IASIT Inenaional Jounal of Enee and Technoloy, Vol., No.5, Ocobe 010 ISSN: 1793-836 Expeimenal Analysis of Paamee Limiaions Hih-Fequency Resonan Gae Dive N. Z. Yahaya, K. M. Beam and M. Awan Absac In meahez swich fequency, he effec of loss is sifican. In diode-clamped esonan ae dive cicui, he esonan duco cuen, duy aio and dead ime ae he limi paamees which b implicaions o he swich loss and hence oal ae dive loss. The expeimenal analysis has been caied ou o validae he simulaion esuls. Fom he pedeemed duco cuen of 9 nh, duy aio of 0 % and dead ime of 15 ns, emakably, he expeimenal esuls show less han 10 % diffeence value compaed o he simulaion. Theefoe, his new fd validaes ha by us coec choice of hese values, he diode-clamped esonan ae dive can opeae bee hihe swich fequency. Index Tems Hih Fequency, Limi Paamees, Resonan Gae Dive, Swich Loss. I. INTRODUTION Thee have been many esonan ae dive (RGD) oduced ecen yeas and mos of hem looked fo soluion swich loss educion and impoved efficiency wih hihe powe oupu densiy. A hih fequency opeaion, specifically above 1 MHz, he effec of ae dive cicui on oveall pefomance becomes ciical. Even houh many cicui opoloies employ L esonan confiuaion echniques ae applied [1-13], deailed analyses on swich loss and eney savs ae ye o be fuhe exploed and veified. A hih fequency self-poweed esonan ae dive has been oduced o impove powe savs, educe ae div loss [1] and ohe sudies concenae on ecove he cicula eney he cicui [4, 6, 9, 1]. Moe issues elaed o solv shoo-houh cuen he applicaion of synchonous buck convee cicui ae well descibed [7]. In addiion, o elae he impoance of RGD, some eviews on hih fequency convee des ae pesened [14]. In he pevious wok, seveal ae dive opoloies have been sudied. Iniially, he ducive coupled ae dive has been poposed whee he powe losses ae vesiaed based on he un-on cycle of ae volae and duco cuen [11]. Then a deailed sudy on he diode-clamped esonan ae dive is caied ou ode o analyze he swich loss fo specific limi paamee values [13]. Subsequenly, he exension of he wok is pesened his pape o veify he simulaion analyses of diode-clamped RGD cicui, Fi. 1 as discussed he pevious wok [15] whee eney ecovey and powe loss savs ae obseved. The simulaion wok has shown ha hee ae paamees which affec he des oucome of diode-clamped RGD cicui. Hee, he des used mus be based on opimized duy aio, D, dead ime, D and duco value, L. These limi paamees ae impoan achiev hih fequency ae dive opeaion. Fom ieaive numeical mehod, i is found ha he D, D and L ae 0 %, 15 ns and 9 nh especively. This evenually ives esuls lowe oal swich loss and yields bee pefomance hih fequency envionmen. i L V s,s1 Fi.1 Diode-lamped RGD icui [15] The simulaion was caied ou accod o cicui opoloy Fi. 1 wih hih powe MOSFETs ae used. V P1 and V P ae he wo sepaae pulse widh modulaos (PWM) which povide complemenay pulses o swich Q 1 and Q especively. The 5-V PWM pu pulses ae used o dive he swiches wih a D eval of 15 ns. The opea wavefoms of he swiches, he duco cuen, i L and he ae volae of S 1, V s,s1 ae shown Fi.. Manuscip eceived July 10, 010. N. Z. Yahaya is wih Univesii Teknoloi PETRONAS, Malaysia. uenly he is expec o si fo he PhD viva. His ma PhD wok concenaes on he developmen of powe eleconics module hih fequency sysem (phone: 605-368-783; fax: 605-365-7443; email: nozaiha_yahaya@peonas.com.my). K. M. Beam is a lecue specializ Physics and baey des. uenly she is aached wih Univesii Teknoloi PETRONAS, Malaysia (e-mail: mumajbeam@peonas.com.my). M. Awan is wih he Elecical Enee Depamen, Univesii Teknoloi PETRONAS, Malaysia. His eseach ees is he aea of Analo I icui Des (e-mail: mohdawan@peonas.com.my). 418
