could still be picked up by Bluetooth receivers, which in general can have a sensitivity exceeding -70 dBm.

-100 -50 0 50 100-100 -50 0 50 100 angle, deg angle, deg

a b (552131

Fiz. 3 Simuluted radiation patterm in two orthogonal planes at 2.24 GHz

a sz-plane b yz-plane (i) LHCP (ii) RHCP

E -30 -35

6-40 H-45 Q -50 .$ -55 f -60

F 7 0 ~ -65

f -75 * -80

2.3 2.4 2.5 2.6 frequency, GHz

1552/41 U

Fig. 4 ‘4verage received power for line-of-sight ( T X I - R X ) and non-line- ofsight (TX2-RX) transmission in indoor environment

TXI-RX, parallel - - ~ TX2-RX, parallel _ _ _ _ TXI-RX, perpendicular . . . . . . . . . . . TX2-RX, perpendicular

Conclusions: We have presented a simple low-cost rectangular ring antenna for operation under the Bluetooth protocol in the 2 . 4 2.5GHz ISM-band. The dimensions of the antenna were opti- mised by use of a field simulator. Excellent agreement was found between the measured and simulated reflection characteristics of the antenna. The transmission characteristics of a transmitter- receiver pair proved to be sufficient in a realistic indoor environ- ment.

0 IEE 2001 Electronics Letters Online No: 20010812 DOI: 10.1049/e1:20010812

C. Vermeeren, H. Rogier, F. Olyslager and D. De Zutter (Depurtment of Infimnution Technology. Sint-Pietersnieuwstraat 41, 9000 Gent, Belgium)

E-mail: hendrikr@intec.rug.ac.be

3 July 2001

References

1 HAARTSEN, J C.: ‘Bluetooth radio systems’, IEEE Pers. Commun., 2000, 7, pp. 28-36

2 GARG, R., BHARTIA, P., BAHL, I , and ITTIPIBOON A.: ‘Microstrip antenna design handbook’ (Artech House Antennas and Propagation Library, Nonvood, 2000)

3 PALANISAMY, v., and GARG, R.: ‘Analysis of circularly polarized square ring and crossed-strip microstrip antennas’, ZEEE. Truns. Antennus Propag., 1986, 34, (ll), pp. 1340-1346

4 CHEN, w.-s., wu, c -K., and WONG, K.-L.: ‘Single-feed square-ring microstrip antenna with truncated corners for compact circular polarisation operation’, Electron. Lett., 1998, 34, (1 l), pp. 1045- 1047

Electrically tunable microwave phase shifter based on layered ferrite-ferroelectric structure

V. Demidov, P. Edenhofer and B. Kalinikos

A novel electrically tunable microwave phase shifter is proposed based on surface spin wave propagation in a layered structure containing ferrite and ferroelectric materials. Operating characteristics of the phase shifter are investigated theoretically. It is shown that waveguiding elements from planar layered ferrite- ferroelectric structures allow the construction of compact devices with high electrical tunability achieving 10”N.

Introduction: Ferroelectrics have recently been investigated as attractive materials for building electrically tunable microwave phase shifters [l - 31. One of the most important features of such phase shifters is their continuous tunability arising from the varia- tions of the dielectric constant of ferroelectric material under a DC electric field. This electrical tunability makes ferroelectric phase shifters attractive for technical applications such as steerable antenna arrays. Usually the principle of operation of ferroelectric phase shifters is based on varying the propagation constant of electromagnetic waves in microstrip transmission lines manufac- tured on ferroelectric substrates. Two serious disadvantages of these conventional microstrip ferroelectric phase shifters are that they cannot be tuned by a large phase range and they demand a relatively high DC voltage. To overcome these disadvantages we suggest the use of layered ferroelectric-ferrite structures as waveguiding elements to build up phase shifters. As shown in [4], the dispersion properties of spin waves propagating in such struc- tures depend on the dielectric constant of the ferroelectric layer due to hybridisation of their dispersion curve with that of electro- magnetic waves. At the same time, spin wave phase and group velocities are orders of magnitude smaller than those of electro- magnetic waves; consequently, their operating phase shift depends on the dielectric constant of the ferroelectric layer being much stronger than the phase shift of electromagnetic waves.

