Transcript Mosfetpart2
NMOS Inverter (E-MOSFET Driver and Load)
Load
iDS
V
1.2V
Th
2
W
1
m
, L
10
m
2
2
W
1
6 A
2
K
6
.
2
x
10
2
L
10
V2
2
Driver
V
1.0V
Th
1
W
m, L11m
1 5
W
4 A
1
5 K
3
.
1
x
10
1
L
V2
1
Transistor Characteristics
2
Electron
mobility
700
cm
/Vsec
n
vDS
6
Gate
oxide
thicknes
s tox20
nm
2
x
10
cm
7
2
Gate
oxide
capacitanc
e C
ox/tox1.77
x
10
F/cm
ox
1
W1
W
2
7
2
K
C
700
cm
/Vsec)(
1
.77
x
10
F/cm
)
n
ox (
2
L 2
L
W
5
2
6
.2
x
10
A
/V
L
ECES 352 Winter 2007
Ch 10 MOS Digital
1
NMOS Inverter (E-MOSFET Load Transistor)
G
*
Load transistor has its gate
connected to its drain so
vDS2 = vGS2 (Always)
*
Triode-saturation boundary at
vDSsat2 = vGS2 - VTh2
*
Since VTh2 > 0, for load transistor
vDS2 > vDSsat2 = vGS2 VTh2
= vDS2 VTh2
*
So load transistor is ALWAYS in
saturation ! It cannot operate in the
triode region !
Load’s drain current is 2
always given
i
K
v
V
2
2 GS
2 Th
2
by D
D
S
iDS2
0.08
*
0.06
V
2
K
v
V
2
DD
o
Th
2
0.04
*
0.02
So load transistor
looks
like a
vDS
vDS
2
2
variable
resistance
R
of size
2
eff
2
iD2
vDS2
ECES 352 Winter 2007
Ch 10 MOS Digital
K
vDS
V
2
2
Th
2
V
v
DD
o
2
K
V
v
V
2
DD
o
Th
2
2
NMOS Inverter (E-MOSFET Driver and Load)
* Voltage transfer characteristic (Vo vs Vi)
* Region I (A to B)
0 < Vi < VTh1
iD1 = 0 since drive transistor Q1 is off,
i.e. in cutoff.
iD2 = 0 since iD2 = iD1 and iD1= 0
* Output voltage is given by
Vo= VDD-VDS2 = VDD-VGS2
= VDD-VTh2
= 5V - 1.2V = 3.8V
Load
V
.2V
Th
2 1
Driver
C
VTh1 1.0V
vo
5V
I
3.8V A
B
0
0
VTh1
=1.0V
ECES 352 Winter 2007
5V
vi
Why not 5V ? Assuming a capacitance load
on the output, when Q1 (driver) turns off,
then Q2 (load) provides current to charge
up the capacitance C so the output Vo can
rise towards VDD.
As Vo rises, VDS2 (still = VGS2) decreases.
When VGS2 decreases to VTh2, then Q2
turns off and stops supplying the charging
current, so Vo cannot rise further.
So Vo cannot rise above VDD - VTh2 = 3.8V
Ch 10 MOS Digital
3
NMOS Inverter (E-MOSFET Driver and Load)
* Region II (B to C)
Driver comes on in saturation mode
Load always in saturation mode.
Load
VTh2 1.2 V
W
m, L2 10
m
2 1
iD1 K1vGS1 VTh1 K1vi VTh1
2
A
K2 6.2x10 2
V
6
iD1 iD2 so
Driver
K1vi VTh1 K2VDDvo VTh2
2
VTh1 1.0 V
W1 5m, L1 1m
K1 3.1x104
2
vo VDDVTh2
A
V2
2
K1
vi VTh1
K2
So vo 3.8V 50vi 1V
vo
iDS1
5V
Driver
B
3.8V A
I
C
A to B
C
vDS1
ECES 352 Winter 2007
II
0
0
Ch 10 MOS Digital
VTh1
=1.0V
5V
4
vi
NMOS Inverter (E-MOSFET Driver and Load)
* Where is point C?