IASIT Inenaional Jounal of Enee and Technoloy, Vol., No.5, Ocobe 010 ISSN: 1793-836 5.0V.5V Swich Q1 - ON Swich Q - OFF Swich Q - ON Swich Q1 - OFF 5.16 4.00 Q1 Swich Q Swich 0V Dead ime.00 13.9 V(Q:,Q:s) V(Q1:,Q1:s) 10.0 Vs of S1 powe MOSFET 0 5.0 0 SEL>> Resonan duco cuen, il Duy Raio -.00 L=1nH L=5nH L=10nH L=0nH L=30nH L=50nH (Hih R Effec) 19.95us 193.00us 193.05us 193.10us 193.15us 193.0us 193.5us 193.30us 193.35us 193.40us 193.45us V(S1:,S1:s) I(L) Fi. Opea Wavefoms of Diode-lamped RGD icui [15] The cicui opeaion explas ha as soon as Q 1 conducs, i L sas o develop. Sce Q is OFF a his ime, boh i L and V s,s1 ae chaed o maximum values. The duaion of cha ime depends on L and he esonan impedance of he newok. Once hey ae fully chaed, i L will hen dischae o zeo houh body diode of Q, L, D 1 and back o V s. This cuen pah dicaes eney ecovey sav mode whee pu cuen oes back o souce a he end of he swich cycle. Then afe D, Q akes un o conduc wih Q 1 is now uned off. Hee, i L is once aa chaed bu he opposie diecion. The V s,s1 appoaches zeo o a po whee i L is maximum. Similaly, he i L is also dischaed o zeo fom neaive peak value houh D, L, body diode of Q 1 and back o V s. This cicui s symmeical opeaion conues he nex subsequen swich cycles. Howeve, he effeciveness of he RGD cicui depends on he values of he limi paamees. II. LIMITING PARAMETERS OF DIODE-LAMPED RGD IRUIT The discussion on he limiaions of he diode-clamped RGD cicui is vey limied. The choice and values of componens used, especially L is impoan. The objecive is o poduce a small scale boad wih smalle size componens. The summaion of cha and discha ime of i L is used as a benchmak deem he mimum ane of duy aio, D. Fom Fi. 3, fo fixed pulse widh V P1 and V P, an cease of L leads o hihe oscillaion couns a he end of is un-off. onsequenly, hihe sess and dissipaion will be expeienced by he diode-clamped RGD cicui. On he ohe hand, educ L will esul slowe elaive swich speed. 193.910us 194.000us 194.00us 194.400us 194.600us 194.800us 194.906us V(Q1:,Q1:s) V(Q:,Q:s) I(L6) I(L5) I(L4) I(L3) I(L) I(L1) Fi. 3 Vay i L fo 0 % pulse widh a D of 15 ns [15] In numeical analysis, he i L equaion is iven (1) whee R is he oal ae esisance of aound 1.5 Ω. The ime aken fo i L o each peak value is iven () which measue he maximum cuen value. The ise and ecovey imes of i L ae iven (3) and (4) especively. These equaions dicae he ansiion speed of he swich cycle. The fase il akes o each peak value, he bee he ansiion speed will be. This also applies o he ecovey ime. i L 00mA 0A ( ) = peak V = s an 1 ise e L ( ec R L s( R L ) ) (1) () π = L (3) = π L (4) -00mA 991.4us 991.44us 991.46us 991.48us 991.50us 991.5us Fi. 4 R du un-off of i L a D = 15 ns fo L = 9 nh Sal oscillaion canno be avoided especially hih fequency cicui. Fi. 4 shows ha hee is a pesence of i L which causes sifican shif powe loss du V s,s1 un-off. This is due o hih paasiic ducance he ae dive. Howeve, he ma is sill low and can be acceped he des. 419
