radar level measurement - wordpress.com...horn antenna (solids) 139 b. radar level installation...

��

��

��

��

��

��

��

��

��

��

Radar level measurement- The users guide

Peter Devine

© VEGA Controls / P Devine / 2000All rights reseved. No part of this book may reproduced in any way, or by any means, without priorpermissio in writing from the publisher:VEGA Controls Ltd, Kendal House, Victoria Way, Burgess Hill, West Sussex, RH 15 9NF England.

British Library Cataloguing in Publication Data

Devine, PeterRadar level measurement - The user´s guide1. Radar2. Title621.3´848

ISBN 0-9538920-0-X

Cover by LinkDesign, Schramberg.Printed in Great Britain at VIP print, Heathfield, Sussex.

written byPeter Devine

additional informationKarl Grießbaum

type setting and layoutLiz Moakes

final drawings and diagramsEvi Brucker

Foreword ixAcknowledgement xiIntroduction xiii

Part I1. History of radar 12. Physics of radar 133. Types of radar 33

1. CW-radar 332. FM - CW 363. Pulse radar 39

Part II4. Radar level measurement 47

1. FM - CW 482. PULSE radar 543. Choice of frequency 624. Accuracy 685. Power 74

5. Radar antennas 771. Horn antennas 812. Dielectric rod antennas 923. Measuring tube antennas 1014. Parabolic dish antennas 1065. Planar array antennas 108Antenna energy patterns 110

6. Installation 115A. Mechanical installation 115

1. Horn antenna (liquids) 1152. Rod antenna (liquids) 1173. General consideration (liquids) 1204. Stand pipes & measuring tubes 1275. Platic tank tops and windows 1346. Horn antenna (solids) 139

B. Radar level installation cont. 1411. safe area applications 1412. Hazardous area applications 144

Contents

The benefits of radar as a level mea-surement technique 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 accu-racy is unaffected by changes in densi-ty, conductivity and dielectric constantof the product being measured or by airmovement above the product.

The practical use of microwaveradar for tank gauging and process ves-sel level 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 for

a 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 ves-sel contents information?

Currently there are two measure-ment techniques in common use forprocess vessel contents measurement.They are frequency modulated continu-ous wave (FM - CW) radar and PULSEradar

In this chapter we explain FM - CWand PULSE radar level measurementand compare the two techniques. Wediscuss accuracy and frequency consid-erations and explore the technicaladvances that have taken place inrecent years and in particular two wire,loop powered transmitters.

47

4. Radar level measurement

radar_applied_to_level_rb.qxd 15.01.2007 18:46 Seite 47

The 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 super-tankers. Subsequently, the same tech-nique was used for custody transferlevel 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, f1 and f2,and the difference between the trans-mitted signal and the return echosignal, fd, is measured. This differencefrequency is directly proportional to the

transit 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-tical problems that need to beaddressed in process level applications.

An FM - CW radar level transmitterrequires a voltage controlled oscillator,VCO, to ramp the signal between thetwo transmitted frequencies, f1 and f2.It is critical that the frequency sweep iscontrolled and must be as linear as pos-sible. A linear frequency modulation isachieved either by accurate frequencymeasurement circuitry with closed loopregulation of the output or by carefullinearisation of the VCO output includ-ing temperature compensation.

48

FM-CW, frequency modulated continuous wave fr

eque

ncy

f2

f1t1

∆ t

fd

time

Transmitted signal

Receivedsignal

Fig 4.1 The FM - CW radar technique is an indirect method of level measurement.fd is proportional to ∆∆t which is proportional to distance


49


��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

�� !

��"��

��

��!

��!

��#��!

��!

��#��!

Fig

4.2

Typ

ical

blo

ck d

iagr

am o

f F

M -

CW

rada

r. A

very

acc

urat

e lin

ear

swee

p is

req

uire

d

FM - CW block diagram (Fig 4.2)The essential component of a fre-

quency modulated continuous waveradar is the linear sweep control cir-cuitry. 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 and

hence towards the product in the ves-sel. The received echo frequencies aremixed with a part of the transmissionfrequency signal. These difference fre-quencies are filtered and amplifiedbefore Fast Fourier Transform (FFT)analysis is carried out. The FFTanalysis produces a frequency spec-trum on which the echo processing andecho decisions are made.