* Driver transistor operating in saturation mode in
region II, so point C is where it leaves saturation,
i.e. where vDS1 = vo = vDSsat1 = vGS1 - VTh1 = vi - VTh1
K
v
(
at
C
)
V
V
1v
at
C
)
V
v
(
at
C
)
v
at
C
)
V
o
DD
Th
2
i(
Th
1 and
o
i(
Th
1so
K
2
V
V
5
1
.
2
V
DD
Th
2
v
at
C
)
V
1
.
0
V
1
.
47
V
i(
Th
1
K
1
50
1
1
K
2
so
v
(
at
C
)
v
at
C
)
V
1
.
47
V
1
.
0
V
0
.
47
V
o
i(
Th
1
iDS1
vo
3.8V A
B
I
C A to B
vDS1
ECES 352 Winter 2007
II
C
0.47V
0
0
Ch 10 MOS Digital
VTh1 1.47V
=1.0V
5
5V
vi
NMOS Inverter (E-MOSFET Driver and Load)
*
*
*
Where is the equivalent resistance that the load
transistor provides in region II ?
Recall the load transistor acts like a load resistor with
an effective resistance
v
V
v
5
v
DS
2
DD
o
0
R
2
eff
2
2
62
i
K
V
v
V
6
.
2
x
10
A
/
V
3
.
8
v
D
2
2
DD
o
Th
2
o
At point B, vo = 3.8 V and the load transistor is off so
v
5
3
.
8
DS
2
R
at
B
2
eff
2
6 2
i
6
.
2
x
10
A
/
V
3
.
8
3
.
8
D
2
*
iDS1
At point C, vo = 0.47 V so
v
5
0
.
47
DS
2
R
at
C
67
K
2
eff
2
6 2
i
6
.
2
x
10
A
/
V
3
.
8
0
.
47
D
2
vo
3.8V A
B
I
C A to B
vDS1
ECES 352 Winter 2007
II
C
0.47V
0
0
Ch 10 MOS Digital
VTh1 1.47V
=1.0V
6
5V
vi
NMOS Inverter (E-MOSFET Driver and Load)
* Region III (C to D)
Driver in triode mode
Load
VTh2 1.2 V
2vV vv and
K
W
m, L2 10
m
2 1
6
K2 6.2x10
2
vGS
iD
K
V
v
v
1
12
1
Th
1
DS
1
DS
1
A
V2
1
i
Th
1 o
2
o
iD
iD2 and
the
load
isstill
in
saturat
so
1
Driver
VTh1 1.0 V
2
vi V
K
v
v
K
V
v
V
12
Th
1
o
o
2
DD
o
Th
2
W1 5m, L1 1m
2
This
gives
v
v
o(
i)
A
K1 3.1x10 2
V
4
vo
iDS1
5V
3.8V A
D
B
I
C
A to B
vDS1
ECES 352 Winter 2007
II
C
0.47V
0
Ch 10 MOS
III
0
VTh1
Digital=1.0V
D
1.47V
7
5V
vi
NMOS Inverter (E-MOSFET Driver and Load)
* Region III (C to D)
2
2
vi V
vovo
V
K
K
v
V
12
Th
1
2
DD
o
Th
2
2
2
K
V
vi V
voK
V
K
K
v
2
V
K
V
0
1
2
o
2
DD
Th
2
1
Th
1
2
DD
Th
2
Point D
4 2
5
4
5
2.36
3.16
8
x
10
v
2
x
10
3
.1
x
10v
1
.0
V
.95
x
10
0
o
i
v
at
D
)5V so
i(
2
1.81
v
1
.96
v
v
0
.28
0
o
i
o
v
at
D
)4
.00
V
3
.96
V0
.04
V
o(
2
1.81
1
.96
v
1
.81
1
.96
v
4
(0
.28
)
i
i
v
o
2
When
v
3
.8V
, v
0
.05
V
i
o
vi0
0.98
20.28
v
.98
v
0
.9
v
0
.9
o
i
i
vo
iDS1
5V
3.8 V A
D
B
I
C
III
0.47 V
A to B
C
vDS1
ECES 352 Winter 2007
II
0.05 V 0
Ch 10 MOS
0
VTh1
Digital=1.0 V
D
1.47 V 3.8 V 5V
8
vi
Noise Margins for NMOS Inverter
(E-MOSFET Driver and Load)
* Noise margin for low state
high
high
low
vo
5V
VOH = 3.8 V
VOL=
0.05V
Vi =VOL
= 0.05V
ECES 352 Winter 2007
NML= VIL - VOL
=1.0V- 0.05V = 0.95V
5V
VIL=VTh1
Vi =VOH
= 1.0V
= 3.8 V
vi
Ch 10 MOS Digital
Measures degree of inverter
sensitivity to noise for the low
state, i.e. how large a noise
signal causes problems.
