Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Cmos Issues

Download as ppt, pdf, or txt
Download as ppt, pdf, or txt
You are on page 1of 180

CMOS VLSI

Integrated Circuits

An IC is a combination of interconnected circuit elements inseparably associated on or within a continuous substrate. Why Design Integrated circuits ?

IC Circuits are much smaller. IC Circuits consume less Power. Easier to design and manufacture. They are more reliable.

Integrated Circuits

Some terminology associated with IC are Substrate : It is supporting material upon or within which an IC is fabricated Monolithic IC : An IC whose elements are formed in place upon or within a semiconductor with at least on element formed within the substrate. Hybrid IC : Consist of two or more IC types Wafer : It is the basic physical unit used in processing. Typically the wafer is circular Chip : one of the repeated ICs on a wafer.

Integrated Circuits

Advantage of IC circuits over discrete components are. Size : Smaller both wires and transistor are shrunk to micro as compared to millimeter or centimeter of discrete components. : Hundred times faster than PCB. : Less power due to small size of circuits on chip.

Speed Power

History of IC Technology

The IC technology can be categorized ,depending on the number of transistors, as :


IC No.of active Devices (BJT/FET) 1-100 FUNCTION YEAR

SSI

Gates, Op-amps, Linear app Registers, Filters

1960

MSI

100-1000

1965

LSI VLSI

1000-10,000 10,000-100,000

Microprocessors, ADC Memory, Computers, Signal Processors

1970 1975

History of IC Technology
Microprocessor 80286 80386 80486 Pentium Date of introduction 2 / 82 10 / 85 4 / 89 3 / 93 # transistors 134,000 275,000 1,200,000 3,100,000 Feature size ( microns ) 1.5 1.5 1.0

0.8 0.6

Pentium Pro

11 / 95

5,500,000

Pentium lll 0.18

Pentium IV 0.13

Feature Size

Typically L=1 to 10 micro m, W=2 to 500 micro m and the thickness of the oxide layer is in the range 0.02 to 0.1 micro m.

Feature Size

Value of the gate length is called as the feature size of manufacturing technology. Thus feature size is function of IC technology.

N-MOSFET layout

Microelectronics
Two basic technologies used for manufacturing ICs are Bipolar MOS

Bipolar : The main technology is low-power-schottky TTL Disadvantage: High power dissipation. Used for SSI and MSI.

MOS : LSI technology uses MOSFETS since these can be packed in small area. P-MOS Technology : It uses P-MOSFET s. Mobility of holes - 240 cm2/v*sec Holes are majority carriers & hence this is relatively slow. N-MOS Technology : It uses N-MOSFET s Mobility of electrons -650 cm2 /v*sec Since electrons are majority carriers this technology is faster than P-MOS. CMOS Technology : It uses combination of P-channel and N-channel MOS.

CMOS Technology
CMOS Technology : Simultaneously provides both NMOS and PMOS. Characteristics of CMOS technology lies between that of P-MOS and N-MOS.

It is faster than P-MOS but slower than N-MOS.


Provides more flexibility It is mainly used in systems that require portability or less static power consumption. It needs more processing steps as compared to N-MOS or P-MOS.

CMOS Technology

CMOS Technology
Wide supply voltage range from 3 to 15 volt. Voltage required to switch gate is fixed percentage of VDD. Packaging density : Requires 2n devices for n input gate Fully restored logic level : Output settles at VDD or Vss. Therefore called as restoring logic. Characteristics of CMOS technology lies between that of PMOS and N-MOS. It is faster than P-MOS but slower than NMOS. Mainly used in systems that require portability or less power consumption. It needs more processing steps as compared to N-MOS or PMOS.

CMOS vs BIPOLAR
FACTORS

CMOS

BIPOLAR

Static power dissipation


Input impedance Noise margin Packaging density Fan-out Direction

Low
High High High Low Bi-directional devices

High
Low Low Low High Unidirectional devices

Trends in VLSI

Minimum geometrical feature size.

Increase in speed of digital circuit. Use of GaAs for very high frequency.
Increase in complexity of a circuit function and device count.

Increasing designer productivity and evergrowing dependence on the computer in the design process.
Continual shift of where design, production, and markets are geographically located. Growing coupling of specific process and its processing equipment.

MOS transistors Types and Symbols


D D

G S

NMOS Enhancement NMOS Depletion D D


G

G S

PMOS Enhancement

NMOS with Bulk Contact

MOS transistors
Top View
Drain L W Source Gate Moat

MOS transistors
n-Channel D Id Vdg + + Vds Ig G + + Vgs Vbb Is p-Channel D Id Vdg + + Vds Ig G + + Vgs Vbb Is

Metal Oxide Semiconductor Field Effect Transistor


Source Gate Gate oxide Drain

Length substrate

MOS stands for the Metal Oxide Semiconductor Consist of four terminals Gate, Source, Drain and substrate

MOS transistors
Some common terms used in the MOS Transistor

Vgs Vds Id

: The gate-tosource Voltage. : The drain-tosource Voltage. (Vds = - Vsd). : The current flowing between the drain to

source.

The constants that determine the magnitude are Vt : The transistor threshold voltage,which is positive for n-type and negative for p-type. K : The transistor transcoductance, which is positive for both types. W/L : The width-to-length ratio of the transistor.

