7/31/2019 Thay Sen 2005

1/5

It has been demonstrated3,4 that the high-est advantage is gained by placing the inductorat the center of each antenna arm, instead ofat the input. In this article, the results of in-vestigations regarding both the location of theinductor as well as the inductance value arepresented. For many practical applications, itis more suitable to place the inductor almostat the input. In this way, no inductor is locatedon the antenna element itself, but rather onthe supporting structure or on the groundplane. In fact, the main objective of this articleis to present the results of numerical and ex-perimental investigations of the size reductionof a PIFA by the use of a lumped inductor. Asan intermediate target, the aim is to reducethe center frequency by 33 percent for a fixedphysical size of the antenna. The antenna per-formance, in terms of the radiation properties,

JESPER THAYSENNokia Denmark

Kbenhavn, DenmarkKAJ B. JAKOBSENTechnical University of Denmark

Lyngby, Denmark

The demand for smaller communicationdevices for personal communicationsystems has led to a constant search formethods to reduce the cellular phone dimen-sions. However, the wavelength does not de-crease with the same speed as the size of themobile phones due to the higher frequencybands used. Even a quarter wavelength anten-na, such as the planar inverted-F antenna(PIFA) tends to become too large, thus creat-ing a demand to decrease the volume of theantenna. Size reduction can be accomplishedsimply by shortening the antenna. However, atlengths shorter than the resonant length, theradiation resistance changes, and the imped-ance at the terminals of the antenna becomereactive. The latter can be compensated for bythe use of one or more inductors connected inseries with the antenna for cancellation of thecapacitance, thus improving the impedancematch1 and ultimately the efficiency.2 Theidea of using a lumped inductor in conjunc-tion with an antenna has often been used in

connection with low frequency antennaswhere the physical size might be several hun-dred meters.3 To date, however, it has foundvery little use in mobile telephony.4

A SIZE REDUCTIONTECHNIQUE FOR MOBILEPHONE PIFA ANTENNASUSING LUMPED INDUCTORSA size reduction technique for the planar inverted-F antenna (PIFA) is presented.

An 18 nH lumped inductor is used in addition to a small 0.3 cm3 PIFA. The PIFA

is located on dielectric foam, 5 mm above a 40 mm 100 mm ground plane. It is

possible to reduce the center frequency (|S11|min) by 33 percent for a fixed

physical size. The measured 6 dB bandwidth is 6.7 percent with a peak

radiation efficiency of 88 percent.

Reprinted with permission ofMICROWAVE JOURNAL from the July 2005 issue.2005 Horizon House Publications, Inc.

7/31/2019 Thay Sen 2005

2/5

scattering parameters, electrical near-field distribution and current distrib-ution, is simulated and verified bymeasurements.5-8 The evaluation ofthe antennas in terms of the electrical

field distribution and current distrib-ution on the antenna element as wellas on the ground plane has been ac-complished using planar near-field

measurements.912 The near-field isusually transformed to far-field data.Nevertheless, it is the raw un-processed near-field data that is pre-sented and used in this article.

MATERIALS AND METHODS

The antenna configuration consid-ered consists of a 40 mm long, 1.5mm wide and 5 mm high PIFA locat-ed on a 40 mm 100 mm groundplane. In all the prototypes, Rohacellmaterial (r = 1.06) is used as the sup-porting structure of the antenna. Theantenna is located at the edge andparallel to the 100 mm edge, as illus-trated inFigure 1. The feed point islocated 5 mm from the edge where a90 bend forms the short to theground plane. In the cases where theinductor is incorporated on the an-tenna element, a 0.5 mm wide gap iscut in the antenna arm. In order todetermine the optimal set-up with re-spect to the antenna performance,the location of the inductor is varied.Hence, the cut is moved from almostat the feed point, the 0.5 mm case,towards the open end, the 33 mmcase.

A planar scanner is used to per-form the measurements.7 The step

size is 4 mm leading to a total of 496measurement points for a 60 mm 120 mm area. This area covers theground plane plus an additional 10mm on each side of the groundplane.9 A three-dimensional E-fieldprobe is used for these measure-ments. The probe is designed forelectrical near-field component mea-surements up to 3 GHz.8 The mea-surements are carried out at 1.06GHz, that is, the measured centrefrequency (|S11|min) of both the 40

mm loaded as well as the 60 mm un-loaded antenna. The measurementfacility gives the total amplitude ofthe electrical fields. These measure-ments are compared to results ob-tained from the IE3D computer pro-gram used.6

In the planar scanning techniquethe probe is moved in a plane situat-ed in front of the antenna and the re-ceived signal (amplitude) is recorded.The position of the probe is charac-terised by the coordinates (x, y, z0) in

the xyz coordinate system of the an-tenna. During the scanning, z0 is keptconstant, while x and y is varied. Thefield is measured at a distance z0 =

3.2 mm, which corresponds to a freespace distance ofo/90, equivalent toan electrical length of 4. It should benoted that the distance between theground plane and the measurementplane is 8.2 mm (o/35), since the an-tenna height is 5 mm.