Assumes identical inverter
providing input signal
Noise Margin =
NML
= VIL - VOL where
VOL = output voltage
when input set to VOH
VIL = maximum input
voltage recognized as a
low input
For this inverter design, NML
is much larger than for
resistor load case!
Can change NML by changing
K’s or VTh1.
9
Noise Margins for NMOS Inverter
(E-MOSFET Driver and Load)
*
low
low
high
Noise margin for high state
Noise Margin = NMH = VOH – VIH where
VOH = output high voltage when
input set to VOL
VIH = minimum input voltage
recognized as a high input
Maybe find VIH by using vo(vi) for
region II (Q1 in saturation). Recall
vo
I
VOH = 3.8 V
Slope
= -1
II
K
then
v
V
V
1
v
V
o
DD
Th
2
i
Th
1
K
2
III
dv
K
o
1
50
1This
is
not
possi
dv
K
i
2
NMH= VOH - VIH
= 3.8 V- 1.53 V
= 2.27 V
vo0.98
vi 0.9 0.98
vi 0.920.28
Vo= 0.32 V
VOL=
0.05V
So must find VIH by using vo(vi) for the
device in region III.
vi
VIH=1.53 V
ECES 352 Winter 2007
Vi =VOH
= 3.8 V
0.98
dv
vi 0.9
o
0.98
1
2
dv
i
0.98
vi 0.9 0.28
vi 1.53
VV
IH
and
vo(vi V
)0.32
V
IH
Ch 10 MOS Digital
10
NMOS Inverter (E-MOSFET Driver and Load)
Load
* Load transistor Q2 acts as a variable
load resistance.
* Since VTh2 > 0 for load transistor
(enhancement type), and vDS2 = vGS2
then
vDS2 > vDSsat2 =
vGS2 - VTh2
= vDS2 VTh2
* So load transistor is ALWAYS in
v
2current is
iD
K
V
saturation
mode,
so
its
2
2
GS
2
Th
2
VTh2 1.2 V
W
m, L2 10
m
2 1
6
K2 6.2x10
A
V2
Driver
VTh1 1.0 V
W1 5m, L1 1m
K1 3.1x104
A
V2
V
2
K
v
V
2
DD
o
Th
2
iDS1
vDS
V
vo
2
DD
* Effective
resistance
ofLoad
QR
is2eff(vo)
R
2eff
2 2
Load line for
variable
load resistance.
iD2
K
V
vo V
2
DD
Th
2
Ex
.
D
C
Atvo 3.5V R
.8M
2eff2
A to B
Atvo 0.5V R
K
2eff60
vDS1
ECES 352 Winter 2007
Atvo 0.05
VR
K
2eff 57
Ch 10 MOS Digital
11
Propagation Delays and Switching Times
for NMOS Inverters
Load
VTh2 1.2 V
iD2
W
m, L2 10
m
2 1
iC
6
K2 6.2x10
A
V2
Driver
C
VTh1 1.0 V
W1 5m, L1 1m
iD1
K1 3.1x104
A
V2
* Output goes from Low to High
Driver Q1 turns off
Load Q2 provides current to
charge up C.
