ICD nertdCir Integrated  Content W PLAs Storage T Pass Standar using Design System Paths Euler and Design Cell Abstraction and Rules Design Layout T of Overview Intr drV ider oduction asso Cir ransistor iew utDesign cuit echnologies nertdCrutDesign Circuit Integrated cuits osi Maharatna Koushik anMcNally Iain Cells d lectur lectur es es anMcNally Iain nertdCir Integrated History    Gor ipelgcfntosfr functions logic Simple akKlyT Kilby Jack oetNyeFair Noyce Robert onBar John Geof oe Resour Notes Books Assessment Moor is Commer First is nertdCir Integrated First is lnrItgae Cir Integrated Planar First is T First nertdCir Integrated httpusersecssotonacukbimnotesicd ACir Design VLSI CMOS of Principles Circuits Integrated Digital o Moor don fr yDme RylRdrEtbihet Establishment Radar Royal Dummer ey ut n ytm Perspective Systems and cuits Law es en W deen ransistor Examination Coursework Informal xsInstr exas Fair e AddisonW W Neil Pr Rabaey Jan utDesign cuit le rtan n W and Brattain alter enticeHall ilICs cial ut Pr cuits hl child hlosre h tr the observes child mns uments seDvdHarris David este ces mT n Fair and TI om cuit Coinventor se esley oposed Coinventor cuit LEi aeLayout Gate LEdit limSoke Bl Labs Bell Shockley illiam child nsi integration in ends pr ttp failed ototype History History Pr 1,000,000,000 100,000,000 dcsepnnilgr exponential edicts 10,000,000 1,000,000 ne a one yGor by founded was Intel obe vr ersne n pr and since year every doubled Gor In te er befor years other pr he In ef Gr Moor In Moor thr 100,000 10,000 cetpcigo components of packing cient 1,000 whwudnwdpn nyo pr on only depend now would owth uht to ough 048080 4004 9018 902000 1990 1980 1970 ecie i nta gr initial his describes e obigEeyY Every Doubling obigEeyT Every Doubling 8008 o Moor don er dce httegr the that edicted vsdgr evised 8086 whi h ubro opnnsprchip per components of number the in owth o Moor don Moor bevdta h ubro opnnsprci had chip per components of number the that observed e lwn u opyia limits physical to due slowing e 286 whpr owth Moor whpr owth n oetNyefr Noyce Robert and e a tIntel at Law es whwudcniu ttesm aefran for rate same the at continue would owth dcin oduln vr w years two every doubling to edictions 386 oY wo ear Law es dcin siiuosypr ridiculously as edictions cs impr ocess dce httetr the that edicted 486 DX ears Pentium mFair om vmnsrte hno mor on than rather ovements child Pentium II Itantium n ol continue would end Itantium 2 Pentium III Pentium 4 Dual Core Itantium 2 ecise 2010 e History Faster Slower Moor the uses industry whole The  rteIntels the or oto aia qimn obidIsdulsappr years doubles ICs build ery to equipment capital of Cost Joneses the with up keep Must o costly too atdinvestment wasted Moor Law es a uv opa e arcto facilities fabrication new plan to curve Law es efflligPr Selffullling a ophesy xmtl ev oximately 1947 Point Contact 1961 Fairchild Bipolar RTL RS Flip-Flop (4 )

1965 Moore’s Law (Mk I) Number of transistor has doubled every year and will continue to do so until 1975

1975 Moore’s Law (Mk II) Number of transistors will double every two years Dual Core Itantium 2 1,000,000,000 Itantium 2 Itantium 100,000,000 Pentium 4 Pentium III 10,000,000 Pentium II Self-fulfilling Pentium 1,000,000 Prophecy 486 DX 286 386 100,000 8086