Simple storage applications usuallyhave a large surface area with very lit-tle agitation, 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 reac-tors where there is agitation and lowdielectric liquids.

Solids applications can be trouble-some because of the internal structureof the silos and undulating product sur-faces which creates multiple echoes.

An FM - CW radar level sensortransmits and receives signals simulta-neously.

50

Pic 2 Typical glass linedagitated processvessel. A radarmust be able tocope with variousfalse echos fromagitatior bladesand baffles


51


In an active process vessel, the vari-ous echoes are received as frequencydifferences compared with the frequen-cy of the transmitting signal. These fre-quency difference 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 high-er amplitude than the real level echo.The system needs to separate and iden-tify these 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 proce-

dure which converts the jumbled arrayof difference frequencies in the timedomain into a frequency spectrum inthe frequency domain.

The relative amplitude of each fre-quency component 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.

Fig 4.3a FM - CW radar level transmitters in an active process vessel

Transmitted signal

f2

fd1, -f-fd2d2, -fd3, -fd4, -fd5

f1t1

Real echo signal

False echo signals


52

Fig 4.3b combined echo frequencies are received simultaneously

Fig 4.3c The individual frequencies must be separated fromthe simultaneously received jumble of frequencies

Sig

nal

am

plit

ud

eS

ign

al a

mp

litu

de

Mixture of frequencies received by FM - CW radar

Combination of mixed difference frequencies received by FM - CW radarIndividual difference frequencies fd1, ffd2d2, fd3, are shown

fd1, fd2, fd3, fd4, fd5 etc combined


Complex 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 are

higher than the real echo. The proces-sors that carry out the FFT analysis areswamped by different amplitude sig-nals across 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


Fig 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 curve

Frequency spectrum echoesEach echo is within an envelope curve of frequencies

amp

litu

de

frequency


PULSE radar level transmitters

54

Pulse radar level transmitters pro-vide distance measurement based onthe direct measurement of the runningtime of microwave pulses transmittedto and reflected from the surface of theproduct being measured.

Pulse radar operates in the timedomain and therefore it does notrequire the Fast Fourier transform(FFT) analysis that characterizes FM -CW radar.

As already discussed, the runningtime for a distance of a few metres ismeasured in nanoseconds. For this rea-son, a special time transformation pro-

cedure is required to enable these shorttime periods to be measured accurately.The requirement is for a ‘slow motion’picture of the transit time of themicrowave pulses with an expandedtime axis. By slow motion we meanmilliseconds instead of nanoseconds.

Pulse radar has a regular and period-ically repeating signal with a high pulserepetition frequency (PRF). Using amethod of sequential sampling, theextremely fast and regular transit timescan be readily transformed into anexpanded time signal.

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 signal


A 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 of

sequential sampling is applied topulse radar level measurement. Theexample shown is a VEGAPULS trans-mitter with a microwave frequency of5.8 GHz.

55


To 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 sam-pled into a memory at a time period T2

which 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.

Fig 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 signal

Fig 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 curve

Periodic Signal(sine wave)

Periodic Signal(radar echoes)

Samplingsignal

Samplingsignal

Expandedtime signal

T1

T2

T1

Emission pulse

Echo pulse

T2


ExampleThe 5.8 GHz, VEGAPULS radar level transmitter has the following pulse repeti-tion rates.

Transmit pulse 3.58 MHz T1 = 279.32961 nanosecondsReference pulse 3.58 MHz - 43.7 Hz T2 = 279.33302 nanoseconds

56

Therefore the time expansion factoris 81920 giving a time expanded pulserepetition period of 22.88 milliseconds.

There is a practical problem in sam-pling the emission / echo pulse signalsof a short (0.8 nanosecond) pulse at 5.8GHz. An electronic switch would needto open and close within a few picosec-onds if a sufficiently short value of the5.8 GHz sine wave is to be sampled.These would have to be very specialand expensive components.

The solution is to combine sequen-tial sampling with a ‘cross correlation’procedure.

Instead of very rapid switch sam-pling, a sample signal of exactly thesame profile is generated but with aslightly longer time period between thepulses.