dv
iC
C o
iD
2
dt
v
2K
V
2
iD
K
V
V
v
2
2
GS
2
Th
2
2
DD
Th
2
o
dv
o K
2
2
V
V
v
DD
Th
2
o
dt C
* Output goes from High to Low
* Driver Q1 turns on to discharge C (So vi = VOH = 3.8 V)
* Driver initially in saturation mode, then triode mode
* Load Q2 produces current that must be absorbed by Q1.
iD1iD2iC
dv
2
iCC o andiD2 K2V
V
DD
Th
2v
o
dt
2
2
Saturation iD1K
V
V
V
1v
GS
1
Th
1 K
1
OH
Th
1 or
Triode
K2V
2
vGS1VTh1vDS1vDS
iD1K
12
1
1
dv
2 K
2
o K
2
V
V
v
1
V
V
or
DD
Th
2
o
OH
Th
1
dt C
C
dv
2 K
2
o K
1
2
V
V
v
V
V
v
v
DD
Th
2 o 2
OH
Th
1
o
o
dt C
C
i
i
driver
can
discharge
the
capaci
e
D
1
D
2 so
2
V
OH
Th
1v
o v
o
ECES 352 Winter 2007
These equations
are not as easily
integrated to find
vo (vi ).
Ch 10 MOS Digital
12
Propagation Delay for NMOS Inverter
*
Load
iD2
iC
Driver
C
iD1
Output goes from High (VOH = 3.8V) to Low (VOL = 0.05V)
Driver Q1 (starts from P R S T)
At outset, Q1 is off (P), and vDS1 = vo = VOH = 3.8V, vi < VTh1
Driver turns on (P to R) when vGS1 is switched to VOH = 3.8 V.
Driver initially in saturation mode, then moves into triode
as capacitor discharges and vDS1 decreases as Q1 moves
along constant vGS1 characteristic (R S T).
Q1 ends at (T) , where vGS1 = vi = 3.8V and
vDS1 = vo = VOL = 0.05V.
Load Q2 (goes from R’ S’ T’) (always in saturation)
Driver
Load
iDS1
iDS2
T’
0.08
vGS2 = vDS2 = VDD-VOL = 4.95V
R
S
0.06
vGS1 = 3.8 V
S’
0.04
0.02
T
P
vo =VOL
vo =VOH = 3.8 V
ECES 352 Winter 2007
= 0.05 V
vDS1
vGS2 = vDS2 = VDD-VOH = 1.2V
vDS2
R’ P’
Ch 10 MOS Digital
13
Approximate Analysis of Propagation Delays
*
*
Load
iD2
iC
Driver
dv
i
Co
i
i
i
C
C
D
2
D
1
dt
where
i
i
discharge
the
C
D
1
D
2 to
C
*
We select point S to be the point where the output
voltage has fallen halfway from its peak value VOH to
its minimum value VOL.
1
1
v
S
V
V
3
.
8
V
0
.
05
V
1
.
93
V
o
OL
OH
2
2
*
We define the high to low propagation time tPHL as the
time it takes for the output to go from vo = vo(R) = VOH =
3.8 V to vo=vo(S) = 1/2(VOH +VOL) = 1.93 V.
v
v(S)vo(R
) C 1
V
V
iCC o Co
V
OL
OH
OH
t
tPHL
tPHL
2
iD1
Driver
iDS1
S
T
Output goes from High (VOH = 3.8V) to Low (VOL = 0.05V)
As an approximation, we use average currents for the
transistors to calculate an average discharge current for
the capacitor.
R
vGS1 = 3.8 V
P
vo =VOL
vo =VOH = 3.8 V
ECES 352 Winter 2007
= 0.05 V
C 1
V
V
0
OL
OH
tPHL
2
C1
C 1
V
tPHL
V
V
V
OL
OH
OL
OH
iC2
iD2iD
12
vDS1
C 1
V
V
0
OH
OL
iD
iD22
1
Ch 10 MOS Digital
14
Approximate Analysis of Propagation Delays
Load
iD2
*
*
Output goes from High (VOH = 3.8V) to Low (VOL = 0.05V)
Average current for the DRIVER transistor.
