Complete VLSI Notes - 6thsem (21scheme)
Complete VLSI Notes - 6thsem (21scheme)
Prepared by : Dr.Savitha.M
Professor
EC Department
KVGCE- Sullia
19.09.2023
03.10.2022
19.09.2023
03.10.2022
19.09.2023
Non ideal characteristics :
The long-channel I-V model , equation of drain current , neglects many effects that are important to
devices with channel lengths below 1 micron. This section summarises the non -deal behaviour of
MOSFET
1. Mobility degradation
2. Velocity Saturation
3. Body effect
6. Sub-threshold Conduction
7. Gate leakage
Mobility Degradation:
ν=μE where ν is the carrier drift velocity and μ is mobility
The mobility term in the above equation was assumed to be independent of the
vertical electric field E. This is a reasonable approximation when the electric field
is weak but it breaks down for stronger electric fields. Stronger vertical electric
fields (produced by a high gate voltage) can attract the charge carriers travelling
along the inversion layer and cause it to collide with the oxide interface. This effect
causes mobility degradation, where the effective mobility decrease with increase in
Vgs.
Velocity Saturation:
Under strong lateral electric fields (produced by Vds), the charge carriers are
accelerated and the collisions between the charge carriers and the silicon lattice
increases. Beyond a certain level of electric field, the carrier drift velocity saturates
at νsat. This is called velocity saturation. Note that transistor saturation and
velocity saturation are two different phenomenons but both of them result in the
saturation of current through the channel (with respect to Vds).
ν=μE
In the above graph, we can see that the long channel device has linear relationship
between Ids and Vds until Vds=Vgs-Vt. After Vds=Vgs-Vt , the long channel
enters into saturation region. In the case of the short channel device, it enters into
velocity saturation region much before the drain pinch-off saturation region. In
short channel devices, depending upon which is lower VdsSAT or (Vgs-Vt), the
device can go into velocity saturation or drain pinch-off saturation. If VdsSAT is
smaller, then the short channel device enters into velocity saturation first. If (Vgs-
Vt) is smaller, it enters into drain pinch-off saturation first.
Body Effect:
So far we considered that source and body terminals are both connected together
and into ground. But what if the body is not connected to the source?
a) A long
channel device. The dotted lines represent the depletion region. b) A short channel device
Sub-threshold conduction:
So far we assumed that Ids=0 for Vgs<Vt. But current doesn’t cut off abruptly once
Vgs becomes less than Vt. As Vgs decreases below Vt, Ids decreases exponentially. If
we were to plot a logarithmic graph of Ids vs Vgs for Vgs less than Vt, it would be a
straight line. This is called the sub-threshold conduction. This effect is exacerbated
by the DIBL effect.
The region below Vt is called the sub threshold region. The graph shows the IV
characteristics for two different Vds: Vds=0.1 and Vds=1. Both the plots have non
zero current in the sub threshold region but Vds=1 plot has a higher current due to
the DIBL effect.
Gate leakage:
The gate and the channel form a capacitor. No current should pass through the
capacitor ideally. But with the decrease in the oxide thickness between the gate and
the channel, the electrons tunnel across and this constitutes the gate leakage
current.
Gate leakage current vs
VDD for different values of oxide thickness
From the above graph, we can see that the gate leakage increases by a factor of 2.7
or more per angstrom reduction in thickness. To keep the gate leakage below 100
A/cm2, the oxide thickness tox must be at least 10.5 A.
Gate leakage current can be reduced by substituting the silicon dioxide with higher
dielectric constant material.
2.5 DC Transfer Characteristics
The DC transfer characteristics of a circuit relate the output voltage to the input voltage,
assuming the input changes slowly enough that capacitances have plenty of time to charge or
discharge. Specific ranges of input and output voltages are defined as valid 0 and 1 logic levels.
Table 2.2, which outlines various regions of operation for the n- and p-transistors. In this table, Vtn is the
threshold voltage of the n-channel device, and Vtp is the threshold voltage of the p-channel device
.
