Jaeger/Blalock 9/25/03 Microelectronic Circuit D esign McGraw-Hill Chap 6 - 1 Chapter 6 Introduction to Digital Electronics Microelectronic Circuit Design Richard C. Jaeger Travis N. Blalock
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 1
Chapter 6Introduction to Digital Electronics
Microelectronic Circuit DesignRichard C. JaegerTravis N. Blalock
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 2
Ideal Logic Gates
• Binary logic gates are the most common style of digital logic
• The output will consist of either a 0 (low) or a 1 (high)
• The most basic digital building block is the inverter
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 3
The Ideal Inverter
The ideal inverter has the following voltage transfer characteristic (VTC) and is described by the following symbol
V+ and V- are the supply rails, and VH and VL describe the high and low logic levels at the output
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 4
Logic Level Definitions
An inverter operating with power supplies at V+ and 0 V can be implemented using a switch with a resistive load
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 5
Logic Voltage Level Definitions
• VL – The nominal voltage corresponding to a low-logic state at the input of a logic gate for vi = VH
• VH – The nominal voltage corresponding to a high-logic state at the output of a logic gate for vi = VL
• VIL – The maximum input voltage that will be recognized as a low input logic level• VIH – The maximum input voltage that will be recognized as a high input logic level• VOH – The output voltage corresponding to an input voltage of VIL
• VOL – The output voltage corresponding to an input voltage of VIH
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 6
Logic Voltage Level Definitions (cont.)
Note that for the VTC of the nonideal inverter, there is now an undefined logic state
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 7
Noise Margins
• Noise margins represent “safety margins” that prevent the circuit from producing erroneous outputs in the presence of noisy inputs
• Noise margins are defined for low and high input levels using the following equations:
NML = VIL – VOL
NMH = VOH – VIH
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 8
Noise Margins (cont.)
• Graphical representation of where noise margins are defined
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 10
NMOS Logic Design
• MOS transistors (both PMOS and NMOS) can be combined with resistive loads to create single channel logic gates
• The circuit designer is limited to altering circuit topology and width-to-length, or W/L, ratio since the other factors are dependent upon processing parameters
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 11
NMOS Inverter with a Resistive Load
• The resistor R is used to “pull” the output high
• MS is the switching transistor
• The size of R and the W/L ratio of MS are the design factors that need to be chosen
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 12
Load Line Visualization
• The following illustrates the operation of the NMOS output (vDS) characteristics where the following equation describes the load line
RiVv DDDDS
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 13
NMOS with Resistive Load Design Example
• Design a NMOS resistive load inverter for – VDD = 3.3 V
– P = 0.1 mW when VL = 0.2 V (chosen to be between 25% and 50% of VTN to ensure that Ms is in cutoff when the input is low and an adequate noise margin)
– Kn = 60 μA/V2
– VTN = 0.75 V
• Find the value of the load resistor R and the W/L ratio of the switching transistor MS
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 14
Example continued
• First the value of the current through the resistor must be determined by using the following:
• The value of the resistor can now be found by the following which assumes that the transistor is on or the output is low:
AV
mWV
PIDD
DD 3.303.3
1.0
kA
VVI
VVRDD
LDD 1023.30
2.03.3
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 15
Example Continued
• For vI = VL = 0.2V, the transistor’s vGS will be less than the threshold voltage, therefore it will be operating in the triode region. Using the linear equation for a MOSFET, the W/L ratio can be found:
11
103.1
2.022.075.03.310603.30
2
6
'
S
S
LL
TNHS
nD
LW
LWA
VVVVL
WKI
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 16
On-Resistance of MS
• The NMOS resistive load inverter can be thought of as a resistive divider when the output is low, described by the following expression:
RRRVV
on
onDDL
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 17
On-Resistance of MS (cont.)