direction of bias magnetic field

H,

direction of wave propagation

/ v

ferroelectric slab

ferrite film

ectric strate

L b metal

screen

1003/11 Fig. 1 Geometry of Iuyered ferrite-ferroelectric waveguiding structure

Geometry und operation: The geometry of the layered waveguiding structure is shown in Fig. 1. It consists of a ferrite film of thick- ness L with dielectric constant E~ grown on a dielectric substrate of thickness b with dielectric constant E ~ , and of a ferroelectric slab of thickness a with dielectric constant E, placed on the ferrite film surface. The layered structure is metallised from the bottom side and covered at the top by an interdigital structure with dis- tance d between the fingers. The interdigital structure serves to create an electric field in the ferroelectric slab. The layered ferrite- ferroelectric structure is magnetised by a constant magnetic field of intensity H, oriented parallel to its plane and perpendicular to the direction of wave propagation. Excitation and reception of surface spin waves propagating in the described structure may be realised by using standard microstrip transducer techniques (see e.g. [5]). Spin waves excited by the input microstrip transducer propagate along the layered waveguiding structure and are received by the output microstrip transducer. During propagation within the ferrite-ferroelectric layered structure these waves accu- mulate a significant phase shift along a relatively short propaga- tion distance (several millimetres) owing to their slow phase

1154 ELECTRONICS LETTERS 13th September2C~ I ‘’01.37 No. 79

velocity. This phase shift can be efficiently controlled by tuning the spin wave dispersion characteristics through the DC voltage applied to the interdigital structure.

9'4 r

100 120 140 160 180 2 0 0 2 2 0 7

wave number, cm-l

[003/21 Fig. 2 Calculated dispersion characteristics of surface spin waves propa- gating in layered structure controlled by applied D C voltage varying jkom 0 to SO V

Curves correspond to values of 0, 10, 20, 30, 40 and 50V

Calculations: To demonstrate the advantages of the layered ferrite- ferroelectric structures for electrically tunable phase shifters we have calculated dispersion characteristics of surface spin waves using the dispersion relation derived in terms of a joint solution of the full system of Maxwell's equations and the linearised equation of motion for magnetisation [4]. The dispersion characteristics, shown in Fig. 2, correspond to different values of the DC voltage applied to the control interdigital structure. The numerical calcula- tions were performed using a geometry for the layered structure which is simple and most appropriate for practical realisation. Typical parameters for yttrium-iron garnet (YIG) films grown on gadolinium-gallium garnet substrates are used for the ferrite film, i.e. saturation magnetisation, 47cM0 = 1750G; L = 5 0 ~ ; E~ = 14; b = 3 0 0 ~ ; &b = 14. The thickness of the ferroelectric layer a is taken to be 1 5 0 ~ . The experimental dependence of the dielectric constant on the applied bias electric field is taken from [l] for ferroelectric ceramics Ba, 65Sr0 35Ti03. The distance d between the fingers of the interdigital structure is 15Opn and the intensity of the bias magnetic field is H0 = 2500 Oe.

As seen from Fig. 2, variations of the bias voltage applied to the ferroelectric layer lead to significant changes in the spin wave dispersion characteristics. The strongest influence occurs in the region of comparatively small wave numbers. With an increase of wave number, all the dispersion curves are approaching the disper- sion curve of magnetostatic surface waves which does not depend on the dielectric constant of the ferroelectric layer. We note that increasing the voltage from 0 to 50 V corresponds to a decrease of the ferroelectric dielectric constant only by 15%. Under such con- ditions a 20m-' displacement of the carrier wave number is achieved at a frequency of 8.8 GHz. This displacement corre- sponds to a change of about 1100" in the phase shift of the carrier wave along a propagation distance of 1 cm.

