Transmission gate and Pass-Transistor Logic
Advanced  logic  function  or  switching  scheme 
are  implemented  using  the  feature  of  transistor 
MOS to work as a simple switch   
  It  has  the  advantage  of  being  simple  and  fast. 
Complex  gates  are  implemented  with  the 
minimum  number  of  transistors  (the  reduced 
parasitic capacitance results in fast circuits  
A
  OUT  
OUT = A  B 
 B   
Prof. Gaetano Palumbo  1           
  The  static  and  transient  performance  strongly  depend 
upon  the  availability  of  an  high  quality  switch  with  low 
parasitic resistance and capacitance  
  A  single  transistor  is  used  as  switch:  Pass 
Transistor  
  N- and P-transistor are used: Transmission Gate   
  Implementation with a single transistor reduces the noise 
margin and causes static power consumption         
V
DD  
V
DD
V
DD
-V
tn
    0 
-V
tp 
V
DD 
V
DD
-V
tn 
OUT    
V
DD 
0   
V
DD    
Prof. Gaetano Palumbo  
2        
1        
Transmission Gate  Transmission Gate Simbol  
C 
  C         
A  B 
A  B        
C 
  C         
  Transmission gate has better noise margin than 
pass transistor   
  0    0 
V
DD 
V
DD 
0  0  
V
DD  
V
DD  
Prof. Gaetano Palumbo  
3           
  Transmission  gate  is  very  efficient  to  implement 
some complex gate (MUX, DEMUX and XOR)   
2-input Multiplexer  4-input Multiplexer   
X  Y  
S  A   
A 
X  Y       
OUT 
B   
S  OUT  
X  
B 
Y  
C        
S  X  Y  
OUT = A  S + B  S  D    
X  Y  
OUT = A  X Y + B  X Y + C  X Y + D  X Y  
Prof. Gaetano Palumbo  4         
2       
4-input MUX as cascade of 2-input MUX  
 X 
A   Y  
X 
B   
Y  OUT 
X 
C    
X  
D      
Y 
X 
OUT = A  X Y + B  X Y + C  X Y + D  X Y  
  More parasitic capacitances on internal nodes, less 
on the output node   
Prof. Gaetano Palumbo  5          
2-input XOR  
 Can be implemented with a 2-input MUX, or:  
  B   
A
   OUT 
A  OUT = A  B + A  B    
 B  
  Only  the  left  side  can  implement  the  XOR  logic 
function,  but  a  threshold  voltages  is  lost  at  the 
output without implementing the right side circuit   
Prof. Gaetano Palumbo  6        
3        
Transmission Gate MUX Layout   
    S  S       
V
DD     
  S        
A 
TG1        
INV        
OUT               
S       
OUT      
TG2  
B 
TG2  TG1 
INV        
  S        
    GND      
    A S  S B       
Prof. Gaetano Palumbo  7          
  Transmission Gate Resistance    
30000.0               
      R
n          
            (W/L)
p
=(W/L)
n
 =  
20000.0 
          1.8/1.2                 
R (Ohm) 
    R
p          
          0    
10000.0    
R
eq    
IN 
OUT                        
0.0
0.0 
1.0  2.0  3.0  4.0  5.0  5 V   
      Vout             
Prof. Gaetano Palumbo  8        
4        
  Despite  the  nonlinear  behavior  of  the  NMOS  and 
PMOS  transistors  varying  the  input  voltage,  the 
resistance  of  a  transmission  gate  is  almost 
constant   
  We  can  approximate  its  value  with  that  given  by 
the parallel of transistor resistances, assuming both 
in linear region   
1 
= G
eq 
= 
dI 
D   
+ 
dI 
D  
= 
R
eq 
dV
DS  
dV
DS       
n  
p               
= |
n
 (V
GSn
 V
tn
 V
DSn
 )+ | 
p
 (V
SGp
 + V
tp
 V
SDp
 )    
Prof. Gaetano Palumbo  9          
  If  the  gate-source  voltage  is  around  V
DD
/2,  the 
two transistors are in linear region (V
DS
~0)   
1 
= | 
|V 
V 
| 
+ | 
  |V 
+V 
|     
DD 
|   
DD 
| 
R  
2  
2  
  n \  tn .  p \  tp . 
eq                         
  Assuming the threshold voltage and the gain factor 
of the NMOS and PMOS are equal (PMOS twice 
the NMOS)   
1 
R
eq 
= 
n
C
ox
 (W / L)
n
 (V
DD
  2V
t
 )  
  To  reduce  parasitic  effects  both  transistors 
are generally minimum size   
Prof. Gaetano Palumbo  10        
5        
Transmission Gate Delay   
0    0    0      
V 
C1 
V
DD 
C2 
V
DD 
C3 
V  
DD 
    DD
Cn  
R1    R2    R3    Rn  
  C1    C2   C3  Cn     
Prof. Gaetano Palumbo  11           
 Applying the open-circuit time constant  
n  i  n    
t = C
i
  R 
j
  = RC i = RC 
n(n 
+1) 
i=1  j=1  i=1 
2      
R=R
eq
 and C=4(C
gs
+C
sb
)  
  Approximating the circuit with a pole-dominant 
behavior  t
PD
=  0.69t  .  It  increase  with  the 
square of n   
  We can introduce buffer to minimize t
PD    
Prof. Gaetano Palumbo  12        
6     
R  
R         
R            
R       
R      
R                                                                                                                                                                                                                                                                                            
        C           C      C            C  C          C                                                                                                    
m                                                                                                                         
                            n          m(m +1)(  | n  |                          
t 
PD
 =  0.69
  RC 
2 
(  +  
 