Metal Oxide Semiconductor Field Effect Transistor

For P-channel MOS substrate is of N type and source, drain are formed with P type material. For N-channel MOS substrate is of P type while source & drain are formed with N type material. Gate is polycrystalline silicon electrode and is insulated from substrate by thin layer of silicon dioxide SiO2 Since the gate is insulated, MOSFETs are also called as Insulated Gate Field Effect Transistors ( IGFET ) It is a voltage controlled device, the current through channel is controlled by voltage applied to gate. MOSFETs can be configured either as

Enhancement type MOSFET OR


Depletion type MOSFET

Depletion type MOS

In Depletion MOS structure, the source & drain are diffused on P- substrate as shown above. Positive voltages enhances number of electrons from source to drain. Negative voltage applied to gate reduces the drain current This is called as normally ON MOS.

Enhancement NMOS

It consists of lightly doped p-type substrate into which two highly doped n+ regions are diffused. When the gate voltage VGS = 0 volt, there is no conduction path between source and drain, hence the drain current is zero. Since the gate is insulated positive voltage may be applied to gate, it will produce electric field across substrate.

Enhancement NMOS
This field will end on induced negative charges in p substrate. These negatively charged electrons, which are minority carriers in p substrate, form an inversion layer. Current flows from source to drain through this induced channel. More the positive voltage, More is the induced charge & hence more current flows from source to drain. This is also called Normally Off MOS, since drain current is zero for zero gate voltage. CMOS integrated circuits use enhancement type transistors only.

Enhancement MOS Cross-section


As VDS is increased, the induced Channel acquires a tapered shape and its resistance increases with Increase in VDS. Here VGS is kept constant at value > VT

Enhancement NMOS-Operation
Gate oxide Drain

Source

Gate

n+
holes

channel

n+
electrons

P-substrate

substrate

Enhancement NMOS-Operation Cut-off Region


Cutoff region (Vgs = 0 and Vds = 0 ) Gate oxide Drain

Source

Gate

n+
holes P-substrate

n+
electrons
No channel

substrate
no current flows between Source and Drain Current-voltage relation : IDS = 0 VGS < VT

Enhancement NMOS-Operation Cut-off Region


Depletion Region (0< Vgs < Vt and Vds = 0 ) Gate oxide Drain Source Gate

n+
holes P-substrate

n+
electrons
Depletion Region

substrate

When positive gate voltage is applied,electron will start to deplete the substrate near the surface under gate. This deplete the p-type substrate in this region and form the depletion region

Enhancement NMOS-Operation Linear Region

Inversion region (Vgs > Vt and Vds = 0 ) Gate oxide Drain Source Gate

n+
holes
P-substrate

n+
electrons
Inversion region

n region

substrate

Enhancement NMOS-Operation Linear Region


Inversion region As gate voltage is increased number of electrons will be attracted to the substrate surface under the gate to make n-type. The n-type region which is created electrically in the p substrate is called the inversion layer. The gate-to-source voltage necessary to create the inversion layer is called the threshold voltage (Vt).

After formation of inversion layer the current flow from drain to source or source to drain if small voltage is applied

Enhancement NMOS-Operation Linear Region

Increase in gate-source voltage(Vgs) beyond threshold voltage (Vt)brings additional electrons under gate causing increase in inversion layer. If large drain-source voltage is applied ,this voltage itself will tend to deplete the inversion layer due to potential drop across the region formed by current flow

Enhancement NMOS-Operation Linear Region

(Vgs > Vt, Vds < Vgs- Vt)

Source

Gate

Gate oxide

Drain

n+
holes
P-substrate

n+
electrons

n region

substrate

Enhancement NMOS-Operation Saturated Region


Vgs > Vt and Vds > Vgs - Vt

Source

Gate

Gate oxide

Drain

n+
holes P-substrate
n region

n+
electrons

substrate

When VDS > VGS VT, VGD < VT, the channel becomes pinched- off & transistor is said to be in saturation. Conduction is brought by drift mechanism of electrons under the influence of positive drain voltage and effective channel length is modulated.

Enhancement NMOS-Operation

some equation Ig = 0 0 Vgs < Vt (cutoff), Vds > 0

Id =

( K W (Vgs Vt- Vds/2) Vds) / L when[Vgs > Vt, 0 < Vds <Vgs Vt(Linear) ] ( K W (Vgs Vt)**2 (1 + Vds)) / 2 L when[Vgs > Vt, Vds > Vgs Vt(saturation) channel length modulation

Enhancement NMOS-Operation

Inverter

Symbol: In Out logic function Out = In

Truth Table of an Inverter:


In 0 1 Out 1 0

STATIC LOAD INVERTERS


Static load inverter are Rationed logic gates where logic levels are determined by relative dimensions of composing transistors. Transfer function of inverter varies with load.

When I = 1, inverter dissipates static power Switching point of inverter depends on ratio of R to RON (on resistance of NMOS device.)

STATIC LOAD INVERTERS

Problems with static load inverters


On chip resistors are large Static power consumption VOL = 0 Large tpLH

ACTIVE-RESISTIVE LOAD
Load device is always on, looks like a load resistor. Dissipates static power when I = 1, VGS = 0V always VOH = 5V; VOL nearly 0V, depending on ratio of RON_dep to RON_enh. Depletion-mode devices were used before it was economical to put both p-type and n-type devices on the same die. Problems with this Extra process step Static power consumption VOL = 0 Large tpLH

The CMOS Inverter


# No static power consumption (almost) # VOH = VDD; VOL = 0 # tpLH = tpLH If properly designed # Low Impedance connection to ground and VDD

- More fab. stages - Lower hole mobility

The CMOS Inverter


Complementary i.e. output have always a low impedance connection to GND or VDD VOH = VDD VOL = 0 VM = f(Ron-n,Ron-p) VM = VDD/2 if Ron-n = Ron-p