*
At point R, vi = vi(R) = VOH =3.8V and driver is in
saturation region since the output has not fallen so
vo = vo(R) = VOH
2
2
i
(
R
)
K
v
V
K
V
V
D
1
1
GS
1
Th
1
1
OH
Th
1
iC
Driver
C
1
i
i
(
R
)
i
(
S
)
D
1
D
1
D
1
2
2
4A
3
.
1
x
10
3
.
8
V
1
.
0
V
2
.
4
mA
2
V
iD1
*
Driver
At point S,
1
1
v
v
S
V
V
3
.8
V
0
.05
V
1
.93
V
o
o
OL
OH
2
2
But
v
v
V
V
V
3
.8
1
.0
2
.8
V
DSsat
1
GS
1
TH
1
OH
TH
1
so
v
v
S
1
.93
V
v
2
.8
V
DS
1
o
DSsat
1
iDS1
sothe
driver
is
in the
triode
region
at
point
S!
*
S
T
R
vGS1 = 3.8 V
P
vo =VOH = 3.8 V
ECES 352 Winter 2007
*
vDS1
So drain current at point S is2
given by
2
i
S
K
2
v
V
v
v
K
2
V
V
v
v
D
1
1GS
1Th
1
DS
1DS
1
1
OH
Th
1
o
o
A
4
2
3
.
1
x
10
2
3
.
8
V
1
.
0
V
1
.
93
V
(
1
.
93
V
)
2
.
2
m
2
V
So average
drain current for1
the driver is
1
i
i
(
R
)
i
(
S
)
2
.
4
mA
2
.
2
mA
2
.
3
m
D
1
D
1
D
1
2
2
Ch 10 MOS Digital
15
Approximate Analysis of Propagation Delays
*
*
Load
iD2
1
i
i
(
R
')
i
(
S
')
D
2
D
2
D
2
2
iC
*
Driver
C
*
At point S’,
1
1
v
v
S
V
V
3
.
8
V
0
.
05
V
1
.
93
V
o
o
OL
OH
2
2
so
v
V
v
S
5
V
1
.
93
V
3
.
07
V
v
DS
2
DD
o
GS
2
*
So drain current at point S’ is given by
Load
T’
0.08
vGS2 = vDS2 = VDD-VOL
= 4.95V
0.06
vGS1 = 3.8 V
0.04
At point R’, vGS2 = VDD -VOH= 5.0V - 3.8V=1.2V and the
load is in saturation but barely on so
2
2
i
(
R
'
)
K
v
V
K
1
.
2
V
1
.
2
V
0
D
2
2
GS
2
Th
2
2
iD1
iDS2
Output goes from High (VOH = 3.8V) to Low (VOL = 0.05V)
Average current for the LOAD transistor.
A
2
2
6
i
(
S
'
)
K
v
V
6
.
2
x
10
3
.
07
V
1
.
2
V
D
2
2
GS
2
Th
2
2
V
5
2
.
2
x
10
A
0
.
022
mA
S’
*
0.02
vGS2 = vDS2 = VDD-VOH = 1.2V
R’
ECES 352 Winter 2007
vDS2
So average
drain current 1
for the load is
1
i
i
(
R
'
)
i
(
S
'
)
0
mA
0
.
022
mA
0
.
01
m
D
2
D
2
D
2
2
2
Ch 10 MOS Digital
16
Approximate Analysis of Propagation Delays
*
*
Load
iD2
i
i
2
.
3
mA
0
.
011
mA
2
.
289
mA
2
.
3
m
D
1
D
2
iC
Driver
Output goes from High to Low.
Using the average currents for the transistors we get
an average discharge current for the capacitor
*
Defining the high to low propagation time tPHL as the
time it takes for the output to go from vo = VOH =
3.8V to vo = 1/2(VOH +VOL) = 1.93V.
C
1
t
V
V
PHL
OH
OL
i
i
2
D
1
D
2
*
For a capacitance load of 10 pF, we get
C
iD1
Driver
*
iDS1
C 1
VOHVOL
tPHL
iD1 iD2 2
10pF 1
3
.
8
V
0
.