IASIT Inenaional Jounal of Enee and Technoloy, Vol., No.5, Ocobe 010 ISSN: 1793-836 In ohe aspec, D value can be obaed fom he adeoff beween speed and powe dissipaion a un-off. When D is se small, hee exiss moe couns. On he ohe hand, if D is applied oo lon, V s,s1 will appea floa lead o hih powe dissipaion. In addiion, he D value can be deemed fom he adeoff beween speed and hih powe dissipaion also a V s,s1 un-off. A lone D oduces hihe oveshoo volae. This paen comes fom he eney soed he paasiic ducance when i is eleased acoss paasiic capaciance of S 1. Theefoe based on he adeoffs, D is opimized a 15 nh, as shown Fi. 5 [15]. Peak cuen (A).36.35.34 peak cuen Fo Induco of 9 nh is used fo ae dive cicui. The oupu swich, S 1 has successfully uned on and he expeimenal seups have shown coec opea wavefoms simila o simulaed wavefoms as illusaed Fi.. In addiion, he PWM oupus fom he funcion eneaos ae fed o MOS dive (EL7104) befoe connec o he ae of MOSFET swiches, Q 1 and Q. IV. EXPERIMENTAL RESULTS Two complemenay PWM sals ae applied o he dives. The PI o DSP chips can be used o eneae hese sals. Howeve, fo simpliciy duy aio and dead ime vaiaion, he abiay funcion eneao is chosen. Fo 1 MHz swich fequency hee will be some noise appeaed he sals. Us MOS ae dives and file cicui, his noise can be educed sificanly. Wih 15 ns D applied beween Q 1 and Q, he oupu PWM sals of 0 % duy-aio esul fom he MOS dives ae shown Fi. 6..33 0 5 10 15 0 5 30 Dead (ns) 5 V Q 1 Q Fi. 5 Opimized D value a lowes peak cuen of i L III. METHODOLOGY The expeimenaion ook place powe eleconics eseach lab wih some of he componens chaned compaed o simulaion, due o he unavailabiliy local make. Sce his is he pelimay expeimenal wok, he diffeence I/V as of he componens used do no ale he esuls much compaed o he simulaion. The chanes ae shown Table I and Table II. TABLE I omponen Name PSMN130-00D/PLP 1N639 IRFP50 V D TABLE II OMPONENTS USED IN SIMULATION Simulaion Ras 00 V / 0 A, R DS(ON) =0.130 Ω 45 V / 60 A 00 V / 33 A, R DS(ON) =0.085 Ω 48 V OMPONENTS USED IN EXPERIMENT Expeimenaion omponen Name Ras STPNF03L 30 V / A, R DS(ON) =0.038 Ω SDP06S60 600 V / 6 A IRFI540NPBF 100 V / 0 A, R DS(ON) =0.05 Ω 5 V V D The PWM sals ae eneaed by dual-oupu abiay funcion eneao (Tekonik AFG 310). The ae dive is applied o he ducive choppe load. Hee, he pimay objecive of he discee expeimenal es-bed is o exame he effeciveness of ae dive cicui meahez swich fequency ove a fixed load condiion. A 1-V pu volae Fi. 6 PWM sals (y:v/div, x:00ns/div) These PWM sals ae fed o wo ae emals of n-channel MOSFETs via dive chips. The expeimenal behavio of cha and discha i L is shown Fi. 7. The posiive peak cuen is abou 3. A and he ise and ecovey imes aken by his cuen ae abou 5 ns and 50 ns especively. This esuls show esemblance of i L behavio found he simulaion. i L(peak) = 3. A ise =5 ns ec =50 ns i L(peak) = -.8 A Fi. 7 Induco cuen, i L (y:a/div, x:100ns/div) The cha of i L yields he cha of V s,s1 o maximum peak value of 5 V as shown Fi. 8. This esul maches he simulaion esul shown Fi.. Howeve, due o say ducance expeienced by he swich, caused by he load duco, some noise is seen. 40