THE UNLOADED PIFAIn order to validate the perfor-

mance of the loaded antennas, an un-loaded prototype having the same di-mension has been fabricated (L W H) = (40 mm 1.5 mm 5 mm).This antenna has a somewhat highercentre frequency compared to theloaded antennas. Therefore, a largerunloaded antenna that has the samecentre frequency (|S11 |min) as theloaded antenna is also presented (60mm 1.5 mm 5 mm). In this way, amore realistic comparison can bemade.

The simulated and measured re-flection coefficient and radiation effi-ciency for two unloaded antennas, 40mm and 60 mm, are shown inFigure

2. For the 60 mm PIFA the simulat-ed centre frequency (|S11|min) is 1.2GHz, 33 percent lower than the cen-tre frequency (|S11|min) for the 40 mmlong PIFA, which is 1.8 GHz.

For both antennas, the measuredfrequencies with the lowest reflectioncoefficient are approximately 10 per-cent lower than the simulated results.This difference could be caused by aslight difference in the simulatedmodel and the prototype. The resolu-tion used in the simulation can alsocause some discrepancy. Here, con-verged results are obtained using 20cells per wavelength and edge cells.6

The measured total electrical fieldcomponents of the radiation patterns

for the 60 mm unloaded PIFA, showninFigure 3, indicate good agreementbetween the simulated and the mea-sured results. Note that the radiationpatterns are obtained at the centrefrequency (|S11|min); therefore, thesimulated patterns are at 1.23 GHzand the measured ones at 1.06 GHz.The measured maximum gain is 3.9dBi, slightly higher than the simulat-ed gain.

The total electrical near-field distri-bution at a distance of z0 = 3.2 mm

above the antenna element is shown inFigure 4. Good agreement in terms ofpeak amplitude and shape of the elec-trical near-field distribution between

TECHNICAL FEATURE

(a)

(b)

1.5 1.6 1.7 1.8 1.9

0

6

12

18

FREQUENCY(GHz)

100

67

33

0REFLECTION

COEFFICIENT

dB

EFFICIENCY(%)

1.0 1.1 1.2 1.3

0

6

12

18

FREQUENCY(GHz)

100

67

33

0REFLECTION

COEFFICIENT

dB

EFFICIENCY(%)

Measured Simulated

v Fig. 2 Simulated and measuredreflection coefficient and radiation efficiencyfor the unloaded 40 mm (a) and the unloaded60 mm (b) long PIFAs.

(a)

(b)

Feed Point

Ground

ShortW

H

L

33mm fromfeed pointcut0mm

=90 =0

=90 =0

=0

+

vFig. 1 Schematic of the PIFA locatedabove a ground plane (a), and antenna

orientation in spherical coordinates (b).

7/31/2019 Thay Sen 2005

3/5

the simulated and the measured re-

sults are obtained; however, more de-tails could be observed from the simu-lated result. For instance, when ob-serving just above the radiatingelement, especially at (x, y) = (0 ,4070), a local minimum is found andthe edge peak radiation is higher abovethe ground when compared to that ob-tained at the side of the ground plane.This is also in accordance with themeasured radiation pattern shown pre-viously. In both cases, the peak valuesare associated with the open end of the

antenna, that is at (x, y) = (0, 40). Also,in both cases a minimum is observed at(x, y) = (40, 40), that is at the oppositeedge of the PIFA.

Either an inductor or a capacitorcan be used to reduce the centre fre-quency (|S11|min) from 1.8 GHz as forthe 40 mm long PIFA to 1.2 GHz.This frequency reduction corre-sponds to a size reduction of 33 per-cent, that is from 60 mm to 40 mm.

INDUCTOR-LOADED PIFATwo different tests are made.

First, for a fixed location of the in-ductor, the inductance is varied be-tween 5 and 100 nH. Hereafter, theoptimal location is found for a fixedinductance value.