The operation of the CMOS inverter can be divided into five regions indicated on Figure
2.26(c). The state of each transistor in each region is shown in Table 2.3. In region A,
the nMOS transistor is OFF so the pMOS transistor pulls the output to VDD. In region
B, the nMOS transistor starts to turn ON, pulling the output down. In region C, both
transistors are in saturation. Notice that ideal transistors are only in region C for Vin =
VDD/2 and that the slope of the transfer curve in this example is Vin is infinity in this
region, corresponding to infinite gain. Real transistors have finite output resistances on
account of channel length modulation, described in Section 2.4.2, and thus have finite
slopes over a broader region C. In region D, the pMOS transistor is partially ON in
region E.
V V V
Vinv
(2.55)
In the limit that the transistors are fully velocity saturated, EQ (2.29) shows
Idn WnCoxvsatn (Vinv Vtn )
Idp WpCoxvsat p (Vinv VDD Vtp )
transistors can be ignored. If several parallel transistors are ON, the collection can be
viewed as a single transistor of greater width.
where
(2.59)
VIH minimum HIGH input voltage
VIL maximum LOW input voltage
VOH minimum HIGH output voltage
VOL maximum LOW output voltage
λ
Module 2 1st chapter
FABRICATION
Transistors are fabricated on thin silicon wafers that serve as both a mechanical support and an
electrical common point called the substrate. We can examine the physical layout of transistors from two
perspectives. One is the top view, obtained by looking down on a wafer. The other is the cross-
section, obtained by slicing the wafer through the middle of a transistor and looking at it edgewise.
We begin by looking at the cross-section of a complete CMOS inverter. We then look at the top view
of the same inverter and define a set of masks used to manufacture the different parts of the inverter.
The size of the transistors and wires is set by the mask dimensions and is limited by the resolution of
the manufacturing process. Continual advancements in this resolution have fuelled the exponential
growth of the semiconductor industry.
The inverter could be defined by a hypothetical set of six masks: n-well, polysilicon, n+ diffusion, p+
diffusion, contacts, and metal. Masks specify where the components will be manufactured on the
chip. Figure 2.3(a) shows a top view of the six masks. The cross-section of the inverter from Figure
2.1 was taken along the dashed line.
1
Fig 2.2 Inverter mask set.
The wafer is first oxidized in a high-temperature (typically 900–1200 °C) furnace that causes Si and O2
to react and become SiO2 on the wafer surface as shown in Figure 2.3(b).
The oxide must be patterned to define the n-well. Figure 2.3(c) shows an organic photoresist that
softens when exposed to light is spun onto the wafer.
Figure 2.3(d) shows, the photoresist is exposed through the n-well mask that allows light to pass
through only where the well should be.
The oxide is etched with hydrofluoric acid (HF) where it is not protected by the photoresist is as
shown in Figure 2.3(e).
Figure 2.3(f) shows the remaining photoresist is stripped away using a mixture of acids called piranha
2
etch. The well is formed where the substrate is not covered with oxide.
Two ways to add dopants are diffusion and ion implantation. In the diffusion process, the wafer is
placed in a furnace with a gas containing the dopants. When heated, dopant atoms diffuse into the
substrate. Notice how the well is wider than the hole in the oxide on account of lateral diffusion
Figure 2.3(g). With ion implantation, dopant ions are accelerated through an electric field and blasted
into the substrate. In either method, the oxide layer prevents dopant atoms from entering the
substrate where no well is intended. Finally, the remaining oxide is stripped with HF to leave the bare
wafer with wells in the appropriate places.
3
The transistor gates are formed next. These consist of polycrystalline silicon, generally called
polysilicon, over a thin layer of oxide. The thin oxide is grown in a furnace. Then the wafer is placed in
a reactor with silane gas (SiH4) and heated again to grow the polysilicon layer through a process
called chemical vapor deposition. The polysilicon is heavily doped to form a reasonably good conductor.
The resulting cross-section is shown in Figure 2.4(a).
Figure 2.4(b) shows the wafer which is patterned with photoresist and the polysilicon mask leaving
the polysilicon gates atop the thin gate oxide. The n+ regions are introduced for the transistor active
area and the well contact.
Figure 2.4(c) shows the well, a protective layer of oxide is formed and patterned with the n-diffusion
mask.
Figure 2.4(d) shows the areas exposed where the dopants are needed.