IASIT Inenaional Jounal of Enee and Technoloy, Vol., No.5, Ocobe 010 ISSN: 1793-836 TABLE III OMPARISON OF DATA Peak of 5 V V s,s1 Expeimen Simulaion Mahad i L(peak) A 3..4 3.6 ise (ns) 5 30 5.81 ec (ns) 50 58 51.6 Fi. 8 V s,s1 oupu volae (y:5v/div, x:500ns/div) The noise also leads o he oscillaion of i L du un-off. Fom simulaion Fi. 4, he ampliude of 00 ma is obseved. The expeimenal esul shown Fi. 9 poves his whee he i L is apped and measued beween -00 ma and 00 ma. 00 ma -00 ma Fi. 9 i L oscillaion du un-off (y:00ma/div, x:50ns/div) In ode o jusify ha swich S 1 conducs coecly, he da cuen and volae ae also measued. Boh of hem coespond o he ducive load cicui whee peak cuen and volae measue A and V especively which ae ue based on iven load paamees. i L V ds,s1 P SW,Q1 (mw) 45. 418.0 40.34 P SW,Q (mw) 438.5 453.47 447.80 The simulaion daa is aken fom pevious wok [15]. The numeical analysis us Mahad is caied ou o veify he esuls us he fomula iven Eqn. (1) o Eqn. (4). Fom Table III, he expeimenal daa validaes he simulaion wok. In addiion, Mahad calculaion has shown and poven ha he analysis is coec only wih he diffeence of less han 10 %. Howeve, he simulaed peak i L value dicaes a sifican diffeen esul compaed o ohes. This is caused by hihe enal ae esisance he dive used he simulaion seup. Ohe han ha, he esuls show pomis dicao fo he RGD cicui o opeae hih swich fequency wih lowe sess and bee pefomance. VI. ONLUSION This pape validaes he simulaion esuls obaed fom PSpice simulaion which was discussed deails [15]. In diode-clamped esonan ae dive cicui, he pedeemed paamee values such as duco cuen, duy aio and dead ime have been jusified via expeimenal analysis. When compa he simulaion esuls wih expeimen and Mahad calculaion, he esuls show ha he diffeence of less han 10 %. Theefoe, his ae dive can be used hihe swich fequency wih he specified paamee values. I ds,s1 AKNOWLEDGMENT The auhos wish o hank Univesii Teknoloi PETRONAS fo povid fancial suppo o publish his wok. Fi. 10 I ds (y:0.5a/div) and V ds (y:5v/div) of S 1 (x:100ns/div) Theefoe, all of he expeimenal esuls validae he simulaion wok excep some of he paamees and componens used ae no of he same ypes wih he simulaion. Neveheless, he oucomes sill show a defie poof of he wok. The analysis of daa is epesened he nex secion. V. RESULTS AND DISUSSIONS All hee ses of daa ae analyzed based on diffeen esuls aken fom expeimen, simulaion and Mahad. The peak i L, ise, ec, swich loss Q 1, P SW,Q1 and P SW,Q ae compaed o validae he simulaion esuls. The compaison is abulaed Table III. 41 REFERENES [1] H. Wan and F. Wan A Self-poweed esonan ae dive fo hih powe mosfe modules IEEE App. Powe Elecon. onf. and Expo., pp. 183-188, 006. [] T. Lopez, G. Saueleande, T. Duebaum and T. Tolle A deailed analysis of esonan ae dive fo PWM applicaions IEEE App. Powe Elecon. onf. and Expo., Vol., pp. 873-878, Feb. 003. [3] D. Sobczynski Acive ae dives 8 h In. Sp Sem., pp. 36-39, 005. [4] Y.H. hen, F.. Lee, L. Amooso and H. Wu A esonan MOSFET ae dive wih efficien eney ecovey IEEE Tans. Powe Elecon., Vol. 19, No., pp. 470-477, Ma. 004. [5] L. Baloh Des and applicaion uide fo hih speed MOSFET ae dive cicuis Texas Insumen App. Noes, pp. 1-37, Sep. 007. [6] Y.H. hen, F.. Lee, L. Amooso and H. Wu A esonan MOSFET ae dive wih complee eney ecovey IEEE Powe Elecon. Moion onol onf., Vol. 1, pp. 40-406, 000. [7] K. Yao and F.. Lee Novel esonan ae dive fo hih fequency synchonous buck convees IEEE Tans.Powe Elecon., Vol. 17, No., pp. 180-186, Ma. 00.