Fig 4.9 compares sequential sam-pling by rapid switching with sequen-tial sampling by cross correlation witha sample pulse.

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 dura-tion is shown in Fig 4.8.

The period of the pulse repetition isshown as T1 in Fig 4.7. Period T1 is

the same for the emission pulse repeti-tion as for any echo pulse repetition asshown.

However, the sampling signalrepeats at period of T2 which is slight-ly longer 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 deter-mined by T1 / (T2-T1).

Fig 4.8 Emission pulse (packet).The wave form of the 5.8GHz pulse with a pulseduration of 0.8nanoseconds


Instead of taking a short voltagesample, cross correlation involves mul-tiplying a 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 the

signal 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 corre-sponds directly 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


Fig 4.9 Comparison of switch sampling with ‘cross correlation’ sampling. The pulseradar uses cross correlation with a sample pulse. This means that rapid ‘picosec-ond’ switching is not required

Sampling with picosecond switching

Sample signal

Emission / Echo pulse

Sampling by cross correlation with asample pulse


The 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 easi-ly within analogue circuits. Simple, butgood quality components such as diodemixers for multiplication and capaci-tors for integration are used.

This method transforms the highfrequency received signal into an accu-rate picture with a considerablyexpanded time axis. The raw valueoutput from the microwave module isan intermediate frequency that is simi-lar to 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

Fig 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 curve

E

IntegralE x M

max

min

0

M

E x M


59

4. Radar level measurementam

plitu

de

transmit pulse

t1 tt22 t3 t4 t5

time

Pulse 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 com-pensation within the microwave mod-ule circuits.

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 particular-ly important 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 small-er echoes to be detected. If the pulseradar is manufactured with welldesigned circuits containing good qual-ity electronic components they candetect echoes over a wide dynamicrange of about 80 dB. This can makethe difference between reliable andunreliable measurement.

Fig 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 analysed

Pulse echoes in a process vessel are separated in time


60

Fig

4.1

1 B

lock

dia

gram

of

PU

LSE

rad

ar m

icro

wav

e m

odul

e


61


Pulse block diagram - (Fig 4.11)The raw pulse output signal (inter-

mediate frequency) 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 for-mer generates pulses with a frequencyof 3.58 MHz. A second variable oscil-lator and pulse former is tuned to a

frequency of 3.58 MHz minus 43.7 Hzand hence a slightly longer periodictime. GaAs FET oscillators are used toproduce the microwave carrier fre-quency of 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 anten-na are amplified and mixed with thesample pulses to produce the raw, timeexpanded, intermediate frequency.

Part of the measurement pulse sig-nal is used as a reference pulse thatprovides automatic temperature com-pensation of the microwave moduleelectronics.

Pic 3 Two wire pulse radar level transmitter mounted in a process reactor vessel


Process 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, low

frequency radar and every frequencyradar in between.

In reality, no single frequency isideally suited for every radar levelmeasurement application. If we com-pare 5.8 GHz radar with 26 GHz radar,we can see the relevant benefits of highfrequency and low frequency radar.

62

Choice of frequency

Fig 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 frequencies

2.6 GHz

5.8 GHz


The 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 approxi-mately the 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 anten-na as well as being inversely propor-tional to the square of the wavelength.

Antenna gain is proportional to:-

Antenna gain also depends on the aper-ture efficiency of the antenna.Therefore the beam angle of a smallantenna at a high frequency is notnecessarily as efficient as the equiva-lent beam angle of a larger, lower fre-quency radar. A 4" horn antenna radarat 6 GHz gives excellent beam focus-ing.

A full explanation of antenna gainand beam angles at different frequen-cies is given in Chapter 5 on radarantennas.

63


Focusing at different frequencies

5 GHz 10 GHz 15 GHz 20 GHz 25 GHz

Fig 4.13 For a given size of antenna, a higher frequency gives a more focused beam

Antenna size - beam angle

diameterwavelength

2

2


A 26 GHz beam angle is morefocused but, in some ways, it has to be.

The wavelength of a 26 GHz radar isonly 1.15 centimetres compared with awavelength of 5.2 centimetres for a5.8 GHz radar.

