An FBAR Circulator

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. IEEE MICR O W A VE AND WIRELESS COMP ONENTS LETTERS 1 An FB AR Circulator Mustaf a Mert T orunbalci , T re vor J. Odelber g, Student Member , IEEE , Suresh Sridaran, Member , IEEE , Richard C. Ruby , F ellow , IEEE , and Sunil A. Bhav e, Senior Member , IEEE Abstract — This letter p r esents the experim ental demonstrat ion of a film bulk acoustic resonator (FBAR) circulator at 2.5 GHz. The circulator is based on spatiotemp oral modulation of the series res onant frequenc y o f FB ARs using v arac tors and e xhibits a lar ge isolation of 76 dB at 2 .5 GHz. The FBAR chip (0.2 5 mm 2 ) consists of three identical FB ARs connec ted in wye config uration. FB AR’ s quality fac tor ( Q ) of 1250 and t he piezoelectric coupling coef - ficient k t 2 of 3% rela x the modulation requir ements, ac hieving nonr ecipro city with sm all modulat ion to RF ratio bette r than 1:800 (3 MHz: 2.5 GHz). Index T erms — Circulators, film bulk acoustic resonators (FB ARs), nonr ecipr ocity . I. I NTR ODUCTION C IRCULA T ORS are nonreciprocal three-terminal devices that transmit electr omagnetic signals entering any port to the next port, only in one direction. These de vices play a crucial role in co mmunication system s : 1) protecting the laser from reflected signals as an isolator in photonic systems and 2) enabling th e transmitter and the r ecei ver to share the same frequency band in full-duple x wireless systems . The most common approach and commercially a v ailable technology for breaking reciprocit y to realize a ci rculator is to use an e xternal magnetic field in a ferromagnetic medium. Howe ver , this approach requires b ulky magnetics that are expensi v e and not CMOS compatible. Recently , alte rnativ e approaches hav e been demonstrated in [1] and [2], where the three identical coupled L C res onators are modulated via spatiotemporal modulation using ring or wye topology . These approaches provide strong nonreciprocity of 50 dB with reasonable insertion l oss and bandwidth. In th is letter , we dem onstrate an alternative solution to the L C circu lator using three identical film bulk acoustic resonators (FB ARs) fabricated in wye configuration in a single packaged die. The FB ARs are superior to the L C lumped element resonators with a smal ler footprint and higher mechanical quality factor ( Q FBAR = 1250 at 2.5 GHz versus Q LC < 10 at 130 MHz) [3]. Th e high mechanical quality factor enable s the use of relati vely low modulation frequency , which will directly reduce the power consumpt ion of the Manuscript recei ved November 23, 2017; revised J anuary 25, 2018; accepted M arch 1, 2018. (Corresponding author: Mustafa Mert T orunbalci.) M. M. T orunbalci, T . J. Odelberg, and S. A. Bha ve are with the OxideMEMS Laboratory , School of Electrical and Computer Engineering, Purdue Uni- versity , W est Lafayette, IN 47906 USA (e-mail : mtorunba@purdue.edu; todelber@purdue.edu; bhav e@purdue.edu). S. Sridaran and R. C. Ruby are with Broadcom Ltd. , San Francisco, CA 95131 USA. Color versions of one or more of the figures in this paper are av ailable online at http://ieee xplore. ieee.org. Digital Object Identifier 10. 1109/LMWC.2018.2815271 Fig. 1. Schematic view (a) FBAR chip a t the center of wye topology connects to three v aractors to form the circulator . (b) The v aractor biasing network identical to [2]. The bpf ensu res that modulation does not leak into the FBAR chip and the high-pass filter ( hpf) allows RF signals to and from the circulator without the modulation leaking into the network analyzer . TA B L E I C OMPONENT L IS T AND V ALUES F OR FB AR C IRCULA T OR circulator . The FB ARs are modulated by varactors with a phase dif ference of 120°, providing an electrical rotation of the RF si gnal with strong isolation of 76 dB. Furthermore, the FB AR circulato r demon strates the appr oach in [2] at a higher frequency than pre viously achieved us ing lumped element inductors and capacitors. II. D ESIGN The operation principle and archi tecture of the FB AR circu- lator are id entical to the L C wye circulator [2]. F ig. 1 presents the schem atic view of the FB AR circulator , co ntrasting th e simplicity o f the FB AR chip with printed circuit b oard (PCB) complexi ty of providing t he modulation signal to the v aractor . T able I provides a list of components used in each branch of th e cir culator . The L C bandpass filter (bpf) has a linear wideband characterist