IASIT Inenaional Jounal of Enee and Technoloy, Vol., No.5, Ocobe 010 ISSN: 1793-836 [8] I. Omua and Y. Shi Gae div cicui and ae div mehod of powe MOSFET U.S. Paen App. Publicaion, pp. 1-9, Feb. 006. [9] B.S. Jacobson Hih fequency esonan ae dive wih paial eney ecovey Powe Elecon. Spec. onf., pp. 131-141, 001. [10] N.Z. Yahaya, K.M. Beam & M. Awan. Simulaion Analysis of an Effecive Gae Dive Scheme fo Sof-Swiched Synchonous Buck onvee In. Jounal of En. & Tech. IJENS, Vol. 9, No. 10, pp. 9-35, Dec 009. [11] N.Z. Yahaya, K.M. Beam and M. Awan Simulaion of an ducive coupled hih fequency esonan ae dive cicui IEEE Ind. Inf., pp. 153-157, Jul. 008. [1] W. Ebele, P.. Sen and L. Yan-Fei A new esonan ae dive cicui wih efficien eney ecovey and low conducion loss IEEE Ind. Elecon. Soc., pp. 650-655, 005. [13] N.Z. Yahaya, K.M. Beam and M. Awan Swich loss analysis of a diode-clamped esonan ae dive newok UMS Sc. Tech. Sem., pp. 471-475, Oc. 008. [14] N.Z. Yahaya, K.M. Beam and M. Awan A eview on VHF powe eleconics convee and des issues IAENG En. Le., Vol. 16, Iss. 3, pp. 88-93, Au. 008. [15] N.Z. Yahaya, K.M. Beam and M. Awan The analysis of paamee limiaion diode-clamped esonan ae dive cicui In. Jounal En. and Tech. IASIT, Vol., No. 1, pp. 17-, Feb. 010 N. Z. Yahaya was bon Lumu, Malaysia. He wen o he Univesiy of Missoui-Kansas iy, USA o sudy eleconics. He aduaed wih a BSc Elecical Enee 1996. Afe ha he seved 5 yeas he dusy Malaysia. In 00, he was awaded his MSc Micoeleconics fom he Univesiy of Newcasle Upon Tyne, UK. uenly he is wai fo his PhD Viva a he Univesii Teknoloi PETRONAS, Malaysia. His ma each/eseach aeas ae he sudy of Powe Eleconics Swich onvees and Analo Powe Devices. K. M. Beam aduaed Physics fom he Madas Univesiy, India 198. Afe aduaion, she joed he pos-aduae couse Physics wih Eleconics specializaion and eceived he Mases Deee fom he Bhaahidasan Univesiy, India 1984. Then she was wok vaious capaciies a he P.S.N.A. ollee of En. & Tech., affiliaed o Anna Univesiy fo 17 yeas. She came o Malaysia he yea 000 and obaed he Docoae fom he Mulimedia Univesiy fo he wok on Solid Sae Devices. uenly, she is aached wih he Univesiy Teknoloi PETRONAS as an Associae Pofesso he Depamen of Elecical and Eleconic En. He eseach ees is Lihium-ion baeies, hybid powe souces, Solid sae devices, and Semiconduco sensos. M. Awan eceived he B App Sc fom USM, Penan, Malaysia, 1980, he MSc (E) fom Univesiy of New Bunswick, Fedeicon. anada, 1984, and he Ph.D fom Univesiy of Souhampon, Enland, 1991. He had woked as es enee a Inel echnoloy, Penan, pio o he pos aduae sudy. He is an Associae Pofesso a he depamen of Elecical and Eleconic Enee, USM, unil 003. uenly, he is an Associae Pofesso a he Elecical and Eleconic Enee, Univesii Teknoloi PETRONAS, Malaysia. He eseach eess clude he des and implemenaion and veificaion of low powe analo RF cicuis and diial Is. 4