2
1' DS
TNGSnD
DSon vVv
LWKi
vR
When the NMOS resistive load inverter’s output is low, the On-Resistance of the NMOS can be calculated with the following expression:
Note that Ron should be kept small compared to R to ensure that VL remains low. Also, notice that its value is nonlinear since it depends on vDS
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 18
Noise Margin Analysis
• The following equations can be used to determine the various parameters needed to determine the noise margin of NMOS resistive load inverters
RKVV
RKV
RKVV
RKVV
RKVV
n
DDOL
n
DD
nTNIH
nDDOH
nTNIL
32
63.11
21
1
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 19
Noise Margin Analysis
To find VIL & VOH
• Assume saturation region operation (VDS = vO=VOH > vGS-VTN= VIL-VTN)• Substitute iD in load-line expression• Set dvO/dvI = 0 and solve for vI=vIL
To find VIH and VOL
• Assume triode –region operation• Repeat above steps
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 20
Load Resistor Problems
• For completely integrated circuits, R must be implemented on chip using the shown structure
• Using the given equation, it can be seen that resistors take up a large area of silicon as in an example 95kΩ resistor
1
9500001.0
10195 4
cmcmkRt
WL
tWLR
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 21
Practice
• P6.46: Design a resistive load inverter to operate from a 2.5V supply with a power dissipation of 50 µW. Assume VTT=0.60V.
• P6.47: Find VH, VL and the power dissipation for vi=VL for the inverters shown.
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 22
Reference Design
Use this design as the reference inverter for complex logic gate problems.
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 23
NOR Gates
Simplified switch model for the NOR gate with A on
Two-input NOR gate
For resistive-load, replace load transistor ML with load resistance
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 24
NAND Gates
Simplified switch model for the NOR gate with A and B on (right)
Two-input NAND gate (left)
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 25
NAND Gate Device Size Selection
• The NAND switching transistors can be sized based on the reference inverter
• To keep the low voltage level to be comparable to the inverter, the desired RON of MA and MB must be 0.5RON of MS,Inverter
• This can be accomplished by approximately doubling the (W/L)A and (W/L)B
• The sizes can also be chosen by using the design value of VL and using the following equation:
DSTNGSS
nDSDSTNGSS
nD vVvL
WKvvVvL
WKi
'' 5.0
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 26
NAND Gate Device Size Selection (continued)
• Two sources of error that arise are the facts that VSB and VGS of the two transistors do not equal. These factors should be considered for proper gate design
• The technique used to calculate the size of the load transistor for the depletion-mode load inverter is the exact same as for this NAND gate
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 27
Complex NMOS Logic Design
An advantage of NMOS technology is that it is simple to design complex logic functions based on the NOR and NAND gates
The circuit in the figure has the logic function:
Y = A + BC + BD
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 28
Complex Logic Gate Transistor Sizing
• There are two ways to find the W/L ratios of the switching transistors
1) Using the worst case (longest) path and choosing the W/L ratio such that the RON of the multiple legs match similar to the technique used to find the W/L ratios in the NAND Gate
2) Partitioning the circuit into series sub-networks, and make the equivalent on-resistances equal
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 29
Complex Logic Gate Transistor Sizing
The figure on the left shows the worst case technique to find the sizes where (W/L)S=2.06 is the reference inverter ratio for this technology and the longest path is 3 transistors are in series
The figure on the right shows the partitioning technique to find the sizes which gives two 4.12/1 ratios in series which is 2(2.06/1)
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 30
Example 6.10
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 31
Prob. 6.96
(a) What is the logic function?(b) Find W/L ratios if design is based on fig. 6.32(d)
(W/L)L=1/2.15 (W/L)S=2.06/1
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 32
Prob. 6.98
a) Find logic functionb) Find (W/L)’s if the gate is to dissipate 3 times the power of reference design in fig. 6.32(d)
(W/L)L=1/2.15 (W/L)S=2.06/1
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 33
Recommended Practice Problems Secs 6.11-6.14
• 92, 93, 95, 97, 98, 99,102,103,110,116,119,122,123,124,125,128, 130,136,137
• Solutions can be found at: ~mtoledo/4207/Pub
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 34
Other configurations: Diode Logic
• Diodes can with resistive loads to implement simple logic gates
Diode OR gate Diode AND gate
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 35
