Combinational Logic Gates in CMOS
References:
Adapted from: Digital Integrated Circuits: A Design
Perspective, J. Rabaey, Prentice Hall © UCB
Principles of CMOS VLSI Design: A Systems Perspective,
N. H. E. Weste, K. Eshraghian, Addison Wesley
Adapted from: EE216A Lecture Notes by Prof. K. Bult ©
UCLA
Design Techniques for Large Fan-In
• Transistor Sizing
• Progressive Transistor Sizing
• Transistor Ordering
• Logic Design to reduce the gate fan-in
Progressive Sizing
• When parasitic capacitance
is significant (e.g., when fan-
in is large), needs to
consider distributed RC
effect
• Increasing the size of M1 has
the largest impact in terms of
delay reduction
• M1 > M2 > M3 > … > MN
ln3
lnN
Out
ln1
ln2
M1
M2
M3
MN
Delay Optimization by Transistor Ordering
ln3
lnN
Out
ln1
ln2
M1
M2
M3
MN
Critical signal next to supply
Critical path
ln3
ln1
Out
lnN
ln2
MN
M2
M3
M1
Critical signal next to output
Critical path
Improved Logic Design
Reduce the fan-in to each gate
Example: F = ((A + B + C).D)
A
B
C
D
A B C
D
output
which path is the slowest ?
Example: F = ((A+B +C).D )
A
B
C
D
A B C
D
Slowest path
Alternatives ?
If minimum-size: tdr 6 tdf,inv
tdf,inv = delay of a minimum inverter
Example: F = ((A+B+C).D)
A
B
C
D
If minimum-size: tdr 7 tdf.inv
this approach is advantageous if driving larger load
Design of an 8-Input AND
Approach 1
Approach 2
Approach 3
3-input NAND Gate with Parasitic Capacitors
inc
out
inb
ina
Cp+load
Ca
Cb
Cc
P1 P2 P3
N3
N2
N1
Macro Modeling for Worst Case Analysis
tdf = [RN1Ca] + [(RN1+RN2)Cb] + [(RN1 + RN2 + RN3)Cc] +
[(RN1 + RN2 + RN3)Cp + [(RN1 + RN2 + RN3)Cload]
Internal delay External load
td = Td, internal + x Cload
Macro Modeling
)(2
2
Lj
N
jNdfCmpC
n
mRCR
mmt
m: fan-in
n: sizing factor of NMOS transistors
p: sizing factor of PMOS transistors
Cinv,min = total gate capacitance of minimum size inverter
k = “fan-out” corresponding to CL
Cj = r Cinv,min
n
mkC
n
pCmCmmRt
invjj
Ndf
min,
22
2
)(
Keep m2 and m2p/n and mk/n reasonable
Macro Modeling for NOR Gate
p
mkC
p
nCmCmmRt
invjj
PNORdr
min,
22
,2
)(
Keep m2 and m2n/p and mk/p reasonable
n
kC
n
mpCmCRt
invj
jNNORdf
min,
,
Keep n as small as possible to minimize the impact on rise-delay
Design Strategy
• Use minimum sized transistors
• Analyze critical path (slowest, maybe more than 1)
• Look at alternative implementations (substitute
NOR’s ?)