10,000 4004 8080 8008 1,000 1970 1980 1990 2000 2010

 R T MOS T Bipolar T of Overview Diode vriwo T of Overview A LIvre n O gate NOR and Inverter TL N-Channel P-Channel ransistors ransistors A Enhancement Z echnologies echnologies V opnnsfrLogic for Components CC Depletion N N P Z S B A SUB G AB PN D S Z N N P P SUB G P Field induced N-channel V Silicon Oxide Insulator CC N D Z vriwo T of Overview l ucin a er be can functions All vriwo T of Overview  otTLfmle ar families TTL Most instead used be could gates NAND ECL hrceitcmlieitrtasso r transistor multiemitter characteristic TTL ie atrsicigta R than switching faster gives savr ihsed ihpower high speed high very a is B A TTL NAND Gate B A echnologies echnologies mor e aie ihasnl O aegate base NOR single a with ealized te ioa T Bipolar Other Z ope hntebscvrinsonher shown version basic the than complex e V CC Z La h xes fgr of expense the at TL oauaigtechnology nonsaturating dcsteoealcmoetcount component overall the educes echnologies Z ECL OR/NOR Gate B A AB Z Z ae complexity eater e V V CC EE Z  The MSaVS technology VLSI a NMOS T of Overview vriwo T of Overview MS CMOS natv MSdvc opeet h MSdvc giving device NMOS the complements device PMOS active An devices the of ratio size the by determined is voltage output low The consumption incr r Resistance load nonlinear as acts transistor Depletion Cir utfnto eemndb eisprle obnto fdevices of combination seriesparallel by determined function cuit elgbesai oe consumption power static negligible swing output rail to rail tt fteart the of state B A AZ ae steehneetdvc un n hsr thus on turns device enhancement the as eases V echnologies echnologies DD Z VLSI B A MSlogic CMOS V V Z DD DD AB A esistor V DD Z B A Z dcn power educing Z iia MSCir CMOS Digital iia MSCir CMOS Digital ttcCO opeetr gates complementary CMOS Static V rossotad ar shorthands arious T ar devices CMOS All NMOS Gnd and PMOS ngnrlsbtaecnetosar connections substrate general In o n e fipt ther inputs of set any For assosata ipedgtlycontr digitally simple as act ransistors lentv r Alternative B A Z V sdfrsmlfigCO cir CMOS simplifying for used e DD epr nacmn mode enhancement e cuits cuits snain o MStransistors CMOS for esentations ileitete aht d rapt oGnd to path a or Vdd to path a either exist will e o rw wher drawn not e C B A le switches olled Z utdiagrams cuit etheyconnecttoVdd N-Channel P-Channel C B A Z iia MSCir CMOS Digital iia MSCir CMOS Digital l opudgtsar gates compound All elsbefntosar functions Realisable Symbol D C B A C B A opudGt Example Gate Compound Z cuits cuits rirr NRexpr ANDOR arbitrary e inverting e opudGates Compound Z Pull Down Network Pull Up Network , , , , , , sin ihivre output inverted with essions D C E B A F Z GND V DD RTL NOR Gate TTL NAND Gate ECL OR/NOR Gate A Z Z A A B Z B B Z

VCC VCC VCC

Z Z

AB A Z Z B AB

VEE

NMOS Compund Gate CMOS Compund Gate VDD VDD A A B B

Z Z C C

ohalwcntuto fNR ADCmon aelasinverting always gate Compound NAND NOR of construction allow Both

MSN ttcpower static No CMOS NMOS

ssos VLSI esistors r no ransistors T MOS

C ORNOR ECS NAND TTL NOR TL R

astr ihRssosMSILSI with ransitors T Bipolar opnnsfrDgtlI Design IC Digital for Components opnnsfrDgtlI Design IC Digital for Components P T NPN Diode T str Real str Ideal et contr Depth ontp implants ntype wo ransistor uctur uctur le implants olled eD eD idsadBplrT Bipolar and Diodes T Bipolar and Diodes N PN P N ransistors ransistors EBC EBC NPN N P P P N N N S opnnsfrDgtlI Design IC Digital for Components ipeNO T NMOS Simple opnnsfrDgtlI Design IC Digital for Components einRule Design ar transistor above not connection Gate l usrtst gr to substrates All wafer of bottom to is connection Substrate is polysilicon implant by The and oxide thick by blocked is It mask extra no has implant Ntype contr also mask Polysilicon Ar Active SUB G D ams ensetn of extent denes mask ea ransistor efAligned Self SD ipeNO T NMOS Simple ound N O T MOS SUB G l xetof extent ols PN ransistors hc Oxide Thick Active Area Mask ea hnOxide Thin Thick Oxide ransistor Polysilicon Mask alias P-substrate Thin Oxide aeOxide Gate S SD N SUB G G Polysilicon D N opnnsfrDgtlI Design IC Digital for Components pr Wher MST NMOS opnnsfrDgtlI Design IC Digital for Components cs sused is ocess iemssms eue odetetransistor the dene to used be must masks Five N mlnspr implants PN T PW op ti o utbefrsbtaecnetost eshar be to connections substrate for suitable not is it e implant P implant N Polysilicon Ar Active PW lfrisolation for ell substrate ransistor ell N+ implant mask P+ implant mask ea connection P well mask dc good oduce MST NMOS O T MOS ohmic ransistors P well SUB SUB ransistor contacts P ++ ++ S SD N G G d mor a ed D N complex e MSPr CMOS MSPr CMOS mln ensPO sour PMOS denes implant P sour NMOS denes implant N r transistors the Thus biased verse PW oeta h ucinbtenci usrt n W N and substrate chip between junction PN the that Note h pr The l n T and ell cs ecie her described ocess ocess ocess i T win Metal Contact Window Polysilicon P implant N implant Active Area N-well bpr ub mi isolated emain cse loexist also ocesses eisan MSInverter CMOS Inverter CMOS edanadNO usrt contact substrate NMOS and cedrain edanadPO usrt contact substrate PMOS and cedrain NW l pr ell ocess ic thsol nNW N an only has it since l ilr will ell N N N P P P N P mi r emain ell e N-well P N Active Area defines Thick Oxide P N Polysilicon defines Thin Oxide P N