Although the n+ regions in Figure 2.4(e) are typically formed with ion implantation, they were
historically diffused and thus still are often called n-diffusion. Notice that the polysilicon gate over the
nMOS transistor blocks the diffusion so the source and drain are separated by a channel under the
gate. This is called a self-aligned process because the source and drain of the transistor are automatically
formed adjacent to the gate without the need to precisely align the masks.
Finally, the protective oxide is stripped as shown in Figure 2.4(f).
F
i
g
2.5 Cross-sections while manufacturing p-diffusion, contacts, and metal
Layout design rules describe how small features can be and how closely they can be reliably packed in
a particular manufacturing process. Industrial design rules are usually specified in microns.
Universities sometimes simplify design by using scalable design rules that are conservative enough to
apply to many manufacturing processes. Mead and Conway popularized scalable design rules basedon
a single parameter, λ, that characterizes the resolution of the process. λ is generally half of the
minimum drawn transistor channel length. This length is the distance between the source and drain
of a transistor and is set by the minimum width of a polysilicon wire.
For example, a 180 nm process has a minimum polysilicon width (and hence transistor length) of
0.18 nm and uses design rules with λ = 0.09 nm3. Lambda-based rules are necessarily conservative
because they round up dimensions to an integer multiple of λ. Designers often describe a process
byits feature size. Feature size refers to minimum transistor length, so λ is half the feature size.
5
2.3.1 Lambda Based Rules
A conservative but easy-to-use set of design rules for layouts with two metal layers in an n-well
process is as follows:
Metal and diffusion have minimum width and spacing of 4 λ
Contacts are 2 λ × 2 λ and must be surrounded by 1 λ on the layers above and below.
Polysilicon uses a width of 2 λ.
Polysilicon overlaps diffusion by 2 λ where a transistor is desired and has a spacing of 1 λ
away where no transistor is desired.
Polysilicon and contacts have a spacing of 3 λ from other polysilicon or contacts.
N-well surrounds pMOS transistors by 6 λ and avoids nMOS transistors by 6 λ.
7
Fig 2.7 Inverter cell layout Fig 2.8 3-input NAND standard cell gate layouts
8
CMOS Inverter
NOTE
Red: Polysilicon
Green: n-diffusion
Yellow: p-diffusion
Blue: M1
Purple (Magenta):M2
Blue (Cyan):M3
Black: Contacts & Taps
9
CMOS NOR Gate
Track Pitch:
Transistors are merely widgets that fit under the wires. We define a routing track as enough space to
place a wire and the required spacing to the next wire. If our wires have a width of 4 λ and a
spacing of 4 λ to the next wire, the track pitch is 8 λ. Therefore, it is reasonable to estimate the
height and width of a cell by counting the number of metal tracks and multiplying by 8 λ
10
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Problem 1: Sketch a stick diagram for a 3 input CMOS NAND gate ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ and estimate the ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ cell width
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
and ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
height. ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Fig 2.14 Stick diagram of 3 input ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ CMOS NAND gate and ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ area estimation̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Problem 2: Sketch a stick diagram for a CMOS̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ gate computing Y= ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ (𝐴 + 𝐵 + 𝐶). 𝐷 ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
and estimate the cell width and height. ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Fig 2.15 CMOS compound gate Function Y= (𝐴 and+area 𝐶). 𝐷 ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
𝐵 +estimation
Wafer Formation
The basic raw material used in CMOS fabs is a wafer or disk of silicon, roughly 75 mm to 300 mm in
diameter and less than 1 mm thick. Wafers are cut from boules, cylindrical ingots of single-crystal
silicon, that have been pulled from a crucible of pure molten silicon. This is known as the Czochralski
method. Controlled amounts of impurities are added to the melt to provide the crystal with the
11
required electrical properties. A seed crystal is dipped into the melt to initiate crystal growth. The
silicon ingot takes on the same crystal orientation as the seed. A graphite radiator heated by radio-
frequency induction surrounds the quartz crucible and maintains the temperature a few degrees
above the melting point of silicon (1425 °C). The atmosphere is typically helium or argon to prevent
the silicon from oxidizing.
The seed is gradually withdrawn vertically from the melt while simultaneously being rotated, as
shown in Figure 2.17. The molten silicon attaches itself to the seed and recrystallizes as it is
withdrawn. The seed withdrawal and rotation rates determine the diameter of the ingot. Growth rates
vary from 30 to 180 mm/hour.
