3. Two-Terminal Pull up MOSFET Inverters: General Characteristics ECE 108 1 Esener ECE108
Dec 20, 2015
3. Two-Terminal Pull up MOSFET Inverters: General Characteristics
ECE 108
1 Esener ECE108
A fluidic inverter
Esener ECE108 2
3. 1. Inverter Currents and Voltages
3 Esener ECE108
S
V DD
LOAD or PULL-UP
D
V G
+
V in = V GS
DRIVER or PULL-DOWN
V 0 = V DS
+
VL=VDD-VO
ID
Device Equations
IL = fL(VL)
ID = fD(VGS; VDS; VT )
Circuit Equations
VL=VDD-VO
ID = IL
fL(VDD-VO) = fD (VIN; VO; VT )
VIN =VGS VO =VDS
=> Transfer characteristics
S
V DD
KL
D
V G
+
V in = V GS
V 0 = V DS
+
IL=KL(VDD-VO) IL
3.2. INVERTER THRESHOLD VOLTAGE
Esener ECE108 4
T T T
T
2
inv inv
inv DD T
2 2V V V V V V 0
B B
1 V V [1 1 2 V V ]
B
DD
B
The inverter logic threshold voltage is the input voltage that results in an equal
output voltage (Vinv =VGS = VDS = VO) (Text book uses VM for Vinv)
Using the constraint, Vinv=VGS = VDS
2
inv
L
DinvDD
invDS
2
invD
DSDDLT
T
VV2K
kVV
VV
VV2
k)V(VK
VIN > Vinv Output Low
VIN < Vinv Output High
V B Note BK
kLet 1-
L
D
3.3. COMPUTING VOH & VOL
Esener ECE108 5
0II LD
Neglecting the leakage current in the pull down MOSFET (off)
IL=KL(VDD-VO) } =>
=> VOH = VDD
VIN=VGS=0 => VGS < VT => Pull down off ID = 0
VOH for VIN=0
VOL for VIN = VDD
TDD
DDOL
2
OL
DDOLTDD
2
OL
2
OLOLDDDOLDDL
VVB1
VV
small very is V sinceand
0B
2VVVV
B
12V
2
VVVVk)V(VK T
VGS > VGS*=> Pull down is Linear
with B.decreases
V seen, can be As OL
fL(VDD-VO) = fD (VIN; VO; VT )
KL(VDD-VO) = 0
3.4. DECIDING ON DRIVER STATE
Esener ECE108 6
V KI assumingand
VV VV
LLL
in
GSDSo
2**
*
2*
22
)(T
T
VVK
kVVV
VVV
VVk
VVKGS
L
DDDTGS
TGSDS
GSD
DSDDL
2* *
-1
*
22 0.........
Note B V
11 2 1
T T
T
L LGS GS DD
D D
D
L
GS DD
K KV V V V V
k k
kLet B
K
V V BVB
Saturation - Linear Transition (Vin=VGS*)
VGS* is the input voltage that biases the driver at the edge of saturation and linear
modes of operation. We need to solve for
Rearranging
fL(VDD-VO) = fD (VIN; VO; VT )
With the constraint T
*
GSDS VVV
Defining
VIN > VGS* pull down Linear
VIN < VGS* pull down Saturated
ESTIMATING
VOLTAGE TRANSFER CHARACTERISTICS (VTC)
Esener ECE108 7
linear.or saturationin down PullVV if
V = V0Ioffdown PullVV if
V and to V toV Compare
TGS
DDDSDTGS
T
*
GSGS
TGSDS
2
TIND
ODDL
*
GSGS
VVV ifcheck sure, makeTo
)V(V2
k )V-(Vf Compute
saturation VV
*
GS GS
DS GS T
V V linear
To make sure and to pick the right result,
check if V <V -V
A. Determine VT and VGS*
]2
V-)VV[(V k)V-(Vf Compute
2
OOTINDODDL
B.
C.
D.
fL(VDD-VO) = 0
3.5. POWER DISSIPATION
Esener ECE108 8
When VGS VT , the driver is "on".
DD D DD L L DD O
2
DD L DD O DD L
P V I =V I ; I G (V -V )
=> P V G (V -V ) ~ V G
L
•
Power dissipation includes the power dissipated in the stable states (Static Power Dissipation)
plus the power dissipated during the transition (Dynamic Power Dissipation).
For MOS inverter with 2-terminal loads, SPD >> DPD
For MOS inverters driving a capacitive load SPD (“Low”) >> SPD (“High”)
Calculating SPD (Low)
For a two-terminal load MOS inverter the power is mostly dissipated at the load
3.6. VTC FEATURES:
3.6.1 Dynamic Range & Slope of VTC
Esener ECE108 9
TDD
OL
OH
TDD
DDOLDDOH
VVB1V
V = Range Dynamic
VVB1
VVand VV Using
As can be seen, the dynamic range increases with B.
Dynamic Range
Dynamic range is the ratio of the output high to outpul low voltage (or current).