N implant NN N aligned to AA and Poly PN

P implant PN N P N P aligned to AA and Poly PN Contact Window PN NP P PN N

Metal P NN PPN PN

Diode CMOS Inverter Twin Tub CMOS Process

P N

P N

PN N P PN NPN Transistor PNI CMOS Inverter N-Well CMOS Processs (with explicit N+ implant mask)

EBC

EBC

N P N

PN N P PN NMOS Enhancement transistor P N NMOS Process CMOS Inverter N-Well CMOS Processs (without explicit N+ implant mask) G

SD

SD

NN P PN N P PN SUB P N Features may be determined by a number of masks e.g. NMOS source drain: ActiveArea AND NOT(NWell OR Poly OR PImplant)

 electron beam Reticle written by scanning Oxide Growth Photoresist Deposition Silicon Wafer Contact/Proximity Mask Photolithography akMaking Mask pia r Optical dcinalw narr allows eduction by step and repeat with optical reduction Pattern reproduced on wafer (or contact/proximity mask) Reticle Mask UV Light wrln widths line ower Lens (positive) Photoresist Mask Pattern Glass Mask SiO (chrome) 2 Photoresist Strip Oxide Etch Photoresist Development Photoresist Exposure UV Light Isotropic Etching Lateral Diffusion N einRules Design T eiaino einRules Design of Derivation opr igelyrr layer Single utae r Multilayer vn hpfailur chip event NN ules ules dsgsms ofr odsg r design to conform must designs e Misalignment can be Cumulative Mask Misalignment is aligned to is aligned to Optical Focus over 3D terrain ules Reticle Mask NN einRules Design Abstraction Pr T r Design T Laborious Design Level Mask Pr Design Level Gate asso ee Design Level ransistor cs neednT independent ocess ocessT lsmycag uigadesign a during change may ules cnlg independent echnology echnologyPr Metal Contact Window Polysilicon P implant Active Area N-well eeso Abstraction of Levels = NOT{ N implant } cnlg dependent echnology m cs dependent ocess MSinverter CMOS btato tc Diagrams Stick Abstraction hl viigsm ftepr the of some avoiding while abstraction of benets the of many us give diagrams Stick iia MSDesign CMOS Digital oeta l Cdsgsms n ttems level mask the at end must designs IC all that note ayt change to Easy Pr specication mask full than easierfaster Much piie aotmyb eeae uhmor much fr generated than be may layout Optimized cs needn vldfrayCO pr CMOS any for valid independent ocess mtasso rgt ee designs level gate or transistor om oblems tc Diagrams Stick ocess aiyfr easily e masikdiagram stick a om Combined contact Tap Contact P+ N+ Polysilicon Metal & tap iia MSDesign CMOS Digital iia MSDesign CMOS Digital Explor ayt xadt ullyu o atclrpr particular a for layout full a to expand to Easy Pr cs needn layout independent ocess ragmn fdvcsadconnections and devices of Arrangement contacts of Number cr of Implications orDsg Space Design your e ossovers tc Diagrams Stick Diagrams Stick ocess tcsadCDSmoi Captur Symbolic CAD and Sticks tcsadCDMagic CAD and Sticks eiatmtccmato pr compaction semiautomatic A ar contacts yblccaptur corr symbolic of generation rapid to leads DRC Online ar err contacts DRC again width with sticks design style Log ar components T assosar ransistors uoaial eetda r as selected automatically e lcdadepiil sized explicitly and placed e ondwt zer with joined e tl opcini vial fdesir if available is compaction style e uoaial eetda r as selected automatically e it wir width o cs ilcr will ocess equir es aeDCcorr DRC eate ed c designs ect r ar ors e equir W=1.1 L=0.5 W=2.2 L=0.5 ed agdimmediately agged e ed c layout ect Processing Optimised Mask Layout Equivalent Stick Diagram