The CMOS Inverter

Rin=infinite ,Rout=0 Noise Margin = VDD/2, Gain = infinite

Actual characteristics
NML = VIL - VOL = 0.75 - 0.50 = 0.25V NMH = VOH - VIH = 3.50 - 2.25 = 1.25V VM = 1.75V

The CMOS InverterVoltage Operation


REGION OF OPERATION CUT-OFF NONSATURATED P-DEVICE VGS >VT VIN > VT + VDD VGS< VT , VIN <VT+ VDD VDS > VGS - VT N-DEVICE VGS < VT VIN < VT VGS > VT , VIN > VT VDS < VGS - VT

VOUT > VIN - VT


VGS< VT , VIN <VT+ VDD VDS < VGS - VT VOUT < VIN - VT

VOUT < VIN - VT


VGS > VT , VIN > VT VDS > VGS - VT VOUT > VIN - VT

SATURATED

Inverter Load Characteristics

Vinp

Vinn

Vds(Vdd - Vdsp)

Taking the absolute value of the p device , Vds , and superimposing yields the following curve. The input/output transfer curve may be determined by the points of common Vgs intersection

CMOS Inverter Voltage Transfer Characteristic

VTC graphically
extracted from the load lines

High noise margin

NMH=VOH-VIH ~ 5-3 = 2V
NML =VIL-VOL ~ 2-0 = 2V

Region: Linear - Saturation


A B

Region: Linear - Saturation


REGION CONDITIO N N-DEVICE A 0 < Vin < VT[n] Cut-off since Vin < VT[n] B VT[n] < Vin <VDD/2 Saturated as (VDS[n] = VOUT) > VDD/2 & (VGS[n] = Vin) < VDD/2

P-DEVICE

Non-saturated (linear) since VSD[p] < VSG[p] |VT[p]|


VOUT = VDD as VDS[P] = 0 As VDS[P] = VOUT VDD

Non-saturated (linear-region) since VSD[p] < VSG[p] - |VT[p]|

OUTPUT

Region: Linear - Saturation


REGION CONDITIO N

C Vin = VDD / 2

D VDD/2 < Vin < VDD VT[p]|

E Vin>VDD||VT[p]|

N-DEVICE

Saturated since VDS[n] > VGS[n] VT[n] & VOUT > Vin VT[n]
Saturated VSD[p] > VSG[p] - |VT[p]| VOUT < Vin + |VT[p]| VOUT f(Vin)

Non saturated.
Saturated

Nonsaturated
Cut-off

P-DEVICE

OUTPUT

VOUT = VSS

Vin - VT[n] < VOUT < Vin + VT[p]|

n/p RATIO
The ratio n (gain of NMOS) / p (gain of PMOS) determines the switching point of the CMOS inverter.

Switching characteristics of CMOS inverter


The switching speed of the CMOS gate is limited by time taken to charge and discharge the load capacitance

Calculation of rise & fall time


Fall Time : The time required for output to fall from 90% to 10% of its steady state value.

Tfall = Rfall * CLOAD Rfall = k * 1/n Tfall 3 CL/VDD n

k is a constant that depends on threshold voltage and VDD.

k = 3 to 4 for values of VDD = 3 to 5 volts & VT = 0.5 to 1 volt.


The gain factor = nox/tox W/L

Calculation of rise & fall time


Rise Time : The time required for the output to rise from 10% to 90% of its steady state value and is given as

Trise = Rrise * CLOAD Rrise = k * 1/p Trise 3 CL/VDD P

Thus to achieve high speed circuits, the load capacitance should be minimized. Lowering the supply voltage on a circuit will reduce the speed of the gates in that circuit.

Calculation of rise & fall time


The delay of transistor decreases as the width of transistor is increased or length is decreased.. For equally sized n and p transistors n = 2p Hence Tfall = Trise /2

To equalize the rise and fall times of an inverter I.e Trise = Tfall must have n = p
This means, that channel width for the p-device must be increased to twice of the n-device. Hence if Lin = Lp then Wp= 2*Wn Thus n = p provides equal current source and sink capabilities. Thus equal charge and discharge times

DELAY TIME
Delay time : It defines the response of gate for change in input. Is measured between 50% transition points of input and output waveforms. Gate displays different response times for rising and falling waveforms. Tplh Defines response time of gate for low to high output transition Tphl Defines response time of gate for high to low output transition The overall propagation delay is average of these two

TD = (Tphl + Tplh)/2

CMOS Inverter - Summary


CMOS inverter uses actively driven P-channel transistor is used as pull-up drive. It allows maximum logic voltage level swing. Eliminates the static power dissipation as no current flows from VDD to ground in steady state. No of transistors is 2N. It is ratio-less logic I.e. Output logic levels are independent of ratio of pull-up and pull-down transistor sizes.

CMOS Inverter - Summary


Resistance of N-channel transistor is Rn Ln /Wn * Kn Resistance of P-channel transistor is Rp Lp /Wp * Kp But n = 2.5 p, hence n = 2.5 p . Hence in order to achieve symmetrical operation(equal rise & all time ) we must have (Ln * Wp)/ (Lp * Wn )= kn/kp =2.5. Thus with this sizing N & P transistors have equal I-V characteristics.

NOISE MARGIN
Logic High Output range

VOH(MIN)
VIH(MIN) VIL(MAX)

Logic High input range

Logic low output range

VOL(MAX)

Logic low input range

Output characteristics

Input characteristics

NOISE MARGIN
Determines the allowable variation in input voltage of gate so that output is not affected. Is specified in terms of two parameters Low noise margin - NML = VIL(MAX) - VOL(MAX) High noise margin - NMH = VOH(MIN) - VIH(MIN)

MOS Transistors As Switch


N-MOS source is tied to ground, used to pull signals down Control output G= 1 G= 0 D=0 D=Z

P-MOS source is tied to VDD, used to pull signals up.

