1. CMOS Inverter: Dynamic V DD RnRn V out = 0 V in = V DD CLCL t pHL = f(R n, C L )  Transient, or dynamic, response determines the maximum speed at which a device can be operated.

2. Sources of Resistance  MOS structure resistance - R on  Source and drain resistance  Contact (via) resistance  Wiring resistance Top view Drain n+Source n+ W L Poly Gate

3. MOS Structure Resistance  The simplest model assumes the transistor is a switch with an infinite “off” resistance and a finite “on” resistance R on  However R on is nonlinear, so use instead the average value of the resistances, R eq, at the end-points of the transition (V DD and V DD /2) R eq = ½ (R on (t 1 ) + R on (t 2 )) R eq = ¾ V DD /I DSAT (1 – 5/6 V DD ) S D R on V GS  V T

4. Equivalent MOS Structure Resistance  For V DD >>V T +V DSAT /2, R eq is independent of V DD (see plot). Only a minor improvement in R eq occurs when V DD is increased (due to channel length modulation)  Once the supply voltage approaches V T, R eq increases dramatically V DD (V)11.522.5 NMOS(k  ) 35191513 PMOS (k  ) 115553831 R eq (for W/L = 1), for larger devices divide R eq by W/L V DD (V) R eq (Ohm) x10 5 (for V GS = V DD, V DS = V DD  V DD /2)  The on resistance is inversely proportional to W/L. Doubling W halves R eq

5. Source and Drain Resistance  More pronounced with scaling since junctions are shallower  With silicidation R  is reduced to the range 1 to 4  /  RSRS RDRD S G D R S,D = (L S,D /W)R  where L S,D is the length of the source or drain diffusion R  is the sheet resistance of the source or drain diffusion (20 to 100  /  )

6. Contact Resistance  Transitions between routing layers (contacts through via’s) add extra resistance to a wire l keep signals wires on a single layer whenever possible l avoid excess contacts l reduce contact resistance by making vias larger (beware of current crowding that puts a practical limit on the size of vias) or by using multiple minimum-size vias to make the contact  Typical contact resistances, R C, (minimum-size) l 5 to 20  for metal or poly to n+, p+ diffusion and metal to poly l 1 to 5  for metal to metal contacts  More pronounced with scaling since contact openings are smaller

7. Inverter Propagation Delay  Propagation delay is proportional to the time-constant of the network formed by the pull-down resistor and the load capacitance t pHL = ln(2) R eqn C L = 0.69 R eqn C L t pLH = ln(2) R eqp C L = 0.69 R eqp C L t p = (t pHL + t pLH )/2 = 0.69 C L (R eqn + R eqp )/2  To equalize rise and fall times make the on-resistance of the NMOS and PMOS approximately equal. V DD RnRn V out = 0 V in = V DD CLCL t pHL = f(R n, C L )

8. Inverter Transient Response V in V out (V) t (sec) x 10 -10 V DD =2.5V 0.25  m W/L n = 1.5 W/L p = 4.5 R eqn = 13 k  (  1.5) R eqp = 31 k  (  4.5)

9. Inverter Transient Response V in V out (V) t (sec) x 10 -10 V DD =2.5V 0.25  m W/L n = 1.5 W/L p = 4.5 R eqn = 13 k  (  1.5) R eqp = 31 k  (  4.5) t pHL = 36 psec t pLH = 29 psec so t p = 32.5 psec tftf trtr t pHL t pLH From simulation: t pHL = 39.9 psec and t pLH = 31.7 psec

10. Inverter Propagation Delay, Revisited  To see how a designer can optimize the delay of a gate have to expand the R eq in the delay equation V DD (V) t p(normalized) t pHL = 0.69 R eqn C L = 0.69 (3/4 (C L V DD )/I DSATn )  0.52 C L / (W/L n k’ n V DSATn )

11. Design for Performance  Reduce C L l internal diffusion capacitance of the gate itself -keep the drain diffusion as small as possible l interconnect capacitance l fanout  Increase W/L ratio of the transistor l the most powerful and effective performance optimization tool in the hands of the designer l watch out for self-loading! – when the intrinsic capacitance dominates the extrinsic load  Increase V DD l can trade-off energy for performance l increasing V DD above a certain level yields only very minimal improvements l reliability concerns enforce a firm upper bound on V DD

12. NMOS/PMOS Ratio  If speed is the only concern, reduce the width of the PMOS device! l widening the PMOS degrades the t pHL due to larger parasitic capacitance  = (W/L p )/(W/L n ) r = R eqp /R eqn (resistance ratio of identically-sized PMOS and NMOS)  opt =  r when wiring capacitance is negligible  So far have sized the PMOS and NMOS so that the R eq ’s match (ratio of 3 to 3.5) l symmetrical VTC l equal high-to-low and low-to-high propagation delays

13. PMOS/NMOS Ratio Effects  = (W/L p )/(W/L n ) t p (sec) x 10 -11  of 2.4 (= 31 k  /13 k  ) gives symmetrical response  of 1.6 to 1.9 gives optimal performance t pLH tptp t pHL