N

Design Rules - width, separation, overlap

Metal Polysilicon N+ P+ Contact Tap Combined contact & tap

 iia MSDesign CMOS Digital iia MSDesign CMOS Digital eo iger single a below l opeetr ae a edsge sn iger single a using designed be may gates complementary All o h aoiyo hs ae ecnda ragmn ftassossuch transistors of arrangement transistors an adjoining nd butt can can we we gates that these of majority the For Referr Car Car dt as to ed floinaino rnitrsour transistor of orientation eful or transistor of selection eful oia appr logical A wo rnitrainda omngt connections gate common at aligned ptransistors of ow ieo diffusion of line ue Path Euler aht aelayout gate to oach dering eaddrain and ce wo rnitr bv or above ntransistors of ow iia MSDesign CMOS Digital Her n ua Algorithm Human One Algorithms Computer iia MSDesign CMOS Digital ether si osbet olwasmlrylble ahthr path labelled similarly a follow to possible it Is Expr idapt hc assthr passes which path a Find designer human the for easy so not is This sear in transistors r is It Y otyaano a iet r ahs hstime this rst path p a try to like may you again try No syuv succeeded youve es s h ahi em ftegt connections gate the of terms in path the ess C B A ear ltvl ayfracmue ocnie l osbearneet of arrangements possible all consider to computer a for easy elatively orpsil ue paths Euler possible four e ho utbeElrpath Euler suitable a of ch Z idn nElrPath Euler an Finding Path Euler an Finding uhalntassoseatyonce exactly ntransistors all ough uhteptransistors the ough Lblpl columns poly Label path Euler Find Design CMOS Digital obndcnatadtap and contact combined A faln fdfuin Wher diffusion of line a of iia MSDesign CMOS Digital a sgo o bu rnitr insuf transistors about for good is tap hsi o h aeasml tap simple a case the not is this e Rueotu node output Route Ruepwrnodes power Route idn nElrPath Euler an Finding ue ahExample Path Euler a eue nywher only used be may cettp a ev hpvleal olatchup to vulnerable chip a leave may taps cient Adtaps Add Ruer Route oe otc xssa h end the at exists contact power a e 1 1 2 2 hudb used be should miignodes emaining 3 3 o MSadNMOS and PMOS for 4 4 5 5 iia MSDesign CMOS Digital iia MSDesign CMOS Digital D C E B A F opsil ahthr path possible No idn nElrPath Euler an Finding Path Euler an Finding Z uhntransistors ough iia MSDesign CMOS Digital G H C D B E A F I or No opsil ahthr path possible No ragmn ilcr will earrangement idn nElrPath Euler an Finding Z uhptransistors ough aeasolution a eate Investigation of Euler paths leads to more efficient layout*