ic which allo ws us to explore dif ferent modulation frequencies and am plitudes. The modulation fre- quency was chosen to be 3 MHz as this relates to the 3-dB bandwidth of the FB AR. The modulation signal is a 7 V pp square wav e with 0-V dc bias that dri ves the varactor into forward b ias with low ON impedance during the positi ve half cycle, and provides 0.2-pF capacitance during the ne gati ve half cycle. The modulation is not allo wed t o leak into the wye network by a second bpf and bleed resistor . An inline high- pass filter i s placed at each RF port to pre vent the high voltage modulation s ignal from damaging recei vers in the network 1531-1309 © 2018 I EEE. Personal u se i s perm itted, but republication/redistri bution requires IEEE permission. See http://www .ieee.org/publications_sta ndards/publicati ons/rights/i ndex.html for more information. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 2 IEEE MICR O W A VE AND WIRELESS COMPONENTS LETTERS Fig. 2. Simulation of the varactor’ s im pact on admittance of an FBAR. FBAR’ s BV D param eters are e xtracted from a fabricated resonator . T uning the capacitance v alue changes the seri es resonance frequency at the cost of decreased adm ittance. Fig. 3. Photos of the PCB where all the circulator components occupy an area of 12 × 8m m 2 . The FBAR die at the center is the smallest component of the circulator: 98% of the area is occupied by the components surrounding the v aractors. analyzer , at the cost of 2-dB R F ins ertion loss. Fig. 2 shows the effect of v aractor modulation on the admittance of an FB AR. The MA COM v aractor was chosen based on its tuning ratio and RF performance characteris tics at gigahertz range in the datasheet. III. E XPERIM ENT AL R ESUL TS Three FB ARs with series resonance frequency of 2.5 GHz were fabricated in a centroid c onfigur ation to achieve identical frequency and impedance characteristics. Fig. 3 shows the photos of the PCB where all the ci rculator c omponents occupy an area of 96 mm 2 . The F B AR die at the center is the smallest component, 98% of the area is occupied by the modulation circuits surrounding the varactors . Fig. 4 show s a schematic view of the experimental set up used to characterize the FB AR circulators. Square wa ve Fig. 4. Experimental setup for the characterization of FB AR circulators. Fig. 5. Comparison of m easured versus simulated S parameters at port 1 when the modulation is OF F . The m easured S21 has 3 dB higher loss than simulation because of PCB parasitics and from the inline high-pass filter. Measured S21 and S31 are identical c onfirming reciprocal behavior with no modulation. modulation signals are generated with a phase difference of 120°, and S-parameters are measured using an Agilent four-port network analyzer . The FBAR die plus varactors have expected port to port impedance of 60  based on an extracted Butterworth-V an Dy k e (BVD) mod el. As the PCB was n ot optimized for high frequenc y using 3-D planar electromagnet ic simulators, as well as the ad dition of the inline high- pass filter to p rotect the n etw ork an alyzer, the measured S21 has 3 dB higher insertion loss for the FB AR chip (Fig. 5). When the modulati on is OF F , S21 and S31 demonstrate identical reciprocal response. Note that the resonance frequency is shifted from FB AR ’ s resonance frequency (Fig. 2) due to the series capacitance of the varactor . As the MA COM varactor has 13-V b reakdo wn v oltage, an i nitial sinusoidal modulation of 6 V pp and 6-V dc bias was applied to achie ve the largest capacitance swing. Less than 10-dB isolation was measured with this scheme. It w as disc overed, howe ver , against our ori g- inal intuition that increasing t he modulation v oltage amplitude such that the v aractor went into forw ard bias, and clipped and improved nonreciprocity to 12 dB. This result revealed that swi tchlike modulation of the v aractor impro ved response, and the best modulation scheme was determined to be a 50% duty cycle square wa ve, 0-V dc bias, and 7 V pp . As soon as the modulation signals are applied, a nonreciprocal response This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. TOR UNBALCI et al. : FB AR CIRCULA TOR 3 Fig. 6. When the modulation is turned ON , the measured S parameters at all ports exhibit forward insertion los s of 11 dB and re verse isolation of 28 dB. The measurem ent was performed with all three ports connected to 50  impedance of the network analyzer . Fig. 7. Measured circulator respons e at port 1, after matching network of L series = 5 . 