Fig. 3 shows the calculated phase shift frequency dependence of the surface spin wave taking the dispersion characteristics as described. The calculations suppose a propagation distance of 5 mm. As seen from this Figure, applying a 50 V control DC volt- age which corresponds to an intensity of the electric field between the interdigital fingers of about 3.3 kV/cm changes the phase shift of the operating spin wave by more than 27c in the frequency range of about 400 MHz. Note from the inset of Fig. 3 that the depend- ence of the phase shift on the applied voltage shows practically linear character over the whole range of control voltages. This is due to the small variation of the dielectric constant of the ferroe- lectric layer necessary for an appreciable tuning of the phase shift. The calculations show that the obtained linear character of this phase shift against voltage is maintained within the whole fre- quency range shown. The slope of the characteristic, which defines the eficiency of electrical tunability of this composite ferrite-ferro-

ELECTRONICS LETTERS 13th September 2001 Va

electric layered phase shifter, decreases from 10.5"N at 8.8 GHz to 9.4"N at 9.4 GHz, i.e. not more than 11% in frequency dispersion. We emphasise that the calculated efficiency of electrical tunability is orders of magnitude higher than that obtained for microstrip ferroelectric phase shifters [l - 31. Another important advantage of these ferrite-ferroelectric tunable structures in comparison with conventional transmission line techniques is due to the electric sep- aration of the microwave circuit and the DC control circuit.

38 -

18 I I I I I I I I

8.80 8.85 8.90 8.95 9.00 9.05 9.10 9.15 9.20 frequency, GHz

(003/31 Fig. 3 Calculated frequency dependence of phase shqt of surface spin wave under increase of applied D C voltage from 0 to 50 V

Inset: Phase shift against applied control voltage at 8.85GHz

Influence of geometrical parameters: The numerical calculations show that increasing the thickness of the ferrite layer leads to an increase of the frequency band where an appreciable phase shift occurs and leads to a reduction of the frequency dispersion of the eficiency of electrical tunability. However, this leads to a decrease in the average value of the efficiency of tunability. Thus, the thick- ness of ferrite film may be varied depending on a specific applica- tion in the range from 10 to 1 0 0 ~ to provide optimal performance. The thickness of the ferroelectric film together with the dielectric constant define the operational frequency range of the phase shifter. However, a significant increase in the thickness of the ferroelectric layer degrades the tunability owing to a decreasing penetration of the electric field into the ferroelectric slab.

Conclusions: A novel, promising construction of tunable ferrite- ferroelectric microwave phase shifters is suggested. The results of numerical calculations show that the ferrite-ferroelectric layered structures can be successfully used as waveguiding elements for microwave phase shifters. This would allow significant improve- ment in the performance of electrical tunability, to decrease the value of the controlling voltage, and to reduce the geometrical size of devices compared with conventional ferroelectric phase shifters.

Acknowledgments: This work was supported in part by the Inter- national Association INTAS, Grant No. 99-1812, by the Russian Foundation for Basic Research, Grant No. 99-02-16370, and by the German Academic Exchange Service (DAAD).

0 IEE 2001 Electronics Letters Online No: 20010746 DOI: IO. 1049/el:2001O746

V. Demidov and P. Edenhofer (Institute for High-Frequency Techniques, University of Bochunz, 0-44780 Bochum, Germany)

E-mail: edh@hf.ruhr-uni-bochum.de

B. Kalinikos (St Petersburg Electrotechnical University, ProJ Popov Str. 5, St Petersburg, 197376, Russia)

S July 2001

References

1 VARADAN, V.K., GHODGAONKAR, D.K., VARADAN, V.V., KELLY. J.F., and GLIKERDAS, P.: 'Ceramic phase shifters for electronically steerable antenna systems', Microw. J., 1992, 35, pp. 116-127

11. 37 No. 19 1155

2 LANCASTER, M.J., POWELL, J., and PORCH, A : ‘Thin-film ferroelectric microwave devices’, Supercond. Sci. Technol., 1998, 11, (1 I), pp. 1323-1334

ROMANOFSKY, R.R., WARNER, J.D., and MUELLER, c.H.: ‘Design and development of ferroelectric tunable microwave components for Ku- and K-band satellite communication systems’, ZEEE Trans. Microw. Theory Tech., 2000, 48, (7), pp. 1181-1189

4 DEMIDOV, v.E., and KALINIKOS, B.A.: ‘Spectra of exchange dipole electromagnetic-spin waves in asymmetric metal-insulator- ferromagnetic-insulator-metal systems’, Tech. Phys., 2001, 46, (2), pp. 219-222