Photolithography
In places covered by the mask, ion implantation might not occur or the dielectric or metal layer might be
left intact. In areas where the mask is absent, the implantation can occur, or dielectric or metal could
be etched away. The patterning is achieved by a process called photolithography. The primary method
for defining areas of interest (i.e., where we want material to be present or absent) on a waferis by the
use of photoresists. The wafer is coated with the photoresist and subjected to selective illumination
through the photomask.
A photomask is constructed with chromium (chrome) covered quartz glass. A UV light source is
used to expose the photoresist. Figure 2.18 illustrates the lithography process. The photomask has
chrome where light should be blocked. The UV light floods the mask from the backside and passes
through the clear sections of the mask to expose the organic photoresist (PR) that has been coated
on the wafer. A developer solvent is then used to dissolve the soluble unexposed photoresist, leaving
islands of insoluble exposed photoresist.This is termed a negative photoresist. A positive resist is
initially insoluble, and when exposed to UV becomes soluble. Positive resists provide for higher
resolution than negative resists, but are less sensitive to light. As feature sizes become smaller, the
photoresist layers have to be made thinner.
12
Fig 2.18 Photomasking with a negative resist (lens system between mask and wafer omitted to
improve clarity and avoid diffracting the reader)
Well and Channel Formation
The following are main CMOS technologies:
n-well process
p-well process
twin-well process
triple-well process
In a p-well process, the nMOS transistors are built in a p-well and the pMOS transistor is placed in
the n-type substrate. p-well processes were used to optimize the pMOS transistor performance.
In the n-well process, pMOS transistors are fabricated in an n-well, each group of pMOS transistors
in an n-well shares the same body node but is isolated from the bodies of pMOS transistors in
different wells. However, all the nMOS transistors on the chip share the same body, which is the
substrate. Noise injected into the substrate by digital circuits can disturb sensitive analog or memory
circuits.
Twin-well processes accompanied the emergence of n-well processes. A twin-well process allows the
optimization of each transistor type.
A third well can be added to create a triple-well process. The triple-well process has emerged to
provide good isolation between analog and digital blocks in mixed-signal chips; it is also used to
isolate high-density dynamic memory from logic.
Isolation
Individual devices in a CMOS process need to be isolated from one another so that they do not have
unexpected interactions. The transistor gate consists of a thin gate oxide layer. Elsewhere, a thicker
layer of field oxide separates polysilicon and metal wires from the substrate. The MOS sandwich
formed by the wire, thick oxide, and substrate behaves as an unwanted parasitic transistor. However,
the thick oxide effectively sets a threshold voltage greater than VDD that prevents the transistor
from turning ON during normal operation. Actually, these field devices can be used for I/O protection.
The thick oxide used to be formed by a process called Local Oxidation of Silicon (LOCOS). A
problem with LOCOS-based processes is the transition between thick and thin oxide, which
extended some distance laterally to form a so-called bird’s beak. The lateral distance is proportional to
the oxide thickness, which limits the packing density of transistors.
STI starts with a pad oxide and a silicon nitride layer, which act as the masking layers, as shown
in Figure 2.19. Openings in the pad oxide are then used to etch into the well or substrate region (this
process can also be used for source/drain diffusion). A liner oxide is then grown to cover the
exposed silicon (Figure 2.19(b)). The trenches are filled with SiO2 or other fillers using CVD that
does not consume the underlying silicon (Figure 2.19(c)). The pad oxide and nitride are removed and
a Chemical Mechanical Polishing (CMP) step is used to planarize the structure (Figure 2.19(d)).
14
Fig 2.19 Shallow trench isolation
Gate Oxide
The next step in the process is to form the gate oxide for the transistors. As mentioned, this is most
commonly in the form of silicon dioxide (SiO2).
In the case of STI-defined source/drain regions, the gate oxide is grown on top of the planarized
structure that occurs at the stage shown in Figure 2.19(d). This is shown in Figure 2.20.
18
2.7 Layout Design Rules
Layout rules, also referred to as design rules, and can be considered a prescription for preparing the
photomasks that are used in the fabrication of integrated circuits. The rules are defined in terms of
feature sizes (widths), separations, and overlaps. The main objective of the layout rules is to build reliably
functional circuits in as small an area as possible. In general, design rules represent a compromise
between performance and yield. The more conservative the rules are, the more likely it is that the
circuit will function. However, the more aggressive the rules are, the greater the opportunity for
improvements in circuit performance and size.
