LPVLSI Unit 2 Notes
LPVLSI Unit 2 Notes
MOS Inverters-Introduction
When any of the noise margins is low, the gate is susceptible to a switching
noise at the input.
Normally, this output is used to drive other gates. , this voltage can be accepted as low level provided
it is less than Vt. So,
Assuming the typical value of threshold voltage Vtn = 0 2. Vdd , we get
This imposes a restriction on the minimum value of load resistance for a successful operation of the
circuit as an inverter.The input–output characteristic of the inverter is shown in Fig
The circuit operates along the load line as shown in Fig. 4.5b.
For Vin = 0 V, the output voltage Vout = Vdd (point A), and for Vin = Vdd, the output
voltage Vout = VOL, as shown by point B.
The transfer characteristic is shown in Fig. 4.5c, which shows that the output is Vdd for Vin
= OV, but for Vin = Vdd the output is not OV(VOL).
This implementation of this inverter has a number of disadvantages:
• As the charging of the output capacitor takes place through the load resistor RL and
discharge through Rc and their values must be different, there is asymmetry in the ON-to-
OFF and OFF-to-ON switching times.
• To have higher speeds of operation, the value of both Rc and RL should be reduced.
• However, this increases the power dissipation of the circuit.
• Moreover, as we shall see later, to achieve a smaller value of Rc, the area of the MOS
inverter needs to be increased.
This implementation of this inverter has a number of disadvantages:
• The resistive load can be fabricated by two approaches—using a diffused resistor approach
or using an undoped poly-silicon approach.
In the first case, an n-type or a p-type isolated diffusion region can be fabricated to realize a
resistor between the power supply line and the drain of the nMOS transistor.
To realize a resistor of the order of few K Ω, as required for proper operation of the circuit,
the length to width must be large.
To realize this large length-to-width ratio in a small area, a serpentine form is used as shown
in Fig. 4.6.
3. Pull-down and pull-up device, both in saturation: This is represented by point C on the
curve. In this situation,
1. Pull-down device in linear region and pull-up device in saturation: This situation occurs
when input voltage is equal to Vdd. Here,
Fig. 4.8 a nMOS inverter with enhance-mode transistor as a pull-up device; b transfer
characteristic. nMOS n-type metal–oxide–semiconductor
Let us consider the output voltage for two situations—when Vin = O and Vin = Vdd.
In the first case, the desired output is Vdd.
But as the output, Vout, approaches the voltage ( Vdd − Vtn), the pull-up transistor turns off.
Therefore, the output voltage cannot reach Vdd.
The maximum output voltage that can be attained is ( Vdd − Vtn), where Vtn is the threshold
voltage of the enhancement-mode pull-up transistor.
The output voltage for Vin = Vdd is not 0 V, because in this case both the transistors are
conducting and act as a voltage divider.
The transfer characteristic is shown in Fig. 4.8b.
From the above discussion, we can make the following conclusion:
• The output is not ratioless, which leads to asymmetry in switching characteristics.
• There is static power dissipation when the output level is low.
• It produces weak low and high output levels.
As a consequence, nMOS enhancement-type transistor is not suitable as a pull-up device for
realizing an MOS inverter.
Substituting
We get,
Region 3: Vin = Vinv At this point, both the transistors are in the saturation condition as
represented by the point C on the superimposed characteristic curves.
In this regions, both the transistors can be modeled as current sources.
assuming
we may equate the saturation currents of the pull-up and pull-down transistors to get
Equating we get,
Equating we get,
As the input voltage has been increased above Vinv, the nMOS transistor moves from the saturation
region to the linear region, whereas the pMOS transistor remains in saturation.
With the increase in input voltage beyond Vinv, the output voltage and also the drain current
continue to drop. A representative point in this region is point D.
In this region, nMOS transistor acts as a resistor and pMOS transistor acts as a current source.
The drain current for the two transistors are given by
or
Region 5:
In this region, the pull-up pMOS transistor remains OFF and the pull-down nMOS transistor
goes to deep saturation. However, the current flow through the circuit is zero as the p
transistor is OFF and the output voltage VO = 0.
Current flows only during the transition period. So, the static power dissipation is very
small.
Moreover, for low and high inputs, the roll of the pMOS and nMOS transistors are
complementary; when one is OFF, the other one is ON. That is why this configuration is
known as the complementary MOS or CMOS inverter.
βn/βp Ratio: As we have mentioned earlier, the low- and high-level outputs of a CMOS inverter are
not dependent on the inverter ratio. However, the transfer characteristic is a function of the βn/βp
ratio. The transfer characteristics for three different ratio values are plotted in Fig.
Here, we note that the voltage at which the gate switches from high to low level ( Vinv) is
dependent on the βn/βp ratio. Vinv increases as βn/βp decreases.
For a given process technology, the βn/βp can be changed by changing the channel
dimensions, i.e., the channel length and width. Keeping L the same, if we increase Wn/Wp
ratio, the transition moves towards the left and as Wn/Wp is decreased, the transition moves
towards the right as show in Fig.