05
V
2.3 mA
2
8
8x10
sec0.8 nsec
S
R
vGS1 = 3.8 V *
*
T
P
vDS1
This propagation delay is small since the driver’s
current is much larger than the load’s current
(since K1 >> K2).
NOTE: The load in this case (high to low) is
delaying (slightly) the transition by supplying current
which tends to charge up the capacitance load.
vo =VOH = 3.8 V
ECES 352 Winter 2007
Ch 10 MOS Digital
17
Propagation Delay for NMOS Inverter
* Output goes from Low (VOL= 0.05V) to High (VOH = 3.8V)
Load
iD2
iC
Driver
C
Driver Q1 turns off and remains off (starts from T P)
At outset, vo = VOL = 0.05V, vDS2 =VDD - vo = 5V-0.05V = 4.95V
As the load Q2 charges up the capacitor, vo increases and
vDS2 (and vGS2 ) decreases, and the load Q2 goes from
T’
S’ R’ (Q2 always remains in saturation).
At S’,
1
1
v
v
S
V
V
3
.
8
V
0
.
05
V
1
.
93
V
o
o
OL
OH
2
2
so
v
S
'
V
v
S
5
V
1
.
93
V
3
.
07
V
v
S
'
DS
2
DD
o
GS
2
iD1
Driver
Load
iDS2
iDS1
T’
0.08
vGS2 = vDS2 = VDD-VOL = 4.95V
0.06
S
R
vGS1 = 3.8 V
0.04
S’ vGS2 = vDS2 = 3.07V
0.02
T
P
vo =VOH = 3.8 V
ECES 352 Winter 2007
vGS2 = vDS2 = VDD-VOH = 1.2V
vDS1
vDS2
R’
Ch 10 MOS Digital
18
Approximate Analysis of Propagation Delays
*
*
Load
iD2
iC
Output goes from Low (VOL = 0.05V) to High (VOH = 3.8V).
As an approximation, we use an average current for the
load transistor to calculate an average charging current
for the capacitor.
iC iD2 iD1 iD2 0
so
Driver
C
dv
iC C o iD2
dt
iD1
*
We select point S’ to be the point where the output
voltage has risen halfway from its low value VOL to
its
peak value VOH.
1
1
v
S
'
V
V
3
.
8
V
0
.
05
V
1
.
93
V
o
OL
OH
2
2
*
We define the low to high propagation time tPLH as the
time it takes for the output to go from vo = vo(T’) = VOL
to vo=vo(S’) = 1/2(VOH +VOL).
v
v(
S
')
v
(
T
') C
1
o
i
C o
Co
V
V
V
C
OL
OH
OL
t
tPLH tPLH
2
Load
iDS2
T’
0.08
vGS2 = vDS2 =
VDD-VOL = 4.95V
0.06
0.04
C
1
V
V
0
OH
OL
tPLH
2
S’
0.02
vGS2 = vDS2 = VDD-VOH = 1.2V
R’
ECES 352 Winter 2007
C
1
1
C
t
V
V
V
V
0
PLH
OH
OL
OH
OL
vDS2
i
2
i
2
C
D
2
Ch 10 MOS Digital
19
Approximate Analysis of Propagation Delays
*
Load
iD2
*
iC
Driver
C
iDS2
At point T’, vGS2 = VDD -VOL=5.0V - 0.05V = 4.95V
and the load is in saturation so
A
2
2
6
i
(
T
'
)
K
v
V
6
.
2
x
10
4
.
95
V
1
.
2
V
D
2
2
GS
2
Th
2
2
V
5
8
.
7
x
10
A
0
.
087
mA
*
At point S’,
1
1
v
v
S
'
V
V
3
.
8
V
0
.
05
V
1
.
93
V
o
o
OL
OH
2
2
so
v
v
V
v
S
'
5
V
1
.
93
V
3
.
07
V
GS
2
DS
2
DD
o
T’
0.08
1
i
i
(
T
')
i
(
S
')
D
2
D
2
D
2
2
*
iD1
Load
Output goes from Low (VOL = 0.05V) to
High (VOH = 3.8V).