ODD 1 5 EVEN 2 EVEN 6 EVEN 3 7 ODD 4 EVEN ODD 6

ODD ODD 7 5 EVEN 4 2 ODD

1 ODD ODD 3

1 2 374 5 6 1 2 374 5 6 1 2 374 5 6

1 2 374 5 6 1 2 374 5 6 1 2 374 5 6

*not all gates will support a common Euler path for both PMOS and NMOS

 iia MSDesign CMOS Digital iia MSDesign CMOS Digital B A S Inter ar outputs and inputs gate All height same of be all should Gates nti aePlslcnvrial n ea horizontally Metal and vertically Polysilicon case this In Pr r All gr and Power sre h eet fhierar of benets the eserves onc svia is connect uigi xenlt cells to external is outing udriswl hnln pwe butted when up line then will rails ound w odco r conductor two Z utpegates Multiple gates Multiple SAB SABZ chy vial ttpadbottom and top at available e outing Z iia MSDesign CMOS Digital ayIsar ICs Many Standar outputs and inputs cell for and vertically Metal and inter rails for metal only use pr VLSI modern Most LAYER 1 METAL plcto pc nertdCir Integrated Specic Application aotwr efre yteAI einri iie notostages two into divided is designer ASIC the by performed work Layout h SCdsge sal a oacs otefl aoto h standar the of layout full the cr to doesnt access and no has usually designer ASIC The cir a library design must designer ASIC The ASIC hsi fe are u yo nbhl ftefoundry the of behalf on or by out carried often is This ellibrary cell A Cr Library Cell Routing Placement elDesign Cell d Design einduigtestandar the using designed e otiigcmol sdlgcgts scr is gates logic used commonly containing aeaynwclsfrtelibrary the for cells new any eate eation cse upr w rmor or two support ocesses cl r cell T outing olyrMetal wolayer cuit sal eahrznal adfrpower for and horizontally Metal Usually utuigtelgcgtsaalbei the in available gates logic the using cuit elmethod cell d LAYERS 2 METAL ea layers metal e ae o pr a for eated h omi to is norm The dcells ocess lcmn Routing Placement gr vertically r the In ar Cells Routing Placement udrails ound uigcanl ewe h el er we cells the between channels outing lcdi n rsvrleullnt ie ihinter with lines length equal several or one in placed e Placement Routing uemtl oiotlyadmetal and horizontally metal oute dgttdpwrand power digitated lcmn Routing Placement Standar themselves h ipeteapeue thr uses example simplest The W t thr ith  In optragrtm a eue oensur to used be can algorithms Computer channel vertical com to human W a even or computer r a the for plete easier considerably life makes This cr appr logical This ute optragrtm a eue ootmz h r the optimize to used be can algorithms computer Further ar ssmtl n metal and metal Uses metal in is T which IO for except metal only Use Standar oCnutrRouting Conductor wo ossing short e utol ensur only must e VLSI A eo mor or ee AOTASRC ROUTING ABSTRACT LAYOUT elDesign Cell d cir Cells d BZ cuits outing ea aesi spsil otk dif a take to possible is it layers metal e we ahmasta esol ee aet or bu signals about worry to have never should we that means oach often httosgaswl o eti h aehrzna or horizontal same the in meet not will signals two that e T ocnutrr conductor wo Mor nd emtllayers metal ee GND! Vdd! that BZ AB ea Layers Metal e inter cell wiring lcmn fclssc htwir that such cells of placement e outing occupies fer GND! Vdd! BZ AB n appr ent mor uigitself outing e ar ea oach than the cells es r appr This ar Nwells r second every ipping By Standar r cell between gaps smaller much to W Standar uigalgorithms outing t hsappr this ith Vdd! GND! Vdd! Vdd! GND! GND! Vdd! GND! ahi omlyascae ihsas r sparse with associated normally is oach emer elDesign Cell d Design Cell d ahw a r can we oach e n oe rgr or power and ged lentv lcmn Style Placement Alternative N!GD GND! GND! GND! d!Vd Vdd! GND! Vdd! GND! Vdd! Vdd! GND! Vdd! d!Vdd! GND! Vdd! wi a epsil oeiiaegp ewe r between gaps eliminate to possible be may it ow Mor d!Vd Vdd! Vdd! GND! Vdd! uesfl vrtecl oteseie isleading pins specied the to cell the over safely oute ea Layers Metal e ows udrisar rails ound N!GND! GND! N!GND! GND! shar e d!Vdd! Vdd! w n o hne based channel non and ows ed d!Vdd! Vdd! GND! GND! ows LAYOUT ABSTRACT ROUTING

A BZ

Vdd! Vdd! Vdd! Vdd! A B Z A B Z NAND2 NAND2 NAND2 1 METAL LAYER

GND! A B Z GND! A B Z GND! GND!

ABZ

A B Z

Vdd! Vdd! Vdd! Vdd! A B Z AB Z NAND2 NAND2 NAND2 2 METAL LAYERS

A B Z AB Z GND! GND! GND! GND!

AB Z Vdd! Vdd! Vdd! Vdd! NAND2 NAND2 NAND2 3 METAL LAYERS A B Z A BZ A B Z GND! GND! GND! GND!