14. Device Sizing for Performance  Divide capacitive load, C L, into l C int : intrinsic - diffusion and Miller effect l C ext : extrinsic - wiring and fanout t p = 0.69 R eq C int (1 + C ext /C int ) = t p0 (1 + C ext /C int ) l where t p0 = 0.69 R eq C int is the intrinsic (unloaded) delay of the gate  Widening both PMOS and NMOS by a factor S reduces R eq by an identical factor (R eq = R ref /S), but raises the intrinsic capacitance by the same factor (C int = SC iref ) t p = 0.69 R ref C iref (1 + C ext /(SC iref )) = t p0 (1 + C ext /(SC iref )) l t p0 is independent of the sizing of the gate; with no load the drive of the gate is totally offset by the increased capacitance l any S sufficiently larger than (C ext /C int ) yields the best performance gains with least area impact

15. Sizing Impacts on Delay S t p (sec) x 10 -11 The majority of the improvement is already obtained for S = 5. Sizing factors larger than 10 barely yield any extra gain (and cost significantly more area). for a fixed load self-loading effect (intrinsic capacitance dominates)

16. Impact of Fanout on Delay  Extrinsic capacitance, C ext, is a function of the fanout of the gate - the larger the fanout, the larger the external load.  First determine the input loading effect of the inverter. Both C g and C int are proportional to the gate sizing, so C int =  C g is independent of gate sizing and t p = t p0 (1 + C ext /  C g ) = t p0 (1 + f/  ) i.e., the delay of an inverter is a function of the ratio between its external load capacitance and its input gate capacitance: the effective fan-out f f = C ext /C g

17. Inverter Chain  If C L is given l How should the inverters be sized? l How many stages are needed to minimize the delay? InOut CLCL  Real goal is to minimize the delay through an inverter chain the delay of the j-th inverter stage is t p,j = t p0 (1 + C g,j+1 /(  C g,j )) = t p0 (1 + f j /  ) and t p = t p1 + t p2 +... + t pN so t p =  t p,j = t p0  (1 + C g,j+1 /(  C g,j )) C g,1 12N

18. Sizing the Inverters in the Chain  The optimum size of each inverter is the geometric mean of its neighbors – meaning that if each inverter is sized up by the same factor f wrt the preceding gate, it will have the same effective fan-out and the same delay f =  C L /C g,1 =  F where F represents the overall effective fan-out of the circuit (F = C L /C g,1 ) and the minimum delay through the inverter chain is t p = N t p0 (1 + (  F ) /  )  The relationship between t p and F is linear for one inverter, square root for two, etc. NN N

19. Example of Inverter Chain Sizing  C L /C g,1 has to be evenly distributed over N = 3 inverters C L /C g,1 = 8/1 f = InOut C L = 8 C g,1 C g,1 1

20. Example of Inverter Chain Sizing  C L /C g,1 has to be evenly distributed over N = 3 inverters C L /C g,1 = 8/1 f = InOut C L = 8 C g,1 C g,1 1 f = 2f 2 = 4 3  8 = 2

21. Determining N: Optimal Number of Inverters  What is the optimal value for N given F (=f N ) ? l if the number of stages is too large, the intrinsic delay of the stages becomes dominate l if the number of stages is too small, the effective fan-out of each stage becomes dominate NN  The optimum N is found by differentiating the minimum delay expression divided by the number of stages and setting the result to 0, giving  +  F - (  F lnF)/N = 0  For  = 0 (ignoring self-loading) N = ln (F) and the effective-fan out becomes f = e = 2.71828  For  = 1 (the typical case) the optimum effective fan-out (tapering factor) turns out to be close to 3.6

22. Optimum Effective Fan-Out  Choosing f larger than optimum has little effect on delay and reduces the number of stages (and area). l Common practice to use f = 4 (for  = 1) l But too many stages has a substantial negative impact on delay  F opt f normalized delay

23. Example of Inverter (Buffer) Staging C L = 64 C g,1 C g,1 = 1 1 C L = 64 C g,1 C g,1 = 1 1 8 C L = 64 C g,1 C g,1 = 1 1 416 C L = 64 C g,1 C g,1 = 1 1 2.8822.6 N f t p 1 64 65 2 8 18 34 15 4 2.8 15.3

24. Impact of Buffer Staging for Large C L  Impressive speed-ups with optimized cascaded inverter chain for very large capacitive loads. F (  = 1) UnbufferedTwo Stage Chain Opt. Inverter Chain 10118.3 1001012216.5 1,00010016524.8 10,00010,00120233.1

25. Input Signal Rise/Fall Time  In reality, the input signal changes gradually (and both PMOS and NMOS conduct for a brief time). This affects the current available for charging/discharging C L and impacts propagation delay. t s (sec) t p (sec) x 10 -11 for a minimum-size inverter with a fan-out of a single gate  t p increases linearly with increasing input slope, t s, once t s > t p  t s is due to the limited driving capability of the preceding gate

26. Design Challenge  A gate is never designed in isolation: its performance is affected by both the fan-out and the driving strength of the gate(s) feeding its inputs. t i p = t i step +  t i-1 step (   0.25)  Keep signal rise times smaller than or equal to the gate propagation delays. l good for performance l good for power consumption  Keeping rise and fall times of the signals small and of approximately equal values is one of the major challenges in high-performance designs - slope engineering.

27. Delay with Long Interconnects  When gates are farther apart, wire capacitance and resistance can no longer be ignored. t p = 0.69R dr C int + (0.69R dr +0.38R w )C w + 0.69(R dr +R w )C fan where R dr = (R eqn + R eqp )/2 = 0.69R dr (C int +C fan ) + 0.69(R dr c w +r w C fan )L + 0.38r w c w L 2 c int V in c fan (r w, c w, L) V out  Wire delay rapidly becomes the dominate factor (due to the quadratic term) in the delay budget for longer wires.