Control Output
G= 0 G= 1 D= 1 D= Z

Why N-MOS Transistor Produces Weak 1?


When G= 0, output is 0 since transistor is non-conducting irrespective of Vin.

As source goes from 0V VGS goes from 5V

5V 0V

When Vs < 4.3 V then VGS < VT Hence VD is left at 5V VT .

Why PMOS Transistor produces weak 0 ?


PMOS will conduct if VGS < VT If VT = - 0.7V then VGS = 0V 5V & C is charged towards VDD since | VGS | > | VT | (5V >0.7V)

When source goes from 5V to 0V,

PMOS will stop conducting when | VGS | < | - 0.7 V | and VD will be left at 0.7V, thus produces weak 0.

MOS Transistors as Switch


SWITCH I/P G=0 G O/P X VDD G=1 Z STRONG 0
INPUT (SOURCE) OUTPUT(DRAIN )

VSS
X

WEAK 1
Z STRONG 1

G=0 VDD G=0

VSS

WEAK 0

Thus NMOS transistor produces active low logic at output. While PMOS gives active high logic at output.

Series Connection Of Switches


If N switches are placed in series Y=0 if A & B are 1. Thus yields an AND function. Y=A*B

If P switches are placed in series Y=1 if A & B are 0.

Thus yields an NOR function. Y = /A * /B

Parallel Connection Of Switches


When N switches are placed in
Thus yields an OR function. Y=A+B parallel Y=0 if either A or B is 1.

When P switches are placed in parallel Y=1 if either A or B is 0. Thus yields an NAND function. Y = /A +/B

Complementary Logic Gate Design


A CMOS gate is combination of two networks as shown below VDD Pull Up Network (PUN) consists of I/P P-MOS transistors. Thus implements PUN the logic function F Y= /F I/P

PDN
VSS

Pull Down Network(PDN) consists of N-MOS transistors. Thus implements the logic function /F.

NAND Gate Design

PDN network consists of two parallel P-MOS transistors PUN network consists of two series N-MOS transistors WHY ???

NAND Circuit

NOR gate design

Y = /(A + B)

PDN = /A * /B

PUN = A + B

NOR circuit

AND Gate Design

Poorly Designed AND gate !!!

AND Gate Design


Instead use this !

Designing compound gates


To implement the function F = /((A+B+C)*D) PDN will provide 0s of function F. Hence equation of PDN is (A+B+C) * D.

For PUN network will provide 1s of function F


Equation for PUN is (/A*/B*/C)+/D

Transmission gate
By combining an N-switch and P-Switch in parallel perfect transmission of both 1s and 0s is achieved. When I=0 both N & P devices are OFF

VIN X

VOUT Z

When I=1 both N & P devices are ON

VIN VDD

VOUT VDD

VSS

VSS

Transmission gate
Schematic icons for transmission gate

Most widely used

Transmission Gate Characteristics


2:1 multiplexer

Y = SA +/SB

This implementation will need total 6 Transistors 4 Transistors for two pass gates 2 Transistors for inversion of S Thus transmission gate logic uses less gates than the design with normal gates.

Transmission Gate Characteristics


But it has got Demerits ! It is non restoring logic output levels may or may not settle at VDD or VSS (as it passes logic level at input to output ) where as CMOS gates provides restoring logic. It has no drive capability, drive comes from original A, B inputs What about CMOS gate? Transmission gates provide no isolation between input and output.

Then Why & Where Transmission gates are used ?

Logic Design Using Transmission gate


2:1 Multiplexer Y = SA + /SB A X B 0 S 0 Y 0(B)

X 0
1

1 X
X

0 1
1

1(B) 0(A)
1(A)

When S= 1, S1 is ON and S2 is OFF. Hence input A is connected to the output. When S= 0, S1 is OFF and S2 is ON. Hence input B is connected to the output.

2:1 Multiplexer

Multiplexer may also be constructed using logic gates However these implementations are larger and consume more power than a transmission gate implementation Design Style Static CMOS Gates Transmission Gates Transistor Count 12 Why ?
4

Comparison of Multiplexers

TRISTATE INVERTER
EN 0 1 1 I X 0 1 O Z 1 0

Total 6 transistors !

D LATCH Positive level sensitive


EN 1 0 Q D Qold

How will the circuit function ?

D-LATCH WITH EN = 1

When EN = 1, switch S1 is closed and S2 is open. Hence Output-Q follows Input D

D-LATCH WITH EN = 0

When EN = 0, switch S1 is open and S2 is closed. Hence, Output-Q is isolated from Input D. Output retains the value of D at the falling edge of EN [WHY ???]

D-latch with asynchronous reset

How to implement asynchronous clear?

Edge Triggered D - Register


By combining two latches in master-slave arrangement, edge triggered register can be constructed. For positive edge triggered first latch(master) is negative level- sensitive latch

Rising Edge Triggered D register

How will the the circuit function ?

D-REGISTER WITH CLK = 0

When CLK= 0 , S1 and S4 are closed /Qm follows D, and Q is stored in the inverter loop.
Qm Qm

D-REGISTER WITH CLK = 1

When CLK= 1. S1 is open and S2 is closed Hence /Qm LATCHES the value of D, that existed on the rising edge of CLK. [ Does it remind you of set-up hold and Metastability] S3 is closed and S4 is open, Hence Q gets the value of /Qm [ I.e. the value of D on the rising edge of CLK].

Q is isolated from changes on D input.

EDGE TRIGGERED DREGISTER


In case of negative edge triggered register master is positive level-sensitive latch.

Guess The Functionality ?