Well Rules
The n-well is usually a deeper implant (especially a deep n-well) than the transistor source/drain
implants, and therefore, it is necessary to provide sufficient clearance between the n-well edges and
the adjacent n+ diffusions. The clearance between the well edge and an enclosed diffusion is
determined by the transition of the field oxide across the well boundary. Processes that use STI may
permit zero inside clearance.
Transistor Rules
CMOS transistors are generally defined by at least four physical masks. These are active (also called
diffusion, diff, thinox, OD, or RX), n-select (also called n-implant, nimp, or nplus), p-select (also
called p-implant, pimp, or pplus) and polysilicon (also called poly, polyg, PO, or PC). The active mask
defines all areas where either n- or p type diffusion is to be placed or where the gates of transistors are to
be placed. The gates of transistors are defined by the logical AND of the polysilicon mask and the
active mask, i.e., where polysilicon crosses diffusion. The select layers define what type of diffusion is
required. n-select surrounds active regions where n-type diffusion is required. p-select surrounds
areas where p-type diffusion is required. n-diffusion areas inside p-well regions define nMOS
transistors (or n-diffusion wires). n-diffusion areas inside n-well regions define n-well contacts.
Likewise, p-diffusion areas inside n-wells define pMOS transistors (or p-diffusion wires). p-diffusion
areas inside p-wells define substrate contacts (or p-well contacts).
Figure 2.23(a) shows the mask construction for the final structures that appear in Figure 2.23(b).
19
Fig 2.23 CMOS n-well process transistor and well/substrate contact construction
Contact Rules
There are several generally available contacts:
Metal to p-active (p-diffusion)
Metal to n-active (n-diffusion)
Metal to polysilicon
Metal to well or substrate
Depending on the process, other contacts such as buried polysilicon-active contacts may be allowedfor
local interconnect.
Because the substrate is divided into well regions, each isolated well must be tied to the appropriate
supply voltage; i.e., the n-well must be tied to VDD and the substrate or p-well must be tied to GND
with well or substrate contacts.
Mask Summary: The only mask involved with contacts to active or poly is the contact mask,
commonly called CONT or CA. Contacts are normally of uniform size to allow for consistent
etching of very small features.
Metal Rules
Metal spacing may vary with the width of the metal line (so called fat-metal rules). That is, above
some metal wire width, the minimum spacing may be increased. This is due to etch characteristics of
small versus large metal wires. There may also be maximum metal width rules. That is, single metal
wires cannot be greater than a certain width. If wider wires are desired, they are constructed by
paralleling a number of smaller wires and adding checkerboard links to tie the wires together.
Mask Summary: Metal rules may be complicated by varying spacing dependent on width: As the
width increases, the spacing increases. Metal overlap over contact might be zero or nonzero.
20
Via Rules
Processes may vary in whether they allow stacked vias to be placed over polysilicon and diffusion
regions. Some processes allow vias to be placed within these areas, but do not allow the vias to
straddle the boundary of polysilicon or diffusion.
Mask Summary: Vias are normally of uniform size within a layer. They may increase in size toward
the top of a metal stack. For instance, large vias required on power busses are constructed from an
array of uniformly sized vias.
Other Rules
The passivation or overglass layer is a protective layer of SiO2 (glass) that covers the final chip.
Appropriately sized openings are required at pads and any internal test points.
Some additional rules that might be present in some processes are as follows:
Extension of polysilicon or metal beyond a contact or via
Differing gate poly extensions depending on the device length
21
Micron Design Rules
Table 2.1 lists a set of micron design rules for a hypothetical 65 nm process representing an
amalgamation of several real processes. Observe that the rules differ slightly but not immensely from
lambda-based rules with = 0.035 µm. A complete set of micron design rules in this generation fills
hundreds of pages. Note that upper-level metal rules are highly variable depending on the metal
thickness; thicker wires require greater widths and spacings and bigger vias.
The two most common metrics for a good chip are speed and power,. Delay and
power are influenced as much by the wires as by the transistors, so Chapter 6
delves into interconnect analysis and design.