The short wavelength of the 26 GHzradar means that it will reflect off many

small objects that may be effectivelyignored by the 5.8 GHz radar. Withoutthe focusing of the beam, the high fre-quency radar would have to cope withmore false echoes than an equivalentlower frequency radar.

Antenna focusing and false echoes

Fig 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 silo

Fig 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 thisproblem

64


High frequency radar transmittersare susceptible to signal scatter fromagitated surfaces. This is due to the sig-nal wavelength in comparison to thesize of the surface disturbance.

The high frequency radar willreceive considerably less signal than anequivalent 5.8 GHz radar when the liq-

uid surface is agitated. The lowerfrequency transmitters are less affectedby agitated surfaces.

It is important that, whatever the fre-quency, the radar electronics and echoprocessing software can cope with verysmall amplitude echo signals. As dis-cussed, pulse radar has an advantage inthis area no matter what the frequency.

65


Agitated liquids and solid measurement

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 levelapplications

Condensation and build upHigh frequency radar level transmit-

ters are more susceptible to condensa-tion and product build up on the anten-na. There is more signal attenuation atthe higher frequencies, such as 26 GHz.Also, the same level of coating or con-densation on a smaller antenna natural-ly has a greater effect on the perfor-mance.

A 6" horn antenna with 5.8 GHz fre-quency is virtually unaffected by con-densation. Also, it is more forgiving ofproduct build up.

Steam and dustLower frequencies such as 5.8 GHz

and 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.


FoamThe effect of foam on radar signals

is a grey area. It depends a great dealon the type of foam including the foamdensity, dielectric constant and conduc-tivity. However, low frequencies suchas 5.8 GHz and 6.3 GHz cope with lowdensity foam better than higher fre-quencies such 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 impor-tant to provide the radar manufacturerwith as much information as possibleon the application.

Minimum distanceHigher frequency radar sensors have

a reduced minimum distance whencompared with the lower frequencies.This can be an additional benefit when measuring in small vessels and stillingtubes.

Fig 4.17 Focusing and radar frequency

Summary of the effects of radar frequency

focu

sing

Better focusing at higher emitting frequency means:

. higher antenna gain (directivity). less false echoes. reduced antenna size


frequency range

66



Fig 4.19 Signal strength from agitated and undulating surfaces and radar frequency

redu

ced

sign

al c

ause

d by

dam

ping

refle

ctio

n fr

om m

ediu

m

Reduced signal strength caused bydamping at higher emitting frequencycaused by:

. condensation. build - up. steam and dust

Higher damping caused by agitatedproduct surface

. wave movement. material cones with solids. signal scattered


frequency range


frequency range

67

Fig 4.18 Signal damping and radar frequency


68

AccuracyThere is no inherent difference in

accuracy between the FM - CW andPULSE radar level measurement tech-niques.

In this book, we are concernedspecifically with process level mea-surement where ‘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 trans-fer level 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 compensa-tion are also used.

Range resolution andbandwidth

In process level applications, bothFM - CW and PULSE radar work withan ‘envelope curve’. The length of thisenvelope curve depends on the band-width of 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 trans-mitters.

Pulse radar bandwidthThe carrier frequency of a pulse

radar varies from 5.8 GHz to about26 GHz.

The pulse duration is importantwhen it comes to resolving two adja-cent echoes. For example, a onenanosecond pulse has a length of about300 mm. Therefore, it would be diffi-cult for 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 a

curve.The shape of this curve is shown in

Fig 4.21. The null to null bandwidth BWnn of

a pulse radar is equal to

where τ is the pulse duration. It is clear from the curve that the

amplitude of frequencies reduces sig-nificantly away from the main pulsefrequency.

sin xx

2τ



Fig 4.20 Pulse radar range resolution.The guaranteed rangeresolution is the length of thepulse. A shorter pulse has awider bandwidth and betterrange resolution

shorter pulse

better range resolution

bandwidth BW nn,equal to

pulse frequency5.8 GHz

6.8 GHz4.8 GHz

Fig 4.21 The null to null bandwidthBWnn of a radar pulse is equalto 2 / ττ where ττ is the pulseduration. Example a 5.8 GHzradar with a pulse duration ofone nanosecond has a null tonull bandwidth of 2 GHz

Fig 4.22 Envelope curve with pulse radar

Pulse radar envelope curveFig 4.22 shows how a pulse radar

echo curve is used in process levelmeasurement.