• Compare and choose best
• Analyze critical path(s) and optimize transistor
sizing
Complementary Logic
• 2n transistors
• Complicated wiring
• No functional sizing required
Ratioed Logic
Ratioed Logic
Reduce the number of devices over complementary logic
Ratioed Logic
• Use PDN to implement the function (which is the
negation of the network)
• Total number of devices: n for the input, 1 for the
static load
• Minimum load is 1 unit-gate load
• Functional sizing is required to optimize noise margin
Functional Sizing in Ratioed Logic
• N transistors
• VOH = VDD
•
• Asymmetrical response
• Static power consumption
• tpLH = 0.69 RLCL
• tpHL = 0.69 (RL || RPDN) CL
DD
LPDN
PDN
OLV
RR
RV
Current Source as the Static Load
• If current source equals VDD/RL
= initial charging current from
the resistive load
• More than 25% reduction
compared to resistive load
ave
swingL
pLHI
VCt
2/
PDN
F
In1
In2
In3 2
LL
pLH
RCt
Load Lines of Ratioed Gates
0.0 1.0 2.0 3.0 4.0 5.0
V ou t (V )
0
0.25
0.5
0.75
1
I L(N
orm
aliz
ed
)
R esistive load
Pseudo-N M O S
D epletion load
C urrent source
NMOS Depletion Load
Use depletion mode NMOS transistor as pull-up
Vtdep of depletion transistor is < 0 V
The depletion mode transistor is always ON:
gate and source connected Vgs = 0
Vin = 0 transistor pull down is off Vout is high
Voltage Output Low
Driver is in linear region with input high
Load is in saturation region
2
2
)(22
)(tdep
loadOL
OLtnDDdriverV
VVVV
Assume: VDD = 5.0V
Vtn = 1.0V = - Vtdep
Proper design: Vol < Vtn
Let: Vol = 0.5V
267.0
load
driver
Gate Threshold Voltage
Assume that both driver and load are in saturation with input Vinv
22
2
)(
)(2
)(2
)(2
dep
load
tinv
driver
tgs
driver
satDS
VVV
VVI
Hence,
driver
load
deptinvVVV
Gate threshold voltage = Vinv
= Input voltage at which Vin = Vout
If driver is increased relative to load then, Vinv decreases
PMOST Load with Constant VGS
Voh = 5.0V
Vol = ???
I = 0.5 p.(Vdd-Vtp)2
I = n Vdd-Vtn)Vol-0.5Vol2
)5.0)((
)(5.0
2
2
ololtndd
tpdd
p
n
VVVV
VV
Sizing for VOL
)5.0)((
)(5.0
2
2
ololtndd
tpdd
p
n
VVVV
VV
Assume: Vdd = 5.0V
Vtn = Vtp = 1.0V
Proper design: Vol < Vth
Let: Vol = 0.5V
26.4
p
n
Sizing for Gate Threshold Voltage
N-device: saturated
2)(
2tnin
n
dsnVVI
)(tninout
VVV
P-device: non-saturated
]2
)())([(
2
DDout
DDouttpDDpdsp
DDgsp
VVVVVVI
VV
Equating the two currents we obtain,
]2
)())([()(
2
2
2 DDout
DDouttpDDptnin
nVV
VVVVVV
Sizing for Gate Threshold Voltage
Solving for Vout
CVVVVtpDDtpout
2)(
Where C = k (Vin - Vtn)2
p
nk
Also,2
22
)(
)()(
tnin
tpouttpDD
p
n
VV
VVVV
To make gate threshold voltage = 0.5VDD
11.6
p
n
Forcing the Voltage Output Low
Propagation Delay of Pseudo-NMOS Inverter
• Use average current
)8
()2
)(()(22
1)(
2
2 DDDD
tpDDPtpDD
P
av
VVVVVVHLI
8)(
2)()()(
22
1)(
2
2 DD
PN
DD
tnDDPtnDDNtnDD
N
av
VVVVVVVVLHI
• Propagation delay
av
DDL
pI
VCt
)2/(
Power Consumption
• Consume power when the output is low
2
,)(
2tpDD
P
lowavVVI
2
,,)(
2tpDDDD
P
lowavDDlowavVVVIVP
Trade-offs to be Considered
• To reduce static power, ILoad should be low
• To obtain a reasonable NML, VOL = ILoadRPDN should
be low
• To reduce tpLH CLVDD/(2ILoad), ILoad should be high
• To reduce tpHL 0.69 RPDNCL, RPDN should be kept
small
Pseudo-NMOS NOR Gate
• Fan-in of N inputs requires only N+1 transistors,
smaller parasitic capacitance and area
• Smaller load to preceding gate
• Static power consumption at output low
• Pseudo-NMOS gates can be used effectively when
speed is importance and majority of the output is high
A B C D
Pseudo-NMOS NAND Gate
VDD
GND
Improved Loads (2)
V DD
V SS
PD N 1
O ut
VDD
V SS
PD N 2
O ut
A
A
B
B
M1 M 2
D ual C ascod e V oltage Sw itch L ogic (D C V S L )
Example
B
A A
B B B
O ut
O ut
X O R -N X O R gate