9n Ha n d C shunt = 1 pF is introduced at each port of the network analyzer . The maximum meas ured isolation is 76 dB. The bandwidth at 40 dB isolation is m easured as 0.4 M Hz. TA B L E I I S UMMAR Y OF R ESUL TS AND C OMP ARISON TO P REVI OUS W ORKS is observed between S21 and S31. S31 achieves an isolation of 28 dB while S 21 is 12 dB with 50-  ter mination pro- vided by the netw ork analyzer to all three terminals (Fig. 6). The sidebands are due to th e v aractor nonlinearity and are spaced by 3 MHz on either side of the signal band; howe ver , the right sideband f alls in the parallel resonance of the FB AR and is greatl y suppresse d. The drop in S21 (modulati on ON ) versus S21 (modul ation OF F ) of 4 dB represents the RF energy that is los t to intermodulation products in this single-ended circulator topology . The measured isolat ion was identical at all three ports. The wye configuration’ s branch impedances necessitate the need for a matching network to optimize the circulator performance. Using a Smith char t, we calculate the matchin g network to be L series = 5 . 9n Ha n d C shunt = 1 pF . Fig. 7 sho ws the response of the circulator at port 1 after the mat ching network was programmed into the network analyzer . S21 has 11-dB inserti on loss, while S 31 achie ves notch depth of 76 dB, demonstrating a nonreci procity between ports 2 and 3 of 65 dB at port 1. The bandwidth at 40-dB isol ation is 0.4 MHz. IV . C ONCLUSION In this letter , we demon s trate an FBAR circulator at 2.5 GHz which achieves 76- dB iso lation with a 7-V pp modulation at 3 MHz. The insertion loss can be impro ved by a lo w loss PCB design using 3-D plana r electrom agnetic sim ulators such as Advanced Design System mo ment um and implementing a dif ferential circulator topology [4], [5]. The FB AR s in this circulator are zero-drift resonators with low Q and k t 2 due to an oxide compensation layer [6]. W e can improv e forward transmis sion and rev erse isolation bandwidths using uncom- pensated FB AR s ( Q FBAR = 3000 and k t 2 > 7%). T able II compares the FB AR circulator w ith first-of-type circulators in L C resonat or and CMOS technologies. Modulating the FB AR using a va ractor not only s hifts the resonator frequenc y , but also its series resistance. W e are currently analyzing if this effect can be harnessed to combine the advantages of both capaciti ve modulati on and resis tance modulation [7] circulator topologies. W e also believ e that nonreciprocity can be achieved using switches and cap acitors in parallel instead of a varactor , simpli fying the de vice further . A CKNO WLEDGMENT The authors would like to thank Prof. D. Perouli s, Dr . M. A. Khater , an d Dr . A. Fisher fo r their in v aluable discus sions on PCB design and experimental testing, and Prof. A. Alu and Prof. H. Krishnasw amy for sharing their insight and answeri ng all our queries. The y w ould al so like to thank sspecially M. Quinn of Maj elac Inc. for PCB assembly . R EFERENC ES [1] N. A. Estep, D. L . Sounas, J. So ric, and A. Alù, “Magnetic-free non- reciprocity and isolation bas ed on param etrically modulated coupled- resonator loops, ” Natur e Phys. , vol. 10, no. 12, p. 923, Nov . 2014. [2] N. A. Estep, D. L. Sounas , and A. Alù, “Magnetless microwa ve circulators based on spatiotemporally modulat ed rings of coupled resonators, ” IEEE T rans. Microw . Theory T echn. , vol. 64, no. 2, pp. 502–518, Feb . 2016. [3] R. Ruby et al. , “Positioning FBAR technology in the frequency and timing domain, ” IE EE T rans. Ultrason., F err oelect., F req. Contr ol , vol. 59, no. 3, pp. 334–345, Mar . 2012. [4] A. Kord, D. L. Sounas, and A . Alù. (2017). “Pseudo-linear- time-inv ariant m agnet-less circulators based on differential spatio- temporal modulation of res onant junctions.” [Online]. A vailable: https://arxiv .org/abs/1709. 08133 [5] Y . Y u et al. , “Magnetic free radio frequency circulator based on spatiotem - poral commutation of MEMS resonators, ” in Proc. IEE E 31st Int. Conf. Micr o E lectr o Mech. Sys t. (M EMS) , Belfast, N orthern Ireland, Jan. 2018, pp. 154–157. [6] R. Ruby , K . Sankaragomathi, S. Srida ran, and R. Parker , “Oven controlled fbar oscillator, ” in Proc. IEEE Int. Ultrason. Symp. (IUS) , T aipei, T aiwan, Oct. 2015, pp. 1–4. [7] N. Reiskarimian and H. Krishnasw amy , “Magnetic-free non-reciprocity based on staggered commutation, ” Nature Commun. , vol. 7, Apr . 2016, Art. 11217.