Diode Transistor Logic
• Since diode gates are limited to AND and OR functions, the diodes can be combined with transistors to complete the basic logic functions such as the following NAND gate
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 36
Using Transistors in Place of a Resistor
• NMOS load w/ a) gate connected
to the source b) gate connected
to ground c) gate connected
to VDD
d) gate biased to linear region e) a depletion mode NMOSNote that a) and b) are
not useful
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 37
Static Design of the NMOS Saturated Load Inverter
Schematic for a NMOSsaturated load inverter
Cross-section for a NMOSsaturated load inverter
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 38
NMOS Inverter with a Depletion-mode Load
• With the addition of a depletion-mode NMOS (VTH < 0V), it is possible to configure an inverter as shown
• VGSL = 0 V for this configuration meaning that ML is always operating in saturation
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 39
Layout of the NMOS Depletion-Mode NOR and NAND Gates
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 40
NMOS Inverter Summary
• Resistive load inverter takes up too much area for and IC design
• The saturated load configuration is the simplest design, but VH never reaches VDD and has a slow switching speed
• The linear load inverter fixes the speed and logic level issues, but it requires an additional power supply for the load gate
• The depletion-mode NMOS load requires the most processing steps, but needs the smallest area to achieve the highest speed, VH = VDD, and best combination of noise margins
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 41
Typical Inverter Characteristic
Inverter w/ Resistor Load
Saturated Load Inverter
Linear Load Inverter
Inverter w/ Depletion-Mode Load
VH 5.0 V 3.4 V 5.0 V 5.0 V
VL 0.25 V 0.25 V 0.25 V 0.25 V
NML 0.34 V 0.32 V 0.02 V 0.69 V
NMH 1.43 V 0.69 V 2.78 V 2.25 V
Area (μm2)
9500 6.92 9.36 4.21
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 42
Dynamic Response of Logic Gates
• An important figure of merit to describe logic gates is their response in the time domain
• The rise and fall times, tf and tr, are measured at the 10% and 90% points on the transitions between the two states as shown by the following expressions:
V10% = VL + 0.1ΔV
V90% = VL + 0.9ΔV = VH – 0.1ΔV
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 43
Propagation Delay
• Propagation delay describes the amount of time between a change at the 50% point input to cause a change at the 50% point of the output described by the following:
• The high-to-low prop delay, τPHL, and the low-to-high prop delay, τPLH, are usually not equal, but can be described as an average value:
2PLHPHL
P
2LH
50%VVV
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 44
Dynamic Response of Logic Gates
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 45
Power Delay Product
• The power-delay product (PDP) is use as a metric to describe the amount of energy required to perform a basic logic operation and is given by the following equation when P is the average power dissipated be the logic gate:
PPPDP
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 46
Static Power Dissipation
• Static Power Dissipation is the average power dissipation of a logic gate when the output is in both the high and low states
• IDDH = current in the circuit for vO = VH
• IDDL = current in the circuit for vO = VL
• Since IDDH = 0 A for vO = VH:
2DDLDDDDHDD
avIVIVP
2DDLDD
avIVP
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 47
Dynamic Power Dissipation
• Dynamic Power Dissipation is the power dissipated during the process of charging and discharging the load capacitance connected to the logic gate
DischargingCharging
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 48
Dynamic Power Dissipation• The energy delivered to the capacitor can be found by:
• Energy stored by the capacitor with C=Q/V is:
• The energy lost in the charging and discharging the C through resistive elements is thus given by:
2)(
)0(0
)()( DD
V
VCDDDDD CVdvtiVCdttiVE
C
C
2
2
21
2
00
DDQQ
SCV
CQdq
CqVdqE
2
2DD
SDLCVEEE
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 49
Dynamic Power Dissipation
• The total energy lost in the charging and discharging of the capacitor through resistive elements is given by:
• Thus it can be seen that for every cycle (frequency) that the gate is changed, the dynamic power dissipation is given by:
222
22 DDDDDD
TD CVCVCVE
fCVP DDD2
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 50
Power Scaling in MOS Logic
• By reducing the W/L of the load and switching transistors of an inverter, it is possible to reduce the static power dissipation by the same factor without sacrificing VH and VL. This same concept works for increasing the power which will increase the dynamic response. For resistive load, the switching transistor W/L and the load resistor should be scaled up or down by the same factor.