Numerical Results

Locating an inductor 10 mm fromthe feed point forms the inductorloading. For this fixed location, thesimulation results for varying the in-ductor value between 5 and 100 nHare shown in Figure 5. Here, the

centre frequency (|S11|min) and thebandwidth are plotted with respect tothe 40 mm unloaded case. The centrefrequency (|S11|min) drops from 1.8GHz towards 0.87 GHz for inductorvalues above 70 nH. For values above35 nH, however, the bandwidth islower than for the unloaded PIFA.This motivates the choice of an in-ductor value below 35 nH. Using 5nH, the bandwidth is 2.3 times thebandwidth for the unloaded PIFA,which is due to the improved imped-

ance match. Between 5 and 35 nH,the optimal inductor value is a trade-off between the decrease in centrefrequency (|S11|min) and the actual

bandwidth. As acompromise, 20 nHis chosen for therest of the work.Here, the centrefrequency (|S11|min)is lowered by 30

percent and thebandwidth is almosttwice the band-width obtained forthe unloaded 40mm case.

For a fixed in-ductance of 20 nH,various locations ofthe inductor havebeen simulated,spanning from al-most at the feedpoint (0.5 mm) to-ward the open end

(33 mm). The simulated centre fre-quency (|S11|min) and relative band-width as a function of the location ofthe inductor, that is the distance fromthe feed point to the inductor, areshown in Figure 6. The lowest re-flection coefficient and the peak effi-ciency as a function of the inductorlocation are shown inFigure 7.

TECHNICAL FEATURE

Measured Simulated

180

135 135

4545

0

9090

(a)

180

135 135

4545

0

9090

(b)

180

225 135

45315

0

90270

(c)

5dB

0

510

15

5dB

0

510

15

5dB

0

510

15

=180 =0

=180 =0

v Fig. 3 Total electrical field componentsof the radiation patterns for the 60 mmunloaded PIFA; (a) cuts for =0, (b) for =90 and (c) cut for =90.

Total E field

60 mm, measured0 20 40

60 mm, simulated0 20 40

(v/m), (dB) Total E field (v/m), (dB)

100

80

60

40

20

0

100

80

60

40

20

0

50

45

40

35

30

25

50

45

40

35

30

25

v Fig. 4 Measured and simulated total electrical near-fielddistribution for the 60 mm long unloaded PIFA at 1.06 GHz.

0 25 50 75 100

1.0

0.8

0.6

0.4

INDUCTOR VALUE (nH)

3

2

1

0

fo/fo

unloaded

BW/BWunloaded

v Fig. 5 Center frequency and bandwidthas a function of inductance for a fixedlocation (10 mm) on the PIFA.

0 5 10 15 20 25 30 35

2.0

1.8

1.6

1.4

1.2

1.0

DISTANCE FROM THE FEEDPOINT TO THE INDUCTOR (mm)

15

12

9

6

3

0CENTERFREQUENCY

GHz

RELATIVEBANDWIDTH(%)

v Fig. 6 Simulated center frequency andbandwidth versus the inductor location.

7/31/2019 Thay Sen 2005

4/5

The centre frequency (|S11|min) in-creases almost linearly from 1.2 to 1.8GHz, when the inductor is moved to-wards the open end, from a position at

0.5 mm to 33 mm from the feed point.Starting with a 45 MHz or 4 percentbandwidth at 1.2 GHz (0.5 mm) thebandwidth drops due to mismatch, forlocations in the range from 5 to 20mm; hence, no 6 dB bandwidth oc-curs. The maximum bandwidth of 14.5percent is obtained at 26 mm, and sta-bilizes around 10 percent when the in-ductor is located at positions near theopen end of the antenna (30 mm to 33mm).

The peak efficiency starts at 75

percent and ends at 60 percent, closeto the feed point (0.5 mm) and theopen end (33 mm), respectively. Atlocations below 5 mm, the efficiencyis higher than 75 percent. Above 5mm, a decrease in the efficiency isobserved, and the efficiency is below50 percent from 10 to 30 mm. Above31 mm, the reflection coefficient is11 dB and the efficiency exceeds 50percent, and reaches a peak of 60percent at 33 mm. In between, theefficiency has dropped to 17 percent

at the 21 mm location.The lowest reflection coefficient

changes from 9 to 1.5 dB when thedistance from the feed point to the in-ductor increases from 0.5 to 20 mm.From 21 to 26 mm the reflection co-efficient peaks at 30 dB, ending at11 dB for locations above 30 mm.