 ttcCO opeetr Gates Complementary CMOS Static asT Pass h astasso sue nmn yai MScir CMOS dynamic many in used is transistor pass The  wher cir uses which logic Dynamic cf asT Pass possibilities these of use make gates str the which in conict Wher ther inputs of set any For Complementary valid appr the After Static ortasiso flgc logic of transmission Poor logic of transmission Good Pr erdto flgcvalue logic of degradation time rise slow ulu spromdb natraiemethod alternative an by performed is pullup e asso Cir ransistor hscniini o e ehv ihrahg meec upto a or output impedence high a either have we met not is condition this e vdsvr opc cir compact very ovides ransistor pit pr opriate cuits netpt uces ttcCMOS Static succeeds path ongest NOUT IN pgto ea h uu eoe ai n r and valid becomes ouput the delay opagation ileitete aht d rapt oGND to path a or Vdd to path a either exist will e GND utcpctnet stor to capicitance cuit Vdd cuits ENABLE PDN PUN OUT tt o hr time short a for state e cuits  Noncomplementary emains asT Pass asT Pass  ic ther since T T asiso aeLayout Gate ransmission Gate ransmission o ttccir static For eitl edtesle oa to themselves lend mediately cir these that note lwrpulldown slower pullup faster balanced asso Cir ransistor asso Cir ransistor IN ear eso nusfrwihteotu snihrple o o high nor low pulled neither is output the which for inputs of sets e EN EN NOUT IN ut ewudnral s MStasiso gates transmission CMOS a use normally would we cuits n OUT and EN EN ut ar cuits p cuits cuits astransistors pass IN o ul complementary fully not e EN EN ieo diffusion of line OUT NOUT IN implementation NEN EN  ec hyd o im not do they hence asT Pass asT Pass  nenlci u hudnvrb loe oa ihimpedance high oat to allowed be never should bus chip internal u W Bus T asiso aeMultiplexor Gate ransmission gr r inverter one only multiplexing distributed inverter dif for transistors few very rm addt o ee metal level for candidate prime IN0 IN1 asso Cir ransistor asso Cir ransistor al ipie lblwiring global simplies eatly cl aotmyof may layout cult iring A3 A2 A1 A0 S SAS AS OUT IN1 IN0 equir cuits cuits stti advantage this fset OUT S S S dprbn ftasiso gates transmission of bank per ed  B3 B2 B1 B0 SCS BS SCS BS IN0 IN1 C3 C2 C1 C0 S S OUT u itiue Multiplexing Distributed Bus mlmnaino isieALU bitslice of Implementation Multiplexing Distributed Bus fr drivers many with signals for Ideal sur db pr by ed  oeta rnmsingtshv odiecpblt ntesleHer themselves in capability drive no have gates transmission that Note A[n] once i itiue multiplexor distributed via Connected cir Separate A M[n] vdn buf oviding M Sequencer Decoder & utfrec function each for cuit 0 0 Operation Code IR Subtract C[n-1] fers Operand A+M A-M FA  C[n] OE ME WR A MDR MAR & A<<1 A&M mdif om Address CACC PC MUX fer A|M 0 1 n modules ent RAM A^M Data fn A[n+1] A>>1 ALU oddiei en is drive good a e A[n-1] A[n]’ u itiue Multiplexing Distributed Bus asT Pass T gates transmission use may implementation Multiplexor distributed not is Multiplexing module ALU optimized Single itt Inverter ristate n aemyhv rtt uptb obnn twt transmission a with it combining by output tristate gate a have may gate Any asso Cir ransistor F0 F1 F2 F3 F4 F5 C[n-1] 3 A[n] FA 0 cuits 01 C[n] M[n] 1 A[n]’ 2 Functions Supported: 0 M A IN 3 A-M A+M A[n+1] 4 A<<1 A>>1 EN EN A|~M A&~M ~(A^M) A^M A|M A&M ~M ~A OUT asT Pass asT Pass T T itt netrLayout Inverter ristate Inverter ristate lentvl h rnmsingt a eicroae notegate the into incorporated be may gate transmission the Alternatively loese osimulate to easier also r is connection one asso Cir ransistor asso Cir ransistor NOUT IN NEN EN cuits cuits mvdese olayout to easier emoved NOUT IN EN EN EN IN OUT asT Pass T itt netrBsDriver Bus Inverter ristate rnitr a eszda r as sized be may transistors buf inverting tristate a asso Cir ransistor A3 A2 A1 A0 SAS AS cuits e sotnue odiehg aaiac u signals bus capacitance high drive to used often is fer B3 B2 B1 B0 equir SCS BS SCS BS ed C3 C2 C1 C0 Pass Transistor CMOS Transmission Gate EN EN IN OUT IN OUT IN OUT IN OUT EN EN EN

Transmission Gate Multiplexor EN EN S OUT Tri-state gates are used for Multiplexing IN1 IN1 Distributed Multiplexing OUT S OUT using transmission gates IN0 IN0 IN1 IN0