A higher frequency pulse with ashorter pulse duration will allow betterrange resolution and also better accura-cy because the leading edge of theenvelope curve is steeper.

2τ

Fig 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 steeper

High frequency, shortduration pulse

Lower frequency pulse withlonger duration

69


70

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 narrower difference frequency ranges for each echo on the frequency spectrum. Thisleads to better range resolution in thesame way as with shorter duration puls-es with pulse radar.This is explained in the following dia-grams and equations.

frequency

amplitude

fd

∆fd

fd

frequency

timeTs

∆F

Fast Fourier Transform

fd =∆F x 2RTs x c

[Eq. 4.1]

∆F bandwidthTs sweep timeR distancefd difference frequencyc speed of light

The FAST FOURIERTRANSFORM produces afrequency spectrum of all echoessuch as that at fd.There is an ambiguity ∆fd for eachecho fd.

∆fd

[Eq. 4.2]

=2Ts


71


From equation 4.3, it can be seenthat with an FM - CW radar the rangeresolution ∆R is equal to:-

Therefore, 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 fre-quency range ∆fd for each echo withinan envelope curve. This translates intorange ambiguity.

amplitude

distance∆R

R

Fig 4.24 to 4.26 - FM - CW range resolution

The ambiguity of the distance R,is ∆ R

∆RR =

∆fdfd

∆RR =

2

∆F x 2 RTs

Ts x c

∆RR =

c∆F x R

∆R = c∆F

[Eq. 4.3]

c∆F


Other influences on accuracyAs we have demonstrated, FM - CW

and PULSE process radar transmittersuse an envelope curve for measure-ment. A wider bandwidth produces bet-ter range resolution. The correspond-ingly short echo will have a steep slopeand therefore a more accurate measure-ment can 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 distur-bance of the real echo curve leading toinaccuracies in the measurement.

Choice of antenna and mechanicalinstallation are important factors inensuring that the optimum accuracy isachieved.

FM - CW frequency spectrum - bandwidth and range resolution

Frequency spectrum - narrow bandwidth of linear sweep

envelope curvesaround echoes

envelope curvesaround echoes

frequency

amplitude

amplitude

Frequency spectrum - wide bandwidth of linear sweep

Fig 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 sweep

frequency

72


73


Fig 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 storagetanks

High 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 discuss-es how a higher accuracy can beachieved.

Pulse radarFor most process applications, mea-

surement relative to the pulse envelopecurve is sufficient. However, if the liq-uid level 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 applica-tions.

FM - CW radarThe fundamental requirement for an

accurate FM - CW radar is an accuratelinear sweep of the frequency modula-tion.

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 charac-teristic of 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 compensa-tion.

Fig 4.30 It is essential that the linearsweep of the FM - CW radar isaccurately controlled

frequency error

f2

f2t1


74

Microwave powerRadar is a subtle form of level mea-

surement. 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 dura-tion and pulse repetition frequency ofpulse radar transmitters.

An increase in the microwave powerwill produce higher amplitude echoes.However, it will produce higher ampli-tude false 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 of

processing power in order to function.This processing power is used to calcu-late the 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 trans-mitter.

Pulse radar transmitters work in thetime domain without FFT analysis andtherefore they do not need powerfulprocessors for this function.

SafetyThe low power output from

microwave radar transmitters meansthat they are an extremely safe methodof level measurement.

Pulse radarThe low energy requirements of

pulse 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 condi-tions. 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 trans-mitter is capable of operating at only14 VDC. This allows it to directlyreplace existing two wire sensors.

Pulsed FM - CWThe low power requirements of

pulse 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.

Power Two wire, loop powered radar


75


Summary of radar level techniques

FM - CW (frequency modulated continuous wave) radar · Indirect method of level measurement · Requires Fast Fourier Transform (FFT) analysis to convert signals into a fre-quency spectrum

· 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 applications


radar level measurement - wordpress.com...horn antenna (solids) 139 b. radar level installation...

Documents