Experimental Results

Based on the results from the pa-rameter study, a prototype 40 mmlong PIFA, loaded with an inductor,

has been measured with respect toradiation efficiency and reflection co-efficient. A lumped 18 nH inductor isused in the experiments.13 The in-

ductor has an inductance of nearly 20nH in the frequency range of inter-est. The results, compared to the un-loaded 60 mm long PIFA, are showninFigure 8.

For the PIFA without any induc-tor, the centre frequency (|S11|min) is1.06 GHz with a peak return loss of16.5 dB. The bandwidth is 7.3 per-cent (78 MHz). The measured effi-ciency is greater than 65 percentwithin this frequency range of inter-est, where the peak efficiency is 85percent. By loading this antenna withan 18 nH inductor soldered at thegap, just 0.5 mm from the feed point,the centre frequency (|S11|min) is 1.07GHz, with a bandwidth of 6.7 per-cent (71 MHz). The peak efficiency is88 percent.

The total electrical field compo-nents of the radiation patterns, showninFigure 9, indicate almost omni-di-rectional properties for the 40 mm longinductor-loaded PIFA. There is goodagreement between the simulation andmeasurement results. The measuredmaximum gain is 3.3 dBi, slightly high-

er than the simulated gain.In Figure 10, the measured totalelectrical field distribution of the 40mm long, 18 nH inductor-loadedPIFA is compared with the one of the60 mm long, unloaded PIFA. It indi-cates a higher amplitude above theinductor-loaded antenna, comparedto the 60 mm one. This originatesfrom the higher current distributionthat is inevitably present, and thushigher radiation from this area.

In the lower half of the pictures,

the electrical field distribution be-haves identically, with slightly highervalues for the 60 mm case. Also, thenull that appears at the opposite side

(x, y) = (40, 50) of the PIFA in the 60

mm case could be found in the 40mm case.The PIFA is basically an inverted-L

antenna that actually comes from amonopole with a bend such that mostof the arm is parallel to the groundplane. This means that the feed point ismoved by a certain distance from theground, here 5 mm from the bend andan additional 5 mm due to the antennaheight; hence, the optimum location ofthe inductor is between 10.5 mm and15 mm from the ground connection,

that is almost one-third the total lengthof 45 mm (length plus height). Collin4

argues that the optimum location of aninductor is at the centre of the arm of

TECHNICAL FEATURE

0 5 10 15 20 25 30 35

0

10

20

30

DISTANCEFROMTHEFEEDPOINTTOTHEINDUCTOR(mm)

100

67

33

0REFLEC

TION

COEFFICIENT

dB

RADIATION

EFFICIEN

CY(%)

v Fig. 7 Simulated reflection coefficientand radiation efficiency versus the inductorlocation.

1.00 1.05 1.10 1.15

0

6

12

18

FREQUENCY(GHz)

100

67

33

0REFLECTION

COEFFICIENT

dB

RADIATION

EFFICIENCY(%)

v Fig. 8 Comparison of the measuredreflection coefficient and efficiency of a 60mm long, unloaded PIFA (dashed line) and40 mm long PIFA loaded with an inductor(solid line).

Measured Simulated

180

135 135

4545

0

9090

(a)

180

135 135

4545

0

9090

(b)

180

=180 =0

=180 =0

225 135

45315

0

90270

(c)

5dB

0

510

15

5dB

0

510

15

5dB

0

510

15

v Fig. 9 Simulated and measured totalelectrical field components of the radiationpattern; (a) cuts for =0, (b) cuts for

=90 and (c) cuts for =90.

7/31/2019 Thay Sen 2005

5/5

the monopole; of course, that cannotbe compared directly to the PIFA.Nevertheless, this actually holds for theimpedance match. If the inductor is lo-cated between 21 and 26 mm, a rathergood simulated impedance match be-low 25 dB is observed. In this case thedecrease in the frequency, with thelowest reflection coefficient, is notoverwhelming, a reduction from only1.8 to 1.7 GHz. Moreover, the radiationefficiency is below 25 percent. Thiscould indicate that the optimum loca-tion for an inductor in the PIFA is clos-er to the feed point.

Above 21 mm no significant fre-quency reduction is obtained. At 30

mm, however, the bandwidth is 200MHz (13 percent), which is higherthan the case of no inductor (50 MHzor 3 percent). Thus, the higher band-width is at the expense of an inductorin terms of reduced efficiency andthe cost of the inductor.