Sizing of NAND and NOR Gates


Symmetrical drive capability of CMOS allows comparable transition time for output voltages irrespective of direction of transition. Primary effect of sizes of pull-up and pull-down transistors is on equivalent resistance of transistors in conduction state. To obtain symmetrical characteristics at output rise and fall time should be same which says RN = RP

Resistance of N-channel is given as RN LN/WN*KN


Resistance of P-channel is given as RP LP/WP*KP

Sizing of NAND and NOR gates

In case of NAND gate equivalent pull-down resistance is twice that of either pull-down alone RDN = Rn3 + Rn4= 2* Lp/Wp *

Kp.

To obtain symmetrical characteristics we must have

Kn/Kp = 1.25 ??

Sizing Of NAND and NOR gates

In case of NOR gate the equivalent pull up resistance is twice that of either pull up alone Rup = Rp3 + Rp4= 2* Lp /Wp * Kp Hence for NOR gate Kn/Kp = 5 will give symmetrical output ?? NOR gates require greater silicon area than NAND gate for symmetric drive operation. Hence NAND gates are always preferred than NOR

Where Does Power Go In CMOS


Power dissipated in a CMOS circuit is categorised as follows, Static dissipation : Due to leakage currents Dynamic dissipation : Due to charging and discharging of internal & load capacitance. Short circuit dissipation : Due to short circuit path between VDD and GND during switching

STATIC DISSIPATION
Static dissipation is due to, Leakage currents in the reversed-biased diodes formed between the substrate (or well) and source/drain regions. Sub threshold conduction, also contributes to static dissipation. Sub threshold leakage increases exponentially as threshold voltage decreases. Total static power dissipation is given as

pstatic=

leakage current * VDD

Where n is the number of devices in a CMOS Circuit.

STATIC DISSIPATION

LOWER BOUND ON THRESHOLD TO PREVENT LEAKAGE

DYNAMIC POWER DISSIPATION

During low to high transition part of energy drawn from supply is dissipated in PMOS. During high to low transition stored energy on capacitor is dissipated in NMOS transistor. Dynamic power dissipation gives measure of this energy consumption

DYNAMIC POWER DISSIPATION

The average dynamic power consumption for input frequency of F is Pdynamic = CL * VDD2 * F Power dissipation is independent of device parameters This can be reduced by decreasing CL , VDD or F

SHORT CIRCUIT DISSIPATION


It is the DC power consumed during switching. A direct current path from VDD to ground exists when both N and P transistors are conducting simultaneously during switching. Short circuit consumption is given by Thus for an inverter without load, assuming Tr = Tf = Trf

Psc = Imean * VDD

Psc = /12 (VDD 2 VT) 3 Trf * F

Short- circuit power depends on W/ L ratios of the transistors Greater the rise-fall time of the signals, larger is the power consumed.

SHORT CIRCUIT DISSIPATION

Input switching waveform & model for short circuit current. Ipeak is determined by the saturation current of devices, hence is proportional to the sizes of the transistor.

Impact of rise/fall times on shortcircuit currents

Power dissipation due to short circuit current is minimized by mminimizing the rise and fall time of input and output signal.

MOS Device Capacitance Estimation


Junction Cap. Bulk Cap. Overlap Cap.

Two types of capacitances - Junction & Gate capacitance.

MOS Device Capacitance Estimation

Gate capacitance :- CG = bulk capacitance + overlap capacitance

CG = CGB + CGD + CGS CG = COX * W* Leff

MOS Device Capacitance Estimation


COX is oxide capacitance & is inversely proportional to oxide thickness COX = eOX/tOX But both source and drain tend to extend somewhat below the oxide by an amount XD called lateral diffusion. Hence effective channel of transistor Leff becomes shorter than the length, the transistor was designed It also gives rise to parasitic capacitance between gate and source (drain) & is called overlap capacitance. This is strictly linear & has fixed value. The reverse biased source-bulk & drain-bulk pn junctions contribute to junction capacitance.

STAGE RATIO

- The object is to maximize the speed with minimum area overhead. - Option I will be slow, since Gate1 will not have drive capability to drive the large inverter. - In Option II, we have a chain of inverters of increasing size (by an order of a) - Gate I will be fast, since it drives a minimum sized inverter (Inverter 1)

STAGE RATIO
When It is desired to drive large load capacitances such as long buses, I/O buffers and off-chip capacitive loads. This is achieved by using a chain of inverters where each successive inverter is made larger than the previous one.

The ratio by which each stage is increased in size is called stage ratio .
The signal delays encountered in driving the off chip load directly from a minimum sized inverter is unacceptable. The optimisation to be achieved here is to minimize the delay between input and output while minimizing the area and power dissipation.

STAGE RATIO

Stage Ratio
Inverter 1 is a minimum- sized device. Subsequent inverter device sizes increase by a factor of a Delay of each stage is a * Td , where Td =delay of minimum sized inverter driving an identically sized inverter Hence total delay ( delay through the n stages ) is n * a * Td Cg is load of first driver which is minimum- sized device

If CgN is the load capacitance of the Nth inverter then CgN = Cg * aN


To guarantee that none of capacitances internal to the chain of inverters exceed Cload [ why ?? ] we must have Cg * an = CL here n = N +1 I.e an = CL/ Cg

Stage Ratio
Hence a = [CL/ Cg]1/n Total delay = n * [CL/ Cg]1/n * Td The optimum value of n is Now optimum value of a i.e aopt can be calculated as;

nopt = ln [CL/ Cg]

Taking natural log of both sides we get a = 2.7 But the actual stage ratio is given by

an = CL/ Cg aln [CL/ Cg] = CL/ Cg

aopt = exp[(k+aopt)/aopt] Where k = Cdrain/Cgate For 1 process k = 0.215, hence aopt = 2.93 For 2.5 process k = 3.57 which gives aopt = 5.32

Stage Ratio Graph

Different processes will have different stage ratios from 2 to 10.