5 ISHAK, w.s.: ‘Magnetostatic wave technology: a review’, Proc, ZEEE, 1988, 76, (2), pp, 171-187

3 MlRANDA, F.A., SUBRAMANYAM, G . , VAN KEULS, F.W.,

2-bit adder: carry and sum logic circuits at 19 GHz clock frequency in InAIAs/lnGaAs HBT technology

T. Mathew, S. Jaganathan, D. Scott, S. Krishnan, Y . Wei, M. Urteaga, M.J.W. Rodwell and S. Long

Carry and sum circuits for a 2-bit adder used in a pipelined 2N- bit adder-accumulator architecture are reported. To obtain high clock rates in a design with multiple gate delays a novel merged AND-OR-Latch structure using four-level series gated current steering logic is employed. These integrated circuits were fabricated in InAIAsLnGaAs transferred substrate heterojunction bipolar transistor technology and operate up to 19GHz clock rate.

Introduction: Adder-accumulators are used as phase accumulators in direct digital frequency synthesis (DDFS) systems [1 - 31. The DDFS frequency resolution is given by Af = hk/2N and the fre- quency tuning range is from DC to approximatelyhk/3 where N is the adder-accumulator digital word width [3] andhk the clock fre- quency [3]. Increasing the DDFS frequency range requires increas- ing the maximum clock frequency of the DDFS components, specifically the phase accumulator (adder-accumulator), Sine ROM, and digital-analogue converter (DAC) [3]. The work reported in this Letter is aimed at increasing the adder-accumula- tor clock rate.

High clock rates are achieved in pipelined adder-accumulators [3]. These circuits are complex because numerous latches are required for synchronisation between stages (Fig. 1) [3]. For adder-accumulators of 8 to 10 bits total resolution, a pipelined architecture using 2-bit adder blocks provides a good compromise between circuit complexity and clock speed (Fig. 1). In this Letter we describe design techniques employed to increase the clock rate of the carry and sum logic circuit of a 2-bit adder.

D

Ai

2 bitadder -t

D

28111 Fig. 1 Pipelined adder-accumulator architecture realised u s i n s b i t udder blocks R: register A: input word S : sum word C: carry bit

Circuit design: We now describe Boolean logic of the 2-bit adder (Fig. 1). To implement an accumulator, the 2-bit sum is fed back

as an adder input, i.e. Bi = Si and B,+I = carry input is C,. The sum (S,, are given by

The 2-bit adder and carry (C,+l, C,+*) terms

S, = Ai CE Bi CB C, Si+l = Ai+l CB Bi+l CE Ci+l

Ci+l = AiBi + AiC, + BiC, Ci+z = Ai+lBi+l + A,+lCi+l + Bi+lCi+1

Ci+l =AiBi+BiCi+AiCi . . . . .

CK

Q 1281/21 Fig. 2 Carry logic circuit of 2-bit udder

For testing, circuit is configured as frequ- divider by setting Bi = B,+l = High, A, = Ai+l = Low and C, = C1+2

I I I I I 100 200 300 400 500

-0.3 I 0

time, ps (281/31

Fig. 3 Output waveforms of carry logic circuit at 19 GHz clock rate

f,,, = 9.5GHz

Gate propagation delays in forming (SI, Sril) and (C,+l, Cl+*) determine the maximum clock frequency, with computation of Cl+* having the longest delay [3]. To reduce delays in evaluating (C[+,, CL+*), the AND-OR logic is realised as a single three-level series-gated ECL gate (Fig. 2). This has approximately 1.41 smaller gate delay than a carry logic circuit using standard two- level series-gated ECL logic, which requires two cascaded gates. The clock frequency is further increased by merging the logic eval- uation and latching (synchronisation) circuits (Fig. 2). This results in a four-level series-gated structure. Simulations indicate a 1.8: 1 improvement compared to a standard series-gated ECL AND-OR logic gate implementation with a cascaded latch stage. A similar four-level series-gated circuit (not shown) generates the sum out-

1156 ELECTRONICS LETTERS 13th September2001 Vol. 37 No. I9