that are less than 300 mmapart. Clearly a shorter pulse durationprovides better range resolution.An effect of a shorter pulse durationis a wider bandwidth or spectrum offrequencies.For example, if the carrier frequencyof a pulse is 5.8 GHz and the durationis only 1 nanosecond, then there is aspectrum of frequencies above andbelow the nominal carrier frequency.The amplitude of the pulse spectrum offrequencies changes according to asin xxcurve.The shape of this curve is shown inFig 4.21.The null to null bandwidth BW nn ofa pulse radar is equal to2τwhere τ is the pulse duration.It is clear from the curve that theamplitude of frequencies reduces significantlyaway from the main pulsefrequency.68

4. Radar level measurementpulse frequency5.8 GHz4.8 GHz6.8 GHzshorter pulsebetter range resolutionbandwidth BW nn,2equal to τFig 4.20 Pulse radar range resolution.The guaranteed rangeresolution is the length of thepulse. A shorter pulse has awider bandwidth and betterrange resolutionPulse radar envelope curveFig 4.22 shows how a pulse radarecho curve is used in process levelmeasurement.A higher frequency pulse with ashorter pulse duration will allow betterrange resolution and also better accuracybecause the leading edge of theenvelope curve is steeper.Fig 4.21 The null to null bandwidthBW nn of a radar pulse is equalto 2 / τ where τ is the pulseduration. Example a 5.8 GHzradar with a pulse duration ofone nanosecond has a null tonull bandwidth of 2 GHzFig 4.22 Envelope curve with pulse radarHigh frequency, shortduration pulseLower frequency pulse withlonger durationFig 4.23 A shorter pulse duration gives better range resolution. The combination ofshorter pulse duaration and higher frequency allows better accuracy because theleading edge of the envelope curve is steeper69

FM-CW radar bandwidthThe bandwidth of an FM - CW radar isthe difference between the start andfinish frequency of the linear frequencymodulation sweep.Unlike pulse radar, the amplitude of theFM - CW signal is constant across therange of frequencies.A wider bandwidth produces narrowerdifference frequency ranges for eachecho on the frequency spectrum. Thisleads to better range resolution in thesame way as with shorter duration pulseswith pulse radar.This is explained in the following diagramsand equations.frequencyf d =∆F x 2RT s x c[Eq. 4.1]∆F∆FT sRf dcbandwidthsweep timedistancedifference frequencyspeed of lightT stimeFast Fourier Transformamplitudef dThe FAST FOURIERTRANSFORM produces afrequency spectrum of all echoessuch as that at f d.There is an ambiguity ∆f d for eachecho f d.∆f d = 2 T s[Eq. 4.2]∆f df dfrequency70

4. Radar level measurementThe ambiguity of the distance R,is ∆ RamplitudeR∆RR=∆f df d∆Rdistance∆RR∆RR==2T s∆F x 2 RT s x cc∆F x RFig 4.24 to 4.26 - FM - CW range resolution∆R =c∆F[Eq. 4.3]From equation 4.3, it can be seenthat with an FM - CW radar the rangeresolution ∆R is equal to:-c∆FTherefore, the wider the bandwidth, thebetter the range resolution.Examples:A linear sweep of 2 GHz has a rangeresolution of 150 mm whereas a 1 GHzbandwidth has a range resolution of300 mm.In process radar applications, eachecho on the frequency spectrum isprocessed with an envelope curve. Theabove equations (Equations 4.1 to 4.3)show that the Fast Fourier Transforms(FFTs) in process radar applications donot produce a single discrete differencefrequency for each echo in the vessel.Instead they produce a difference frequencyrange ∆f d for each echo withinan envelope curve. This translates intorange ambiguity.71

FM - CW frequency spectrum - bandwidth and range resolutionFrequency spectrum - narrow bandwidth of linear sweepamplitudeenvelope curvesaround echoesFrequency spectrum - wide bandwidth of linear sweepamplitudefrequencyenvelope curvesaround echoesfrequencyFig 4.28 Illustration of envelope curve around the frequency spectram of FM - CWradars. The same four echoes are shown for radar transmitters with differentbandwidths. An improvement in the range resolution is achieved with a widerbandwidth of the linear sweepOther influences on accuracyAs we have demonstrated, FM - CWand PULSE process radar transmittersuse an envelope curve for measurement.A wider bandwidth produces betterrange resolution. The correspondinglyshort echo will have a steep slopeand therefore a more accurate measurementcan be made. Other influences onaccuracy include signal to noise ratioand interference.A high signal to noise ratio allowsmore accurate measurement whileinterference effects can cause a disturbanceof the real echo curve leading toinaccuracies in the measurement.Choice of antenna and mechanicalinstallation are important factors inensuring that the optimum accuracy isachieved.72