S S S S *note distictive polysilicon crossover using tri-state inverters

Tri-state Inverter

using tri-state buffers

IN OUT IN OUT EN EN

Bi-directional I/O EN IN OUT Tri-state Buffer

IN OUT IN OUT

EN EN *this is another form of multiplexing

 ace n FlipFlops and Latches ace n FlipFlops and Latches T latch gate transmission CMOS ipetranspar simple A asiso aelthlayout latch gate ransmission D w netr ar inverters two low is load when latched transpar opc aoti osbeuiglyrmetal layer using possible is layout compact a LOAD D LD n hnla shigh is load when ent LD n ac a ebidar build be can latch ent er equir dsnetetasiso aecno rv itself drive cannot gate transmission the since ed Q udatasiso aemultiplexor gate transmission a ound Q DQ LD LD D LOAD 0 1 Q ace n FlipFlops and Latches ace n FlipFlops and Latches conguration o s nsml synchr simple in use For inverters tristate using achieved be may layout simpler A pact h cir the transpar a which in transpar condition race the avoids this r design this D utbhvsa iigeg trigger edge rising a as behaves cuit DLD LD n ac prtn ntesm lc phase clock same the on operating latch ent equir D stoadtoa rnitr u a elb mor be well may but transistors additional two es nu cir onous CLOCK 0 1 ut eueapi flthsi atrslave master a in latches of pair a use we cuits Q CLOCK 0 1 dDtp ipop type D ed D LD n ac rvsasecond a drives latch ent Q LD Q ecom ace n FlipFlops and Latches ace n FlipFlops and Latches lentv conguration Alternative T T itt netrimplementation inverter ristate implementation gate ransmission Implementation CLK CLK CLK CLK D D CLK CLK D D CLOCK 0 1 CLOCK 0 1 20 Transistors Q 16 Transistors 16 Transistors Q Q Q ace n FlipFlops and Latches ace n FlipFlops and Latches CLK D CLK CLK o h aefntoaiyw ol s neg trigger edge an use could we functionality same the For type D slave master of Layout D mor mor few a conguration alternative using compact very ipe lc distribution clock simpler ope wiring complex e transistors e 26 Transistors Q 16 Transistors CLK Q D D L L L CLK CLK CLK CLK dDtype D ed Q Q pc saving space Wher File Register AZ odsgasms egic fr glitch be must signals Load deT Edge B ehv lar have we e Reg3 Reg2 Reg1 rigger dDtp pr Dtype ed eaonso trg h s fidvda ace a edto lead can latches individual of use the storage of amounts ge vnsarc odto condition race a events A ewt ihl contr tightly with ee B Clock DQ Clock le timing olled Reg Z LDr3 LDr2 LDr1  Load D Load Load D D Reg Reg1 Reg1 Reg2  Q Q Q Reg ENBr3 ENBr2 ENBr1 ENAr3 ENAr2 ENAr1  Latch Master Slave D-Type Flip-Flop

D D 1 1 Q Q 0 0 D Q D Q

CLOCK CLOCK LD LD

LD LD CLK CLK CLK CLK

D D 1 1 Q Q 0 0 D Q D Q

CLOCK CLOCK

LD LD

LD LD CLK CLK CLK CLK *muliplexor based latches and flip- include distictive polysilicon crossover

Edge Triggered D-Type Flip-Flop

D

Q

CLK 26 Transistors

CLK D Q Euler path analysis is applied creatively to these multi-gate cells - gates are often linked via the common gnd/pwr node Final layouts will be more complex where clock buffers, reset circuitry and metal 2 i/o are included

 Ls OsadRAMs and ROMs PLAs load depletion NMOS the of place in str PLA Most RAMs and ROMs PLAs mlmnigmlil O Smo Pr of Sum SOP multiple implementing Pr C B A gambeLgcArystr Array Logic ogrammable h n hne eie utb aie nor in ratioed be must devices channel N and P on The always is device P the since conditions static cir CMOS complementary Unlike pe n ttcpwrdissipation power static and speed r ratioing This voltage output low SOP = AND-OR uctur slsi lwrgtatog ther although gate slower a in esults 1I2 I1 sepo suoNO O ae sn hne device Pchannel a using gates NOR pseudoNMOS employ es P4 P3 P2 P1 suoNO O gate NOR PseudoNMOS L str PLA 3I 5I I7 I6 I5 I4 I3 uctur Y X Z spr es utteegtswl ispt oe under power dissipate will gates these cuits dcs rPSexpr POS or oducts uctur vd oia n opc ehdof method compact and logical a ovide C B A es AND plane satradeof a is e e ocr to der essions P4 P3 P2 P1 aeter the eate OUT ewe gate between f OR plane equir ed Y X Z Ls OsadRAMs and ROMs PLAs Ls OsadRAMs and ROMs PLAs C B A aoti ipyamte fslcigadpaigr placing and set selecting ited of matter a simply is Layout expr intermediate for Ar glrlyu sepoewt oun o nusadotusadr and outputs and inputs for columns with employed is layout egular B XYZ ABC P4 P3 P2 P1 essions P1 P2 P3 P4 L str PLA str PLA Y X Z uctur uctur B XYZ ABC e e N ln el OR plane cells AND plane cells caglrclsfr cells ectangular Peripheral cells P1 P2 P3 P4 malim a om ows Ls OsadRAMs and ROMs PLAs Ls OsadRAMs and ROMs PLAs A structur RAM mization to Conversion O a ipyb L ihe eoe plane decoder xed with PLA a plane be data simply may ROM A A2 scnmk s ftesm eoeplane decode same the of use make can es A2 eoePaeData Plane Decode Plane drs ie Data Lines Address Lines sticks VDD A1 GND A1 ssrih forwar straight is A0 AB A0 L str PLA A ROMs Only AND plane cells are shown here D3 BC uctur D2 ihopruiisfrfrhropti further for opportunities with d D1 C P1 P3 P4 P2 e D0 VDD GND W0 W1 W2 W3 W4 W5 W6 W7 A n pr and Word Lines A B ogrammable B C C P1 P2 P3 P4 Ls OsadRAMs and ROMs PLAs RMStr SRAM hr ie oc uigwieNO rnitr of transistors NMOS write during conict lived Short standar a on storage density high for Used Dif Din[2] Din[1] Din[0] nWE nOE A[1] A[0] fer I BIT BIT nilapiesar ampliers ential uctur e Standar sdfrsed r speedy for used e rnitrsai A cell RAM static transistor d ttcRAM Static WORD I BIT BIT MSpr CMOS d ead ocess e str fer ne path onger GND WORD Vdd Dout[2] Dout[1] Dout[0] PLA and ROM SRAM