1|t 
PDinv
 
     
                       
                           m             \ m  .                    
 
      d 
t 
PD 
= 0 
                                                 
 
   
dm 
                                             
t 
PDinv 
   
                                                               
                                                                m
opt  = 
2   
 
0.69RC 
n
  
n 
                                               
0.69RC 
 
 
  t
 PDinv  
=
 
0
 
                                           
                                                           
                                                            
2    m
2
                                                         
 
                        Typical value 3-4                       
 
                                                                  
 
                        
Prof. Gaetano Palumbo 
             
13 
 
                                                                                           
 
 
 
 
 
 
 
 
 
Source of errors 
 
 Clock feedthrough due to capacitive coupling 
 
 
Transistor ON  Transistor OFF 
 
 
 
V  V  V  V  V 
V
OUT 
A  A  A  A  A     
          V
t
 
 
V 
C
L
    C
L
    C
L
 
C
GS   
C
GS    C   
DD   
V
A
+V
t
 
  GS 
 
         
 
 
 
 
 
Prof. Gaetano Palumbo  14 
 
 
 
 
 
 
 
7 
 
 
 
 
 
 
 
 
 
Charge conservation yields 
 
(C
L
 + C
GS
 )V
OUT
  = C
L
V
A
  C
GS
V
t
 
 
 
 
 
 
 
AV  = V  V 
A 
=  
C
GS 
(V 
A 
+V ) 
 
   
OUT1  OUT     C
L
 + C
GS
  t 
 
                   
 
Reduce with the reduction of C
GS
 
 
 
 
Prof. Gaetano Palumbo  15 
 
 
 
 
 
 
 
 
 
 
  Charge  Injection  due  to  redistribution  of  the 
charge in the channel  
 
 
Transistor ON (triode region) 
 
V 
V
DD 
V
A
 
A     
 
Oxide 
n+       n+ 
 
p 
 
V
A
  V
A
 
 
 V
DD
 
C
L
 
 
 
 
Q
ch
 = C
ox
WL(V
DD
 V
A
 V
t
 ) 
 
 
 
Prof. Gaetano Palumbo  16 
 
 
 
 
 
 
 
8 
 
 
 
 
 
 
 
When  the  transistor  switch  off  the  charge  in  the 
channel is lost. If the clock edge are sufficiently steep 
the  charge  distribute  equally  on  the  two  diffusion 
node 
Q
ch
 /2 
Q
ch
/2 
Oxide 
n+  n+ 
p 
 
 
 
 
       
 
AV  =  
1
 
C
ox
WL(V
DD
 V
A
 V
t
 )   
 
       
  OUT 2  2  C
L
   
 
         
 
   
Prof. Gaetano Palumbo  17 
 
 
 
 
 
 
 
 
 
 
 
  Clock feedthrough and Charge Injection are lower 
reducing the transistor dimensions (i.e., W and L)  
 
  Both the errors are reduced in Transfer Gates if the 
transistor are of equal dimensions.  
 
  the  charge  in  the  channel  of  equal  NMOS 
and  PMOS  transistors  are  equal  but  with 
opposite sign  
 
  If NMOS and PMOS transistors are equal  
and switch off symmetrically the charge in 
the C
GS
 are equal but opposite  
 
 
Prof. Gaetano Palumbo  18 
 
 
 
 
 
 
 
9 
 
 
 
 
 
 
 
NMOS-Only TG with Level Restorer 
 
 
  Noise  margin  reduction  of  a  pass  transistor  is  avoided 
with  a  PMOS  transistor  in  positive  feedback  which 
restores the output value from V
DD
- V
t
 to V
DD
  
 
V
DD
   V
DD
 
 
 
 
OUT 
 
 
 
 
 
Prof. Gaetano Palumbo  19 
 
 
 
 
 
 
 
 
 
 
 
  Only a PMOS transistor is needed for a set of TG which 
have a common output node  
 
  Clock feedthrough  and  Charge Injection are higher than 
than in complementary TG  
 
  Reduce the static power consumption of the next inverter 
but introduce a contribute due to itself  
 
  Design strategy is close to that of ratioed logic  
 
 
 
 
 
Prof. Gaetano Palumbo  20 
 
 
 
 
 
 
 
10 
 
 
 
 
 
 
 
Design of an NMOS-Only TG 
 
  The  critical  case  is  when  the  inner  node  was  previously 
restored  to  V
DD
  (output  voltage  equal  to  0),  the  first 
inverter has the input to V
DD
 and MR goes from 0 to V
DD
  