4. Radar level measurementHigh accuracy radarHigh accuracy of the order of+ 1 mm is generally meaningless in anactive process vessel or a solids silo.For example, a typical chemical reactorwill have agitators, baffles and otherinternal structures plus constantlychanging product characteristics.Although custody transfer levelmeasurement applications are not in thescope of this book, this section discusseshow a higher accuracy can beachieved.Pulse radarFor most process applications, measurementrelative to the pulse envelopecurve is sufficient. However, if the liquidlevel surface is flat calm and theecho has a reasonable amplitude, it ispossible to look inside the envelopecurve wave packet at the phase of anindividual wave.However, the envelope curve of ahigh frequency radar with a short pulseduration is sufficiently steep to producea very accurate and cost effective leveltransmitter for storage vessel applications.FM - CW radarThe fundamental requirement for anaccurate FM - CW radar is an accuratelinear sweep of the frequency modulation.As with the pulse radar, it is possibleto look inside the envelope curve of thefrequency spectrum if the applicationhas a simple single echo that is characteristicof a liquids storage tank. This isachieved by measuring the phase angleof the difference frequency. However,this is only practical with custodytransfer applications where fast andexpensive processors are used withtemperature and pressure compensation.f 2frequency errorf 2t 1Fig 4.30 It is essential that the linearsweep of the FM - CW radar isaccurately controlledFig 4.29 Higher accuracy of pulse radarlevel transmitters can beachieved by looking at the phaseof an individual wave within theenvelope curve. This is onlypractical in slow moving storagetanks73

PowerMicrowave powerRadar is a subtle form of level measurement.The peak microwave powerof most process radar level transmittersis less than 1 milliWatt. This level ofpower is sufficient for tanks and silosof 40 metres or more.The average power depends on thesweep time and sweep repetition rate ofFM - CW radar and on the pulse durationand pulse repetition frequency ofpulse radar transmitters.An increase in the microwave powerwill produce higher amplitude echoes.However, it will produce higher amplitudefalse echoes and ringing noiseas well as a higher amplitude echoesoff the product surface. The averagemicrowave power of a Pulse radar canbe as little as 1 microWatt.Processing powerFM - CW radars need a high level ofprocessing power in order to function.This processing power is used to calculatethe FFT algorithms that producethe frequency spectrum of echoes.The requirement for processing powerhas restricted the ability of FM - CWradar manufacturers to make a reliabletwo wire, intrinsically safe radar transmitter.Pulse radar transmitters work in thetime domain without FFT analysis andtherefore they do not need powerfulprocessors for this function.SafetyThe low power output frommicrowave radar transmitters meansthat they are an extremely safe methodof level measurement.Two wire, loop powered radarPulse radarThe low energy requirements ofpulse radar enabled the first ever twowire, loop powered, intrinsically saferadar level transmitter to be introducedto the process industry in mid-1997.The VEGAPULS 50 series of pulseradar transmitters have proved to bevery capable in difficult process conditions.The performance of the two wire,4 to 20 mA, sensors is equal to the fourwire units that preceded them.The pulse microwave module onlyneeds a 3.3 volt power supply witha maximum power consumption of50 milliWatts. This drops down to5 milliWatts when it is in stand-bymode. The difference between the twowire pulse and the four wire pulse isthat the two wire radar sends out burstsof pulses and updates the output aboutonce every second. The four wire sendsout pulses continuously and updatesseven times a second.With high quality electronics, thecomplete 24 VDC, 4 to 20 mA transmitteris capable of operating at only14 VDC. This allows it to directlyreplace existing two wire sensors.Pulsed FM - CWThe low power requirements ofpulse radar have allowed two wireradar to become sucessful. FM - CWradar requires processing power andtime for the FFT's to be calculated.Power saving has been used to producea ‘pulsed’ FM - CW radar. However,this device is limited to simple storageapplications because the update time istoo long and the processing too limitedfor arduous process applications.74

4. Radar level measurementSummary of radar level techniquesFM - CW (frequency modulated continuous wave) radar· Indirect method of level measurement· Requires Fast Fourier Transform (FFT) analysis to convert signals into a frequencyspectrum· FFT analysis requires processing power and therefore practical FM - CWprocess radars have to be four wire and not two wire loop powered· FM - CW radars are challenged by large numbers of multiple echoes (causedby the parabolic effects of horizontal cylindrical or dished topped vessels)PULSE radar· Direct, time of flight level measurement· Uses a special sampling technique to produce a time expanded intermediatefrequency signal· The intermediate frequency is produced in hardware and does not require FFTanalysis· Low processing power requirement mean that practical and very capable twowire, loop powered, intrinsically safe pulse radar can be used in some of themost challenging process level applications75