SIZING OF TRANSISTOR
- The speed of a CMOS gate depends on how fast the load capacitance can be charged/ discharged - Load capacitance depends on the fan- out of the gate, and the size of the gates connected to the output

- Increasing the size of the p- transistors decreases the rise- time of the gate
- Increasing the transistor sizes of the N and P- block also increases the input capacitance of the gate (i. e. the input delays will increase at the expense of the output delays)

SIZING OF TRANSISTOR
Progressive Sizing: I.e. Increase size downwards m1>m2 > m3 >m4 m 4 discharge CL m 3 discharge CL + C3 m 2 discharge CL + C3 + C2 m 1 discharge CL + C3 + C2 + C1

SIZING OF TRANSISTOR

SCALING OF MOS TRANSISTOR DIMENSIONS

Sub-micron Considerations
When dimensions of MOS device go below 1 then its behaviour deviates substantially than actual MOS operation that has been discussed so far. For sub-micron range the channel length becomes comparable to other device parameters such as depth of drain and source junctions, and width of their depletion regions. Such devices are called short channel transistors & represents the deviation form ideal model.

Sub-micron Considerations

Variation in I-V characteristics


The I-V characteristics of short channel device deviate considerably from the ideal equations.

ID= {( VGS -VT)VDS (VDS2)/2}


ID= (kn/2) W/L(VGS -VT)2 (1+ VDS)

-----Linear region
----Saturation region

The most important reasons for this difference are the Velocity saturation and mobility degradation. Velocity saturation : Carrier velocity is given as; n = n Ex = n*dv/dx This states that carrier velocity is proportional to electric field & is independent of value of that field, i.e it is constant.

Variation in I-V characteristics


But when the electric field along the channel reaches a critical value Emax , velocity of carriers tend to saturate( i.e carriers reach their maximum limited velocity ). Current under the velocity saturated condition is IDSAT = SAT* Cox * W (VGS - VDSAT - VT)

Thus the saturation current linearly depends on the gatesource voltage.


Also, ID is independent of L in velocity saturated devices. Reduction in the channel length causes reduction in the electron mobility even at normal electric field levels. This is called mobility degradation

Mobility Variation
The mobility M describes the ease with which carriers drift in the substrate material It is defined as ratio of average carrier drift velocity V to the Electric field E.

Mobility can vary in number of ways viz :


According to the type of charge carrier. [ WHY ??] Increase in doping concentration decreases mobility. Increase in temperature decreases mobility.

Threshold Voltage Variation


As device dimensions are reduced threshold voltage becomes function of L,W and VDS
The threshold voltage is not constant with respect to the voltage difference between the substrate and the source of the MOS transistor. This is known as substrate bias effect or body- effect .

Impact Ionization
As the length of the gate decreases, electric field intensity at the drain increases.

In sub- micron devices, the field intensity can become very high, to an extent that electrons are imparted with enough energy to become hot.
These hot electrons can impact the drain, dislodging holes.

These free holes will escape into the substrate creating a substrate current. This effect is known as impact ionization
This will degrade the transistor performance and can trigger latch-up.

Impact Ionization
The high- energy (hot) electrons can also penetrate the gate oxide causing a gate current. This can lead to degradation of MOS device parameters. Hot- electron effects can be minimized by decreasing the supply voltage in smaller devices.

Tunnelling
The gate is separated from the substrate by an oxide of thickness tox , Generally the gate current in a MOS Transistor is zero When the gate oxide is very thin, a current can flow from gate to source/drain This happens due to electron tunnelling through the gate oxide This gate-current is proportional to the area of the gate. This effect puts a limit on the minimum thickness of the gate oxide layer, as processes are scaled

Drain Punch-Through
If the drain is at a very high voltage with respect to source , the depletion regions around the drain and source will meet This will cause a channel current to flow, irrespective of the gate voltage, even if it is zero. This is known as punch-through condition.

Punch-through can be avoided with,


Thinner Oxides. Larger Substrate Doping.

Shallower Junctions.
Longer Channels.

Channel Length Modulation


In the saturation region, the ideal characteristics of a MOS transistor shows a constant current region, i.e.drain current IDS remains constant, with increase in VDS. This characteristics assumes that carrier mobility is constant.It does not take into account the variation in channel length due to change in VDS. However, in saturation, as VDS increases, channel- length L decreases by a very small amount such that Leff = L - DL This decrease in L, increases the [W/ L] ratio, and hence increases IDS due to increase in . For long channel lengths, influence of channel variation is of little consequence, but as devices are scaled down this variation has be taken into account.