 
  During  the  transient  a  direct  path  between  the  power 
supply and ground exists  
V 
V
DD   
V
 
 
DD  DD   
 
M2
  MR   
V
DD  MA   
  X  OUT 
  M1   
  IN   
 
Prof. Gaetano Palumbo  21 
 
 
 
 
 
 
 
 
 
 
  To  change  the  output  state  node  X  must  be  lower  than 
the  inverter  threshold  voltage  V
DD
/2  (the  positive 
feedback complete the inverter commutation)  
 
  To  guarantee  the  above  condition  consider  the  Pseudo 
NMOS gate M1-MA-MR  
 
V
DD
 V
DD
 
MR 
V
DD                MA 
X 
 
M1 
V
DD 
 
 
Prof. Gaetano Palumbo  22 
 
 
 
 
 
 
 
11 
 
 
 
 
 
 
 
Transistor PMOS is in triode region 
 
W 
|2(V
DD
+V
tp
)(V
DD
V
OL
)|(V
DD
V
OL
)
p
 
  n   
L  +L   
 
M1  MA= 
|2(V  V )V  |V 
   
(W/ L)   
  MR  DD   tnOL  OL  n 
 
W 
n 
L
M1
 +L
MA=
p 
(W/ L)
MR
   
n
 
 
 If M1-MA are minimum size transistors 
 
  |W |  W 
min 
 
 
 
  |  s 
     
 
L
min 
   
  \  L .
MR
   
 
              
 
Prof. Gaetano Palumbo  23 
 
 
 
 
 
 
 
 
 
 
CPL 
(Complementary Pass-Transistor Logic) 
 
  The  logic  is  differential  since  two 
complementary data path are implemented  
 
I1 
pass transistor 
I1 
complementary 
 
I2  I2 
 
In 
network 
In 
  pass transistor 
 
F 
  network 
 
I1  I1   
F 
 
I2  I2 
 
 
       
In    In     
 
  F      F 
 
 
Prof. Gaetano Palumbo  24 
 
 
 
 
 
 
 
12 
 
 
 
 
 
 
 
 
  Although  differential  signal  requires  extra 
circuitry,  the  differential  style  results 
advantageous  in  term  of  transistor  number  to 
implement some complex gates such as adders  
 
  The  availability  of  both  polarity  of  every  signal 
eliminates  the  need  for  extra  inverter  (addition 
speed up).  
 
  The  design  is  very  modular.  All  gates  use 
exactly  the  same  topology,  only  the  input  are 
permutated.  Complex  gates  are  implemented 
with cascade of standard modules  
 
 
 
Prof. Gaetano Palumbo  25 
 
 
 
 
 
 
 
 
 
 
  NMOS  transistors  with  V
tn
<-V
tp
  are  needed  to 
avoid  static  power  consumption.  (The  reduced 
threshold improve the switching speed)  
 
  The low V
tn
 leads higher subthreshold current  
 
 
 
 
higher power supply 
 
 
 
 The low V
tn
 determines higher noise margin 
 
 
 
Prof. Gaetano Palumbo  26 
 
 
 
 
 
 
 
13 
 
 
 
 
 
 
 
2-input OR/NOR  2-input XOR/XNOR 
A    B  B    A 
A   A  A    A 
 
     
B 
 
B 
 
 
       
B 
 
B 
 
 
       
A+B  A+B 
A + B  A + B 
 
     
  2-input AND/NAND   
 
 
B    A  BA 
 
 
 
B 
 
B 
 
 
AB  AB 
 
Prof. Gaetano Palumbo  27 
 
 
 
 
 
 
 
 
 
 
 
3-input AND/NAND  3-input OR/NOR 
CA   B  AB    C  CA   B  AB    C 
A    A   
B    B   
A    A   
B    B   
ABC  ABC  A+B+C  A+B+C 
 
 
 
 
Prof. Gaetano Palumbo  28 
 
 
 
 
 
 
 
14 
 
 
 
 
 
 
 
Layout of a 4-input NAND CPL 
 
 
A   X 
B 
B  A  OUT   C  D 
       
 
 
 
 
 
    X   Y   
 
    X   Y   
 
B  A  OUT    C  D 
 
A       
 
B  X 
Y  ABCD 
     
    X   
 
B  B     
 
C  Y   Y   
 
D 
 
X  ABCD 
C 
 
X  X 
 
     
D  Y 
   
 
D  D 
   
 
 
 
 
Prof. Gaetano Palumbo  29 
 
 
 
 
 
 
 
 
 
 
 
 
A       
 
B 
X     
 
     
 
A 
     
 
B 
X     
 
 
Y  ABCD 
 
     
B  B 
X 
 
 
   
 
C   
Y 
 
 
D 
Y   
 
 
X 
 
 
    ABCD 
 
C 
     
 
D 
Y  X  X 
 
       
D  D 
   
 
 
Prof. Gaetano Palumbo  30 
 
 
 
 
 
 
 
15