Average current for the load transistor.
vGS2 = vDS2 = VDD-VOL = 4.95V
* So drain current at point S’ is given by
A
2
2
6
i
(
S
'
)
K
v
V
6
.
2
x
10
3
.
07
V
1
.
2
V
D
2
2
GS
2
Th
2
2
V
0.06
0.04
S’
5
2
.
2
x
10
A
0
.
022
mA
vGS2 = vDS2 = 3.07V
*
0.02
vGS2 = vDS2 = VDD-VOH = 1.2V
R’
ECES 352 Winter 2007
vDS2
So average
1drain current for the load is
i
i
(
T
')
i
(
S
')
D
2
D
2
D
2
2
1
0
.
087
mA
0
.
022
mA
0
.
055
mA
2
Ch 10 MOS Digital
20
Approximate Analysis of Propagation Delays
*
*
Load
iD2
dv
iCC o iD2
dt
iC
Driver
*
Defining the low to high propagation time tPLH
as the time it takes for the output to go from
vo = VOL = 0.05V to vo = 1/2(VOH +VOL) = 1.93V.
C
1
t
V
V
PLH
OH
OL
i
2
D
2
*
For a capacitance load of 10 pF, we get
C
1
pF
1
10
t
V
V
3
.
8
V
0
.
05
V
PLH
OH
OL
i
2
.
055
mA
2
0
D
2
6
3
.
4
x
10
sec
340
n
sec
*
This propagation delay is very large since the
load’s current is much smaller than the driver’s
current (since K2 << K1).
The load transistor in this case is causing an
excessive delay in the transition from low to high
output by supplying only a very small current to
charge up the capacitance load.
Also, as the output voltage rises, vgs2
decreases so the load transistor supplies less
and less current to charge up the capacitor.
C
iD1
Load
iDS2
Output goes from Low to High.
Use the average current for the load transistor
to calculate the charging time.
T’
0.08
vGS2 = vDS2
= VDD-VOL = 4.95V
0.06
vGS1 = 3.8 V
0.04
S’
0.02
vGS2 = vDS2 = VDD-VOH = 1.2V
R’
ECES 352 Winter 2007
vDS2
*
Ch 10 MOS Digital
21
Average Propagation Delay for NMOS Inverter
*
Load
iD2
iC
*
Driver
C
iD1
*
Driver
Output goes from High (VOH = 3.8V) to Low (VOL = 0.05V)
Driver on, providing discharge current (iD1 = 2.3 mA)
Load on, (iD2 = 0.011 mA) delaying output fall
Propagation delay tPHL=0.8 nsec.
Output goes from Low (VOL = 0.05V) to High (VOH = 3.8V).
Driver off (iD1 = 0).
Load providing charging current (iD2 = 0.055 mA)
Propagation delay tPLH = 340 nsec.
Inverter’s average propagation delay is tp
0
.
8
nsec
340
nse
t
t
PHL
PLH
t
17
ns
p
2
2
Load
iDS1
iDS2
T’
0.08
vGS2 = vDS2 = VDD-VOL = 4.95V
R
S
0.06
vGS1 = 3.8 V
S’
0.04
vGS2 = vDS2 = 3.07V
0.02
T
P
vo =VOH = 3.8 V
ECES 352 Winter 2007
vDS1
vGS2 = vDS2 = VDD-VOH = 1.2V
R’
Ch 10 MOS Digital
22
vDS2
Static Power Dissipation for NMOS Inverter
*
Load
iD2
iC
Driver
*
C
*
*
iD1
Driver
Output High (VOH = 3.8 V)
Driver off (iD1=0), Load on, but iD2= iD1 = 0.
No power dissipation in static high output mode (PH=0).
Output Low (VOL = 0.05 V).
Driver on and operating at point T (vo = 0.05 V).
Load on and operating at point T’ where iD2(T’) = 0.087 mA.
Power dissipation PL = iD2(T’)VDD = (0.087 mA) 5 V = 0.42 mW.
Inverter’s average power dissipation is PD= 1/2(PL+PH) = 0.21 mW.
Power delay product DP
DP = PD tp
-11
= 0.21 mW (170 nsec) = 3.6x10 J = 36 pJ.