Scaling Methods
Scaling is method in which device geometries migrate to lower sizes while still maintaining the same device characteristics. This is done by scaling the critical parameters of a device in accordance to a given criteria. Scaling methods include Lateral scaling Constant field scaling Constant voltage scaling

Lateral Scaling
LATERAL SCALING : - Here the only parameter that is scaled is the gate length L - This method of scaling is also called gate- shrink - It can be easily done to an existing mask design - Power dissipation increases by the factor - Input capacitance of the transistor decreases by the factor

CONSTANT FIELD SCALING


- A dimensionless scaling factor is applied to, All dimensions (including vertical dimensions such as tox ) are decreased by . Device voltages are decreased by Concentration densities are increased by As a result, Depletion thickness d. Threshold voltage VT Drain Current IDS also get scaled by the same factor Since the voltage VDD is scaled, the electric field in the device remains constant Hence, the operating characteristics of the device remain the same even after scaling Power dissipation decreases by 2

CONSTANT VOLTAGE SCALING


Similar to constant field scaling, except that voltage VDD is kept Constant The current I DS increases by the factor Speed of the device increases by the factor Number of transistors per unit area increases by the factor 2

As a result, the current density increases by the factor 3


Proportionately wider metal wires are required for more densely packed structures Power dissipation increases by the factor

This will increase the need for cooling devices/ structures for the IC
Power dissipation of above 1- 2 Watts require specialized cooling fins or packaging

CMOS PROCESSING TECHNOLOGY

IC Fabrication
An IC fabrication process contains a series of masking steps to Create successive layers of insulating, conducting and semiconducting material that define the transistors and metal interconnect. Techniques such as oxidation, implantation,deposition are used to build these layers.

The starting material used for IC fabrication is silicon wafer. Wafer is a disk of silicon, 4" to 8" in diameter, < 1mm thick,
Wafers are very brittle, the larger the diameter, the more susceptible to damage. Surface of the wafer is polished to a very flat, scratch free surface. The single crystal silicon used as substrate is obtained from polycrystalline silicon generally by CZ (czochralski ) process. Controlled amount of impurities are added to the melt to provide

IC Fabrication
The crystal with required electrical characteristics. After the crystal has been developed several steps are involved to achieve mirror like structure. Oxidation is used to deposit Silicon Dioxide (SiO2) on surface of wafer to be used as insulting material - heat wafers inside of an oxidation atmosphere such as oxygen or water vapour. To build the micro(semiconductor) devices, we need junctions formed by N and P type region. To create these regions on silicon wafer what we need is process to introduce impurity atoms into the substrate.This may be achieved by using Epitaxy, Deposition and Ion-implantation.

IC Fabrication
Epitaxy involves growing single crystal film (of the required dopant) on silicon surface by heating wafer and exposing it to a source of the dopant. Deposition is to evaporate the dopant onto the surface, then heat the surface to drive the impurities in the wafer Ion implantation involves exposing surface to highly energized dopant atoms. When these atoms impinge on the surface, they travel below the surface forming the regions with varying doping concentration. During fabrication of the transistors or other structures it is needed to block some regions from receiving the dopants. Hence special material called as mask is used to block the impurities in particular region.

MASK

Common material used for masks are Photoresist, Polysilicon, Silicon dioxide, Silicon nitride.

To create mask: (a) deposit mask material over entire surface (b) cut windows in the mask to create exposed areas (c) deposit dopant (d) remove un-required mask material Masks plays important role in process called selective diffusions. The selective diffusion involves 1. Patterning windows in a mask material on the surface of the wafer. 2. Subjecting the exposed areas to a dopant source. 3. Removing any un-required mask material.

Fabrication of CMOS Devices


Technologies used for CMOS fabrications include N-well process P-well process

Twin-tub process
Silicon on insulator.

P-Wells and N-Wells


IN

P substrate contact

D
n+

OUT S

N substrate contact p+ n+

n+ p+
P-well

p+

N-well

A p- transistor is built on an n- substrate and an n- transistor is built on a p-substrate

P-Wells and N-Wells


In order to have both types of transistors on the same substrate, the substrate is divided into well regions (Shaded

region in the standard cells)


Two types of wells are available - n- well and p- well In a p- substrate, an n- well is used to create a local region of n type Substrate, wherein the designer can create p- transistors In a n- substrate, a p- well creates a local p- type substrate region, to accommodate the n- transistors.

Hence, every p- device is surrounded by an n- well, that must be connected to VDD via a VDD substrate contact.
Similarly, n- devices are surrounded by p- well connected to GND using a GND substrate contact.

The P-Well Process

Mask1 Defines the areas in which the deep P-Well diffusions are to take place. Mask2 Defines the thinox regions, namely those areas where the thick oxide is to be stripped & thin oxide grown to accommodate p & n transistors and diffusion wires. Mask3 Used to pattern poly-silicon layer which is deposited after the thin oxide. Mask4 - A p+ mask is now used to define all areas where pdiffusion is to take place. Mask5 This is usually performed using the negative form of p+ mask & with mask2 defines those areas where n-type diffusion is to take place. Mask6 Contact cuts are now defined. Mask7 _ The metal layer pattern is defined by this mask Mask8- It is needed to define the openings for access to bonding pads.

The N-Well Process


Typical fabrication processes for n-well are similar to that of p-well process except that n-well is implanted rather than pwell. Nwell CMOS circuits are superior to P-well because of lower substrate bias effects on transistor threshold voltage & inherently low parasitic capacitances associated with source & drain regions.

The N-Well Process


1. Steps in N-well process : Formation of n-well regions

2. Define N-MOS and P-MOS active areas. 3. Field and gate oxidations (thinox) 4. Form and pattern poly-silicon. 5. P+ diffusion 6. N+ diffusion 7. Contact cuts

8. Deposit and pattern metallization


9. Over glass with cuts for bonding pads

LATCH-UP IN CMOS Circuits


Latch-up is condition in which parasitic components gives rise to establishment of low-resistance conducting path between VDD & VSS This results in chip self-destruction or system failure.

Latch-up may be induced by the glitches on the supply rails or by incident radiation.

Physical Origin Of Latch-up


Vss
IN OUT

Physical Origin Of Latch-up

VSS

Latch-up Mechanism
If sufficient current is drawn from NPN emitter then NPN ( Q2 )turns on when VBE 0.7V. When NPN turns on, note that emitter current increases exponentially with VBE Current flowing through the parasitic nwell resistors will eventually turn on the parasitic PNP. As PNP turns on, the NPN base current increases and voltage drop across Rsubstrate also increases, further increasing the NPN emitter current (Q2 turns on harder), which further increases the PNP base current, which again further increases NPN base current.

