Professional Documents
Culture Documents
5.2 CMOS Logic Gate Design: Serial or Parallel in Rs Transisto of No
5.2 CMOS Logic Gate Design: Serial or Parallel in Rs Transisto of No
(a).
(2). RTL/logic gates(carry look-ahead) (3). Ckt level (4). Layout level =>Most leveraged way is achieved by completing a (1).good architecture. (2). RTL/logic level (check pipelining, fan-in, fan-out etc).
(a).fan-in: number of inputs e.g.:4-input nand has a fan-in of 4 ,2-input nand gate has a fan-in of 2.(known) (b).fan-out: total number of gate inputs that are driven by a gate output. Default gate size=minimum sized inverter as unity. =>Fan-in & Fan-out will be affected by stage ratio and transistors in parallel or serial tr(rise time)for an m-input nand gate:
tr = Rp (m n Cd + C r + K C g ) n
. }
a b c
(1).Rp: effective resistance of a p-device in a unit inverter. (2).n: width multiplier of PMOS (3).k:fan-out (4).m:fan-in (5).Cg:gate capacitance of an unit . inverter. (6).Cd:drain capacitance (7).Cr:Routing capacitance.
Tr =
Rp n
( m n C r Cg + q(k) Cg + k Cg )
{ { {
Drain cap
Routing cap
fan-out
where
R: Cd/Cg: ratio of the intrinsic drain capacitance of an inverter to the gate Capacitance: q(k):Represent routing cap in terms of Cg
5.2.2
example: (spice simulation)
Wn = 6 , Ln = 1 W p = 12.3, L p = 1
Example of an 8-input NAND gate construction Approach1:An 8-input NAND + an Inverter. Approach2:Two 4-input NAND +2-input NOR Approach3 : see figure
Transistor sizing: stage ratio to drive large Cl (such as clock & global reset) Guideline: start with minimum sized devices then optimize
paths from a critical-path-timing analysis. Optimizing paths can be done at different levels - use nand structures where possible - place (big)inverters at high fan-out nodes if possible - Avoid the use of NOR structures in high-speed circuits(fan-in>4) fan-out is large) - use a fan-out below 5-10 - use minimum-sized gate on high fan-out nodes to minimize Cl presented to the driving gate. - Keep rising and falling edges sharp - When designing with power or area as a constraint, remember that large fan-in Complementary gates will always work given enough time.
( ) Z = A ( B + C ) + (D E )
a
d e b c
(+1)
clk=1 evlauate
{ {
0,n-block short 1,n-block open clk=0,z=1(Cl is charged to vdd) clk=1,Z is conditionally evaluated clk is a single phase clock pull-up time is improved. pull-down time is increased due to the ground switch.
problems: (a). inputs can only change during the precharge phase and must be stable during the precharge phase -> charge sharing may corrupt the o/p mode voltage. (b).simple single-phase dynamic CMOS gates cannot be cascaded(some delay between N1&N2)
Evaluate precharge
Vi Vj
F = P1 (V1 ) + P2 (V 2 ) + + Pn (V n )
Pi = control _ signals V i = pass _ signals V i {0 ,1 , X i , X i , Z }
F = i PiVi
(c).Logic function
F = A ( B) + A ( B) (d) implementation
(a)Complementary (b)NMOS (b) Cross-coupled
(b)Implementation:
the apparent advantages of pass-transistor networks in cmos should be studied carefully.(e.g: how to achieve good logic levels) -5.4.7 CMOS domino logic (1).during preharge(clk=0),PZ=1 ,and inverter o/p=0 (2)transistors in subsequent logic blocks will be turned off during he precharge phase
(3).each gate in sequence can make at most one transition(1 to 0)=>can be used in cascaded logic gates
limitations: (1).each gate must be buffered (an advantage ,too) (2)only non-inverting structures are possible (3)common in dynamic CMOS-charge sharing charging sharing in dynamic CMOS
Vn 1 =
C1 V DD ( C i ) + C1
i= 2 7
sol: (a) place clocked NMOS at the bottom (b) provide immediate nodes with (c) their own precharging transistors
during pre-charging .Advantage of D. CMOS (1). Smaller area (2). C , speed (3)glitch free if design carefully
5.5.2
.setup time: the time before the clock edge that the D input has to be stable .hold time: the time after the clock edge that the Dinput has to main stable .clock-to-Q delay(Tq):the delay from the positive clock input to the new value of the Q output. latches (a).negative levelsensitive latch (b)positive level-sensitive latch (c)positive edge-triggered register(master-slave) (d).operation of the master-slave register (e)CMOS circuit implementation.
RS latch R(reset)=active Q->0 S(set)=active Q->1 Two implementations: Nand gate Nor gate
Introduction to VLSI
Analog Ic design (Prof Wang) Vlsi system Design() 1.During summer Have VLSI cad tools 2.Implement a Small IC Computer organization
Testing(Prof Su) Digital signal Processing (DSP) (Prof Chang) 1.filter 2.FFT 3.Finite-word length effect
Communication ic design(by ql )
Latches
T-Register
D Q
clk
( )
system timing(in pipelined system) (a) Tc Tg + Td + Ts in a edgetriggered system (b) Level-sensitive system (1)Tda < Tc1 Tga Tsb ( 2)Tdb < Tco Tqb Tsc Tc1 = clk = 1 Tc 0 = clk = 0
5.5.8 Single-phase logic structures (1) Improve speed (2) Reduce area (3) Reduce dynamic power consumption (Example : N-P Dynamic CMOS Logic) Pipelined Structure: #CLK sections are precharged. #(-CLK) sections are evaluated when CLK=0 CLK=1
Basic rules : (1)During precharge : logic blocks must be switched off (Avoid internal (2)During evaluation : the internal inputs can make only one transition races)
5.510 Two-phase clocking 1. Two forms of clock skew : (1) clock overlap (2) slow rise/fall time 2. Two-phase dynamic register