CONCLUSIONA small 0.3 cm3 PIFA, located on di-

electric foam 5 mm above a 40 mm 100 mm ground plane, is investigated.Adding an inductor on the arm of the

PIFA improves the performances forthe shown PIFA. The best case with re-spect to centre frequency (|S11|min) re-duction is obtained when an 18 nHlumped inductor is placed within the

first few millimetres from the feedpoint. Here, the measured frequencypoint with the lowest reflection coeffi-cient is decreased by 33 percent, from1.60 to 1.06 GHz, the reflection coeffi-cient is 16.5 dB, the measured 6 dBbandwidth is 6.7 percent and the radia-

tion peak efficiency is 88 percent.When comparing the 40 mm in-

ductor-loaded antenna with the 60mm unloaded antenna, the majorbenefit includes the reduced size fora fixed centre frequency (|S11|min).This, however, comes at the expenseof reduced efficiency and bandwidth.

By the use of inductor loading, it isshown that for a fixed size the centrefrequency (|S11|min) can be decreased.The principle could also be used for afixed frequency, where a 30 to 40percent size reduction is expected.

The PIFAs shown are not fully op-timized with respect to the occupiedvolume or frequency, nor is any othercomponent or cover included. Thus,for practical use, both the shape ofthe antenna and the shape of the dis-tributed inductor should be carefullychosen in order to get the best fre-quency bandwidth and efficiency per-formance for a given application. s

ACKNOWLEDGMENTThis work has been supported by

Nokia Denmark.

References1. C.W. Harrington, Monopole with Induc-

tive Loading, IEEE Transactions on An-te nn as an d Pr op ag at io n, July 1963,pp. 394400.

2. G.S. Smith, Efficiency of ElectricallySmall Antennas Combined with MatchingNetworks, IEEE Transactions on Anten-nas and Propagation, Vol. 25, May 1977,pp. 369373.

3. G. Hall (Ed.), The ARRL Antenna Book,

15th Edition, Chapter 7, The AmericanRadio Relay League, Newington, CT,1990.

4. R.E. Collin, Antenna s and RadiowavePropagation, McGraw-Hill, New York, NY,1985, pp. 97104.

5. www.satimo.com.6. www.zeland.com.7. www.semcad.com.8. www.speag.com.9. A.W. Rudge (Ed.), The Handbook of An-

tenna Design, Chapter 8, Antenna Mea-surements, Peter Pelegrinus Ltd., 1982.

10. R.C. Baird, A.C. Newell and C.F. Stuben-rauch, A Brief History of Near-field Mea-surements of Antennas at the National Bu-reau of Standards, IEEE Transactions onAntennas and Propagation, Vol. 36, No. 6,1988.

11. Y. Gao and I. Wolff, A New MiniatureMagnetic Field Probe for MeasuringThree-dimensional Fields in Planar HighFrequency Circuits, IEEE Transactionson Microwave Theory and Techniques, Vol.44, No. 6, June 1996, pp. 911918.

12. Y. Gao and I. Wolff, Miniature ElectricNear-field Probe for Measuring 3-D Fieldsin Planar Microwave Circuits, IEEETransactions on Microwave Theory andTechniques, Vol. 46, No. 7, July 1998,

pp. 907913.13. www.coilcraft.com.

Jesper Thaysenreceived his MScdegree from theTechnical University ofDenmark, Kgs.Lyngby, Denmark, in2000. He is currentlyworking toward hisPhD degree. He hasbeen with NokiaDenmark since 2001.His current interests

include broadband antennas, small antennas

and multi-elements antennas.Kaj B. Jakobsenreceived his MScEEdegree from theTechnical University ofDenmark, Kgs.Lyngby, Denmark, in1986, his PhD degreefrom the University ofDayton, OH, in 1989,and his graduatediploma in businessadministration,

organization and management fromCopenhagen Business School, Denmark, in

2000. His research interests include GPR,antennas, microwaves, electromagnetics andrelated interdisciplinary topics. Since 1990, hehas been a professor at the TechnicalUniversity of Denmark, from where hereceived the Teacher of the Year award in1994.

TECHNICAL FEATURE

v Fig. 10 Measured current distributionfor the inductor loaded versus unloaded.

(v/m), (dB) (v/m), (dB)Total E field Total E field

60 mm, unloaded 40 mm, unloaded

100

80

60

40

20

0

100

80

60

40

20

0

60

55

50

45

40

35

30

60

55

50

45

40

35

30

0 20 40 0 20 40