Rwell
Q1

Q2

Rsubstrate

Remedies for the latch-up problem


One way is to keep the p-substrate tied very closely (i.e.close proximity) to GND (most negative supply) to reduce substrate

resistance (RS1 &RS2), and the n-well tied very closely to VDD
to reduce RW1 & RW2. Each well must have a substrate contact of appropriate type (ntype for n-well). Place substrate contacts as close as possible to the source connection of transistors connected to the supply rails. Place a substrate contact for every 5- 10 transistors

Lay out n and p- transistors with packing of n- devices towards Gnd and p- devices towards VDD

CMOS Process Layers


Layer Well (p,n) Active Area (n+,p+) Select (p+,n+) Polysilicon Color Yellow Green Green Red Blue Magenta Black Black Black Representation

Metal1 Metal2
Contact To Poly Contact To Diffusion Via

STICK DIAGRAM
Before the cell can be constructed from a transistor schematic it is necessary to develop a strategy for the cell's basic layout. Stick Diagrams are a means for the design engineer to visualize the cell routing and transistor placement.

Steps involved in stick diagram construction.


STEP 1 : Identify each transistor by a unique name of its gate signal Identify each connection to the transistor by a unique name

STICK DIAGRAM

Figure 1: Schematic and Graph

STICK DIAGRAM

Figure.2 Euler path

STICK DIAGRAM
STEP2 :
Eulers paths : A path the traverses each node in the path, such that each edge is visited only once. The path is defined by the order of each transistor name. The Euler path of the Pull up network must be the same as the path of the Pull down network. Euler paths are not necessarily unique. It may be necessary to redefine the function to find a Euler path. F = E + (CD) + (AB) = (AB) +E + (CD)

STICK DIAGRAM

Next step is to lay out the stick diagram : Trace two green lines horizontally to represent the NMOS and PMOS devices. The gate contact to the devices are represented by vertical strips. Surround the NMOS device in a yellow box to represent the surrounding Pwell material.

STICK DIAGRAM

Figure 3: Connection label layout

STICK DIAGRAM
Surround the PMOS device in a green box to represent the surrounding N-well material. Trace a blue line horizontally, above and below the PMOS and NMOS lines to represent the Metal 1 of VDD and VSS.

Label each Poly line with the Euler path label, in order from left to right.
Place the connection labels upon the NMOS and PMOS devices. Place the VDD, VSS and all output names upon the NMOS and PMOS devices

STICK DIAGRAM

Figure 3: Connection label layout

Figure 4: Stick Diagram, Interconnected STICK DIAGRAM

SCHEMATIC

STICK-DIAGRAM NAND GATE

STICK DIAGRAM NOR GATE

Layout Design Rules


Design rules provide guidelines for preparing the photo masks used in the fabrication of IC & consists of minimum-width ( line widths ) and minimum-spacing ( interlayer ) constraints. The main goal of layout rules gives a circuit with optimum yield in as small an area as possible without compromising reliability of the circuit. Design rules specify the fundamental unit as minimum line

width.

It stands for the minimum mask dimension that can be safely transferred to the semiconductor material. The minimum line width is set by resolution of the patterning process ( lithographic process ) The interaction between different layers. Even for the same minimum dimension, design rules tend to differ form company to company and from process to process.

Layout Design Rules


One approach to address this is to use scalable design rules. These include 1. Lambda ( ) based rules. 2. Micron rules. Lambda ( ) based rules : These defines all the rules as function of single parameter called . Scaling of the minimum dimension is accomplished simply by changing the value of .. This results in linear of all dimensions. When mapping the transistor schematic/layout to a particular technology, the actual W, L will be calculated as:

W' (actual, microns ()) = W (microns)


L' (actual, microns) = L (microns) (lambda ) is dimensionless unit called as scaling factor.

Layout Design Rules


Scaling factor allows transistor W/L values specified in schematic to be technology independent.(feature-size independent way of setting out mask dimensions to scale.) Scaling factor allows transistor W/L values specified in schematic to be technology independent.(feature-size independent way of setting out mask dimensions to scale.)

Scaling factor lambda is foundry/silicon vendor dependent


For MOSIS foundry vendors, 1.2 technology = 0.6

0.8 technology = 0.4.


Typically the minimum gate length is set to 2 and width is varied.

Layout Design Rules


But linear scaling is possible over a limited range of dimensions, hence these rules are not used by industry. Normal industrial practice is to deal with micron rules . Micron rules expresses the design rules in absolute dimensions and hence can exploit the features of a given process to a maximum degree. These are usually given as list of minimum feature sizes and spacing for al the masks required in an given process.

Design Rules
Examples from AMS 0.6micron technology

Intra-Layer Design Rules

Vias and Contacts

N Transistor Layout
Bulk Source Gate Drain

Thin-Oxide

P Transistor - Layout
Bulk Source Gate Drain

nThin-Oxide

Parallel/Series Transistors

Large MOS Transistors

AND Gate Layout

AND Gate - Layout

Inverter Layout

Inverter Layout

4-Input NAND Gate

Pseudo NMOS NAND Gate

Logic Graph for F = (A+B)C

Layout F = (A+B)C

Logic Graph for F = /(AB+CD)

Pass Transistor Based Multiplexer

Pass Transistor Based Multiplexer

Pass Transistor Based Multiplexer


Gate Source

Gate oxide

Drain

substrate

n+
holes

channel

n+
electrons

P-substrate

You might also like