It is a measure of the separation of logic levels and can affect the noise performance
Slope of VTC
The slope determines the width of the transition region
DD
2
inO
2
inD
ODDL VVV2
BV VV
2
k)V(VK TT
OIN T
IN
dVslope of VTC B(V -V )
dV
with B.increases slope theseen, can be As
Assuming pull down saturated
Slope= -B(Vin-VT)
3.6.2 NOISE MARGINS
Noise margin is the maximum noise voltage added to the input signal that does not cause an undesirable change at the output.
• Low input Vin < V*IL
• Transition V*IL< Vin < V*IH
• High input Vin > V*IH
NML = V*
IL -V*OL NMH = V*
OH -V*IH
dV o
dV i = - 1
dV o
dV i = - 1
V iL * V iH
* V OH
* V OL *
V OH *
V OL *
V DD
V DD
VOL= VIL
VIH
Both input and output axis have same units
VOH*
VOL*
NMH NML
VOL* VIL* VOH* VIH*
VOL= VIL
VIH
NOISE MARGINS - LOGIC REGENERATION
Assumptions: •Both inverters have
identical VTC •Both input and output
axis have same units
VOH*
VOL*
NMH NML
VOL* VIL* VOH* VIH*
VOL= VIL
VIH
LOGIC REGENERATION
VOH*
VOL*
NMH NML
VOL* VIL* VOH* VIH*
VOH= VIH
VIL
LOGIC REGENERATION
VOH*
VOL*
NMH NML
VOL* VIL* VOH* VIH*
VOL= VIL
VIH
LOGIC REGENERATION
VOH*
VOL*
VOH*
VOL*
NMH NML
IMPACT OF VTC ON NOISE MARGINS
VOH*
VOL*
VOH*
VOL*
NMH=0 NML
IMPACT OF VTC ON NOISE MARGINS
VOH*
VOL*
VOH*
VOL*
NMH=0 NML
IMPACT OF VTC ON NOISE MARGINS
VOH*
VOL*
VOH*
VOL*
NMH< 0 NML
IMPACT OF VTC ON NOISE MARGINS
NOISE MARGINS
Esener ECE108 19
Finding VIL* (Assume pull down Saturated)
O D DD
L
1V I V
K
* * **
IL T2
*DD IL T
11
B V V 1k
I V V2
O O D D
LIL D IL IL
dV dV dI dI
dV dI dV K dV
*
IL T T
* 2DOH T T DD
L
*
OH DD
V V +1/B V
kV (V +1/B-V ) V
2K
1V V
2B
Then,
v in
VO
VOH
VOL
VDD VT
*
OHV
*
ILV
VTC FEATURES: NOISE MARGINS
Esener ECE108 20
3.5.3. Finding VIH* (Assume pull down Linear)
VVVk=V
I ;Vk
V
I
2
VVVVkI
1
OTin
O
DOD
in
D
2
OOTinDD
)V
I)/(
V
I(
dV
dV
O
D
in
D
in
O
D
L
DDO
2
O23
DTOin
OTin
O
IK
1V=Vand
kVIand V2VV
1VVV
V
Theorem: If f(x,y) is continuous in a region and its partial derivatives with respect to x and y are also
continuous in this region, and y can be expressed in function of x then
dy/dx = -(df/dx)/( df/dy); Using y= VO, x = Vin and f = ID; we obtain,
=> V*OL = VDD - B (3/2 V*
OL2)
TDD
*
IH
DD
*
OL
V16BV13B
2V
16BV13B
1V
=>
]2
V-)VV[(V kI
2
OOTINDD
v in
VO
VOH
VOL
VDD VT
*
OLV
*
IHV
VTC FEATURES: NOISE MARGINS
Esener ECE108 21
Using noise margin definitions let’s compute
NMH VOH
* VIH
* and NML VIL
* VOL
*
*
IL T T
*
OH DD
V V +1/B V
1V V
2
B
TDD
*
IH
DD
*
OL
V16BV13B
2V
16BV13B
1V
H DD DD T
L T DD
1 2NM V 1 6BV 1 V
2B 3B
1NM V +1/B- 1 6BV 1
3B
v in
VO
VOH
VOL
VDD
*
OHV
*
IHV*
OHV
*
OLV
*
ILV*
OLV
NML NMH
Non Ideal Transfer Characteristics of a real inverter
NM H
V in (V)
NM L
V M
0.0
1.0
2.0
3.0
4.0
5.0
1.0 2.0 3.0 4.0 5.0
V out (V)
VO
L = VIL
VIH
Non Ideal speed response
Typically switching does not occur instantaneously • Rise and fall times are finite • Leading to a significant inverter
propagation delay
Ideal speed response with Negligible rise and fall time And no propagation delay
Bit Error Rate
• BER is closely related to the distortion introduced by the inverter including finite rise and fall times, overshoots and signal jitter. One way to measure BER is through eye diagrams that are obtained by integrating the output of an inverter that is fed by a random sequence of bits that is sufficiently long. By looking at the closing of the eye one can approximately determine the BER.
• A direct measurement can also be performed by counting the errors at the output of the inverter. This can take a very very long time and is generally only practiced in communication circuits where the BER is around 10-9 and bit rates exceed 1Gb/s
Vin
Vout
t
10%
90%
Vdd
0V
t
m1
m0
0
1
BER = Qm1 - m0
+1 0
Speed Response -Bit Error Rate - Eye Diagram