Load
iDS1
iDS2
T’
0.08
vGS2 = vDS2 = VDD-VOL = 4.95V
R
S
0.06
vGS1 = 3.8 V
S’
0.04
0.02
T
P
vo =VOH = 3.8 V
ECES 352 Winter 2007
vDS1
vGS2 = vDS2 = 3.07V
vGS2 = vDS2 = VDD-VOH = 1.2V
vDS2
R’
Ch 10 MOS Digital
23
Comparison of NMOS Inverters
* Noise Margins
NML = 0.05 V, NMH = 3.17 V
* Propagation Delay
* Noise Margins
NML = 0.95 V, NMH = 2.27 V
* Propagation Delay
1
1
1
1
t
t
t
4
.
5
n
sec
14
n
sec
10
n
sec
t
t
t
0
.
8
n
sec
340
n
sec
17
n
s
PD
PHL
PLH
PD
PHL
PLH
2
2
2
2
* Power Dissipation
* Power Dissipation
* Power-Delay Product
* Power-Delay Product
(
0
10
mW
)
(
0
0
.
42
mW
)
1
1
P
P
P
5
mW
P
P
P
0
.
21
mW
HL
H
L
2
2
2
2
DP
P
t
5
mW
(
10
n
sec)
50
pJ
PD
ECES 352 Winter 2007
DP
P
t
0
.
21
mW
(
170
n
sec
36
p
PD
Ch 10 MOS Digital
24
How to Improve the NMOS Inverter?
* How to reduce the tPLH ?
* Increase the load current by
increasing the load’s W/L ratio.
2
2
i
K
v
V
K
V
v
V
D
2
2
GS
2
Th
2
2
DD
o
Th
2
W
1
m
, L
10
m
2
2
W
1
6A
2
K
6
.2
x
10
2
2
L
V
2 10
* However, that increases the fall time
* Noise Margins
tPHL since the load provides current
NML = 0.95 V, NMH = 2.27 V
that the driver must absorb to
* Propagation Delay
discharge the capacitor. The net effect
1
1
may be some improvement in the tPD.
t
t
t
0
.
8
n
sec
340
n
sec
170
n
sec
PD
PHL
PLH
2
2
* However, there will be an increase
in the power dissipation.
(
0
0
.
42
mW
)
* Power Dissipation
1
P
P
P
0
.
21
mW
H
L
2
2
* Power-Delay Product
* Also, the noise margin for the high
state will be degraded.
DP
P
t
0
.
21
mW
(
170
n
sec)
36
pJ
PD
ECES 352 Winter 2007
Ch 10 MOS Digital
25
A Modified NMOS Inverter
* Increase the load’s W/L ratio from
1/10 to 2 (20 X increase in K2 ).
1 W
A
A
6
4
K
C
6
.
2
x
10
1
.
2
x
10
2
n
ox
2
2
2 L
V
V
2
2
i
K
v
V
K
V
v
V
D
2
2
GS
2
Th
2
2
DD
o
Th
2
* New tPLH is 17 nsec vs previous 340 nsec.
* New tPHL is 9 nsec vs previous 0.8 nsec.
* New tPD is 13 nsec vs previous 170 nsec.
* Noise Margins
NML = 0.95 V, NMH = 2.27 V
* Propagation Delay
* New power dissipation is 8.4 mW vs
previous 0.21 mW.
* New power-delay product is 109 pJ vs
1
1
previous 36 pJ.
t
t
t
0
.
8
n
sec
340
n
sec
170
n
sec
PD
PHL
PLH
2
2
* New VIH = 3.64 V vs previous 1.53 V.
* New noise margin for the high state
(
0
0
.
42
mW
)
1
P
P
P
0
.
21
mW
H
L
NMH = 3.8 - 3.64 = 0.16 V vs
2
2
previous 2.27 V.
* Power-Delay Product
* Power Dissipation
DP
P
t
0
.
21
mW
(
170
n
sec)
36
pJ
PD
ECES 352 Winter 2007
Ch 10 MOS Digital
26