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 51
Power Scaling in MOS Logic
a) Original Saturated Load Inverterb) Saturated Load inverter designed to operate at 1/3 the powerc) Original Depletion-Mode Inverterd) Depletion-mode inverter designed to operate at twice the power
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 52
Dynamic BehaviorCapacitance in MOS Logic Circuits
• The MOS device has the capacitances CSB, CGS, CDB, and CGD that need to be considered for dynamic response analysis, but depending on the configuration, some of them will be shorted out as seen in the first figure
• The capacitance seen at a node can be lumped together as seen in the second figure
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 53
Fan-out Limitations
• Static design constraints are not usually important for MOS logic circuits since they normally drive capacitive loads (i.e. the gate of a MOS)
• As the number of gates the output (fan-out) of a logic device has to drive, the load capacitance increases, and the time response decreases
• This notion implies that the fan-out that a logic circuit can drive will be limited to time delay tolerances of the circuit
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 54
Dynamic Response of the NMOS Inverter with a Resistive Load
• The rise and fall times and propagation delays are given by the relationships:
where R and C are the resistance and capacitance seen at the output
RC
RCtt
PHLPLH
fr
69.0
2.2
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 55
Dynamic Response of the NMOS Inverter with a Resistive Load
• There are four important times that need to be considered when characterizing the dynamic response of a logic circuit which are denote t1 – t4 in the figure
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 56
Dynamic Response of the NMOS Inverter with a Resistive Load
• It is also possible to calculate τPHL and tf by a piecewise analysis technique, and are given by the following equations:
TNSH
TNS
H
TNSHonSf
TNSH
TNS
LH
TNSHonSPHL
VVVV
VVVVVCRttt
VVV
VVVVCRtttt
1.0229.0
9.02ln
214ln)(
14
2233
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 57
Comparison of Load Devices
• The current has been normalized to 50 μA for vo=VOL=0.25 V is the figure for the various types of inverters
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 58
Comparison of Load Devices
• Body effect degrades the performance of the load device
• The saturated load devices have the poorest τPLH since they have the lowest load current delivery
• The saturated load devices also reach zero current before the output reaches 5 V
• The linear load device is faster than the saturated load device, but still slower than the resistive load inverter.
• The fastest τPLH is from the depletion-mode device
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 59
Comparison of Load Devices
Simulated fall times for a 0.1 pF load
Simulated rise times for a 0.1 pF load
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 60
PMOS Logic
• PMOS logic circuits predated NMOS logic circuit, but were replaced since they are usually operate at slower speeds (note the change in the power supplies and that VL = -5V, VH=-0.25 for (a,c,d), VL = -4V, VH=-0.25 for b)
Resistive Load Saturated Load Linear Load Depletion-mode Load
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 61
PMOS
• For , transistor is off.• For more negative vGS, drain
current increases in magnitude. • PMOS is in triode region for small
(neg) values of VDS and in saturation for larger (neg) values.
TPVGSV
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 62
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 63
PMOS NAND and NOR Gates
NOR Gate NAND Gate
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 64
Silicon Art
• In the earlier days of IC design, chip designers were allowed to artistically express themselves on the wafer by creating images with various processing steps
• However, today’s modern foundries have stopped this since the graphics did not pass the design rules and were causing fabrication problems
Jaeger/Blalock9/25/03
Microelectronic Circuit DesignMcGraw-Hill
Chap 6 - 65
Silicon Art Examples