BIT BIT

WORD select

AND Plane OR Plane

ROM is PLA with fixed AND (decoder) plane programmable OR (data) plane Cells are designed to butt together in two dimensions leading to efficient layout PLA layout efficiency will depend on the actual function implemented (e.g. number of common product terms)

 ytmDsg Choices Design System ytmDsg Choices Design System piiainlmtdb speedpowerar by limited optimization Pr Pr eiCso Design SemiCustom ulCso Design Custom Full Design SemiCustom ulCso Design Custom Full gambeLogic ogrammable Logic ogrammable eg PLD iiu kl r skill Minimum market to time Best pr for Cheapest turnar design possible Best eg Pr Mask eg Pr Field otsilr skill Most Density Highest Power Lowest Fastest pr mass for Cheapest eg Standar leaHar Altera Array Gate CMOS ECS EPC Cyclone Altera XC P lae itcMC MTC Mietec Alcatel LV C ELV atc ispGALV Lattice PEELCV ICT V AL elDesign Cell d gambeGt ra FPGA Array Gate ogrammable gambeGt Array Gate ogrammable equir ototyping equir ed Cp Istr II dCopy oduction ed udtime ound uctur ataeof trade ea dASICs ed m f ellibrary cell ST AR HERE T Pr il Pr Field 

I/O Buffers Logic ogrammable iixX a has XC Xilinx Pr Blocks IO Blocks Logic Congurable Repr pr Fuse use time One gambeInter ogrammable CLB CLB Horizontal Routing Channel gambeUlcrclyErasable UVElectrically ogrammable gambeGt Array Gate ogrammable CLB CLB I/O Buffers I/O Buffers connect ogrammable Lsadu o to up and CLBs Vertical Routing Channel

I/O Buffers  iixXC Xilinx srIOpins IO user Point Interconnection Programmable CLB Switching Matrix il Pr Field il Pr Field pins IO user to up and  leaCcoeE a oi lmnsragda LgcAryBokand Blocks Array Logic as arranged Elements Logic has EPC Cyclone Altera oi ra lcs a lcsIOElements IO Blocks Ram MK Blocks Array Logic Pr gambeInter ogrammable gambeGt Array Gate ogrammable gambeGt Array Gate ogrammable connect leaCyclone Altera iixX CLB XC Xilinx  il Pr Field akPr Mask gambeGt Array Gate ogrammable gambeGt Array Gate ogrammable arranged as 4 columns 9 Output Pads of 17 sites each. 68 Gate Sites 8 Input Pads GND Pad leaCcoeLE Cyclone Altera 8 Input Pads V Pad 9 Output Pads DD akPr Mask Standar O C A B X  W uoGnrtdMacr Generated Auto Functions Logic W Contact and Metal Customize ytmLvlBlocks Level System l upr ytmO hpapplications Chip On System support ill RAM ROM PLA Micr GND elDesign Cell d gambeGt Array Gate ogrammable opr cso cor ocessor Blocks o e  Vdd no ak only masks indow B C X A O O C A B X GND Vdd GND AC B C A B Vdd O A B C O X ulCustom Full Wher designers custom full need styles design All Hnedcmue aechip game computer Handheld eg techniques semicustom alongside used odsg uligbok o semicustom for blocks building design to pr base the design to Standar ar optimum for crafted hand bitslice custom Full Macr elar lc AROM RAM block o elcontr cell d eAIsuefl utmtcnqe hyar they techniques custom full use ASICs ge oller gambelgcchips logic ogrammable aef ea cec n o oe consumption power low and ciency ieyt be to likely e Programmable Logic Semi-Custom Full Custom

Skill Required Little Lots

Time to Market Quick Slow One-off Cheap Manufacturing Cost Expensive Unit Cost Cheap In Production Expensive

Speed Fast Slow

Area Small Big

Power Low High

All design styles need full custom designers A large ASIC (especially SoC) may mix Semi-Custom and Full Custom