As the carrier mobility depends on temperature, it is expected that the transfer characteristics
will be affected with the temperature. However, both βn and βp are affected in the same
manner (βαT−15 . ) and the ratio βn/βp remains more or less the same.
On the other hand, both the threshold voltages Vtn and Vtp decrease with increase in
temperature leading to some shrinkage of the region I and expansion of region V.
a An nMOS inverter driven by another inverter; b inverter with Vin = Vdd; and c inverter
with Vin = Vdd – Vt.
Equating we get
where Z is known as
the aspect ratio of the MOS devices
Substituting typical values Vtn = 0.2Vdd , Vtdp = −0.6Vdd and Vinv = 0.5Vdd ,
This ratio Zpu / Zpd is known as the inverter ratio ( Rinv) of the inverter, and it is 4:1 for an inverter
directly driven by another inverter.
Therefore,
Now, for the depletion-mode pull-up transistor, Vgs = 0 ,
In a similar manner, we can consider the case of inverter 2, where the input
voltage is ( Vdd−Vtp). As in case of inverter 1, we get
If we impose the condition that both the inverters will have the same output,
Substituting typical values Vtn = 0.2 Vdd and Vtp = 0.3 Vdd , we get
Switching Characteristics
• Delay-Time Estimation
• Ring Oscillator
•
Fig:a Parasitic capacitances of a CMOS inverter. b CMOS complementary metal–
oxide– semiconductor
The delay td is the time difference between the midpoint of the input swing and
the midpoint of the swing of the output signal.
The load capacitance shown at the output of the inverter represents the total of
the input capacitance of driven gates, the parasitic capacitance at the output of
the gate itself, and the wiring capacitance.
The capacitances Cgd and Cgs are mainly due to gate overlap with the diffusion
regions, whereas Cdb and Cgb are voltage-dependent junction capacitances.
The capacitance Cout is the lumped value of the distributed capacitances due to
interconnection and Cgn and Cgp are due to the thin oxide capacitances over
the gate area of the nMOS and pMOS transistors, respectively
Switching Characteristics-Delay-Time Estimation
The capacitor CL discharges from voltage Vdd to Vdd /2 during time t0 to t1.
This expression gives us a simple analytical expression for the delay time. It is
observed that the delay is linearly proportional to the total load capacitance
CL.The delay also increases as the supply voltage is scaled down, and it
increases drastically as it approaches the threshold voltage.
To overcome this problem, the threshold voltage is also scaled down along with
the supply voltage. This is known as the constant field scaling.
Another alternative is to use constant voltage scaling, in which the supply
voltage is kept unchanged because it may not be always possible to reduce the
supply voltage to maintain electrical compatibility with other subsystems used
to realize a complete system.
The designer can control the following parameters to optimize the speed of
CMOS gates.The width of the MOS transistors can be increased to reduce the
delay. This is known as gate sizing, which will be discussed later in more detail.
The load capacitance can be reduced to reduce delay. This is achieved by using
transistors of smaller and smaller dimensions as provided by future-generation
devices.Delay can also be reduced by increasing the supply voltage Vdd along
and/or reducing the threshold voltage Vt of the transistors.
Switching Characteristics-Ring Oscillator
Fig:Output waveform of a three-stage ring oscillator
The time period can be expressed as the sum of the six delay times
For an n-stage (where n is an odd number) inverter, the time period T = 2•n .td
Therefore, the frequency of oscillation f = 1/ 2ntd or td = 1/ 2nf .
It may be noted that the delay time can be obtained by measuring the frequency of
the ring oscillator.For better accuracy of the measurement of frequency, a suitable
value of n (say 151) can be chosen.
This can be used for the characterization of a fabrication process or a particular
design. The ring oscillator can also be used for on-chip clock generation.
However, it does not provide a stable or accurate clock frequency due to
dependence on temperature and other parameters. To generate stable and accurate
clock frequency, an off-chip crystal is used to realize a crystal oscillator.
Delay Parameters
• Resistance Estimation
• Area Capacitance of Different Layers
• Standard Unit of Capacitance Cg
• The Delay Unit
Resistance Estimation:
The delays of CMOS and BiCMOS inverters are compared in Fig. 4.23 for
different fan-outs.
Here, it is assumed that for CMOS, Wp = 15 μm and Wn = 7 μm and for the
BiCMOS, Wp = Wn = 10 μ m and WQ1 = WQ2 = 2 μ m.
It may be noted that for fan-out of 1 or 2, CMOS provides smaller delay
compared to BiCMOS due to longer delay through its two stages.
Buffer Sizing
• It may be observed that an MOS transistor of unit length (2λ) has gate
capacitance proportional to its width ( W), which may be multiple of λ.
• With the increase of the width, the current driving capability is increased.
• But this, in turn, also increases the gate capacitance. As a consequence, the
delay in driving a load capacitance CL by a transistor of gate capacitance Cg
is given by the relationship (cL / cg ) , τ where τ is the unit delay, or delay in
driving an inverter by another of the same size.
• Let us now consider a situation in which a large capacitive load, such as an
output pad, is to be driven by an MOS gate. The typical value of such load
capacitance is about 100 pF, which is several orders of magnitude higher
than Cg.
• If such a load is driven by an MOS gate of minimum dimension (2λ x 2λ),
then the delay will be 103τ.
• To reduce this delay, if the driver transistor is made wider, say 103 × 2λ, the
delay of this stage becomes τ, but the delay in driving this driver stage is
1000τ, so, the total delay is 1001τ, which is more than the previous case.
• It has been observed that the overall delay can be minimized by using a
cascaded stage of inverters of increasing size as shown in Fig. 4.24.
•
• For example, considering each succeeding stage is ten times bigger than that
of the preceding stage, each stage gives a delay of 10τ.
• This results in a total delay of 31τ instead of 1001τ.
• Now, the question arises about the relative dimension of two consecutive
stages. If relative dimension is large, fewer stages are needed, and if relative
dimension is small, a large number of driver stages is needed.
• As each stage gives a delay of fτ, the total delay of n stages is
•
• For a large value of n, let buffers are introduced after each k stages, as
shown in Fig. 5.6 for k = 3.
• Assuming a propagation delay of each buffer is tbuf , the overall propagation
delay can be computed as follows:
Two nMOS pass-transistor logic network (one for each rail).Two small pull-
up pMOS transistors for swing restoration.Two output inverters for the
complementary output signals.
Fig:Basic complementary pass-transistor logic (CPL) structure
2-to-1 Multiplexer Realization using CPL Logic
This is the basic and minimal gate structure with ten transistors.
All two-input functions can be implemented by this basic gate structure.
(a) Basic swing-restored pass-transistor logic (SRPL) configuration; and (b) SRPL
realization of 2-input NAND gate
The SRPL logic style is an extension of CPL to make it suitable for low-power
low-voltage applications.
Output inverters are cross-coupled with a latch structure, which performs both
swing restoration and output buffering.The pull-up pMOS transistors are not
required anymore and that the output nodes of the nMOS networks are the
gate outputs.
As the inverters have to drive the outputs and must also be overridden by the
nMOS network, transistor sizing becomes very difficult and results in poor
output-driving capacity, slow switching, and larger short circuit currents. This
problem becomes worse when many gates are cascaded.
Double Pass-Transistor Logic
Gate Logic
• Fan-In and Fan-Out
• nMOS NAND and NOR Gates
• CMOS Realization
• CMOS NAND Gates
• CMOS NOR Gates
• Switching Characteristics
• CMOS NOR Gate
• CMOS Complex Logic Gates
Logic-Fan-In and Fan-Out:
• Fan-in is the number of signal inputs that the gate processes to generate some
output.Fan-out is the number of logic inputs driven by a gate.
•
nMOS NAND and NOR Gates:
• (a) n-input nMOS NAND gate; (b) equivalent circuits; and (c) n-input nMOS
NOR gate.
• Realizations of nMOS NAND gate with two or more inputs.
• Let us consider the generalized realization with n inputs with a depletion- type
nMOS transistor as a pull-up device and n enhancement-type nMOS transistors
as pull-down devices. In this kind of realization, the length/width (L/W) ratio of
the pull-up and pull-down transistors should be carefully chosen such that the
desired logic levels are maintained.
• The critical factor here is the low-level output voltage, which should be
sufficiently low such that it turns off the transistors of the following stages.
• To satisfy this, the output voltage should be less than the threshold voltage, i.e.,
Vout ≤ Vt = 0.2 Vdd.
• To get a low-level output, all the pull-down transistors must be ON to provide
the GND path.
For the Boundary Condition
• It may be noted that, not only one pull-down transistor is required per input of
the NAND gate stage but also the size of the pull-up transistor has to be
adjusted to maintain the required overall ratio.
• This requires a considerably larger area than those of the corresponding nMOS
inverter.Moreover, the delay increases in direct proportion to the number of
inputs.
• If each pull-down transistor is kept of minimum size, then each will represent
one gate capacitance at its input and resistance of all the pull-down transistors
will be in series. Therefore, for an n-input NAND gates, we have a delay of n-
times that of an inverter, i.e.,
CMOS Realization
(a) Equivalent circuit of n-input complementary MOS (CMOS) NAND gate; and
(b) Transfer characteristics of n-input CMOS NAND gate
To determine the inversion point, pMOS transistors in parallel may be equated
to a single transistor with the width n times that of a single transistor.
And nMOS transistors in series may be considered to be equivalent to have a
length equal to n times that of a single transistor
This makes the trans-conductance ratio =βn/ βp n2 .
(a) n-input complementary MOS (CMOS) NOR gate and (b) The equivalent Circuit
Tans-conductance ratio is equal To n2βn/βp
Therefore, with the increase in fan-in the inversion point moves towards left for
NOR gates, whereas it moves towards right in case of NAND gates.
Switching Characteristics
Intrinsic time constant of this network is given by