A timing analyzer computes the arrival times, i.e., the latest time
at which each node in a block of logic will switch. The nodes are classified as inputs,
outputs, and internal nodes. The user must specify the arrival time of inputs and the time
data is required at the outputs. The arrival time ai at internal node i depends on the
propagation delay of the gate driving i and the arrival times of the inputs to the gate:
4.2 Transient Response 3
Figure 4.2 shows nodes annotated with arrival times. If the outputs are all required at 200 ps,
the circuit has 60 ps of slack.
.
RC Delay Model:
RC delay models approximate the nonlinear transistor I-V and C-V characteristics
with an average resistance and capacitance over the switching range of the gate.
Effective Resistance:
4.2 Transient Response 4
The RC delay model treats a transistor as a switch in series with a resistor. The
effective resistance is the ratio of Vds to Ids averaged across the switching interval of
interest.
A unit nMOS transistor is defined to have effective resistance R. The size of the unit
transistor is arbitrary but conventionally refers to a transistor with minimum length
and minimum contacted diffusion width (W/L ratio).. A unit pMOS transistor has
greater resistance, generally in the range of 2R–3R, because of its lower mobility.
4.2 Transient Response 5
4.2 Transient Response 6
Module -3 Semiconductor Memories
• Objectives:
Introduction to Semiconductor Memories
Dynamic Random Access Memory (DRAM)
Static Random Access Memory (SRAM)
Nonvolatile Memory (ROM)
Flash Memory
Ferroelectric Random Access Memory(FRAM)
Introduction
• Semiconductor memory arrays capable of storing large quantities of
digital information are essential to all digital systems.
• The amount of memory required in a particular system depends on
the type of application, but, in general, the number of transistors
utilized for the information (data) storage function is much larger
than the number of transistors used in logic operations and for other
purposes.
• The area efficiency of the memory array, i.e., the number of stored
data bits per unit area, is one of the key design criteria that determine
the overall storage capacity and, hence, the memory cost per bit.
• Another important issue is the memory access time, i.e., the time
required to store and/or retrieve a particular data bit in the memory
array
• The access time determines the memory speed, which is an important
performance criterion of the memory array.
• Finally, the static and dynamic power consumption of the memory
array is a significant factor to be considered in the design, because of
the increasing importance of low-power applications.
• Memory circuit s are generally classified according to the type of data
storage and the type of data access.
Overview of semiconductor memory types
ROM
• Read-Only Memory (ROM) circuits allow, as the name implies, only
the retrieval of previously stored data and do not permit
modifications of the stored information contents during normal
operation.
• ROMs are non-volatile memories, i.e., the data storage function is not
lost even when the power supply voltage is off.
• Depending on the type of data storage (data write) method, ROMs
are classified as mask-programmed ROMs, Programmable ROMs
(PROM), Erasable PROMs (EPROM), and Electrically Erasable PROMs
(EEPROM)
RAM
• Read-write (R/W) memory circuits, on the other hand, must permit
the modification (writing) of data bits stored in the memory array, as
well as their retrieval (reading) on demand.
• This requires that the data storage function be volatile, i.e., the stored
data are lost when the power supply voltage is turned off.
• The read-write memory circuit is commonly called Random Access
Memory (RAM),
• Based on the operation type of individual data storage cells, RAMs are
classified into two main categories:
Static RAMs (SRAM) and Dynamic RAMs (DRAM).
A typical memory array organization is shown in Figure
• The data storage structure, consists of individual memory cells arranged in
an array of horizontal rows and vertical columns.
• Each cell is capable of storing one bit of binary information.
• Also, each memory cell shares a common connection with the other cells in
the same row, and another common connection with the other cells in the
same column.
• In this structure, there are 2𝑁 rows, also called word lines, and 2𝑀 columns,
also called bit lines. Thus, the total number of memory cells in this array is
2𝑁 x 2𝑀 .
• To access a particular memory cell, i.e., a particular data bit in this array,
the corresponding bit line and the corresponding word line must be
activated (selected).
• The row and column selection operations are accomplished by row and
column decoders, respectively.
• The row decoder circuit selects one out of 2𝑁 word lines according to an N-
bit row address,
• while the column decoder circuit selects one out of 2𝑀 bit lines according
to an M-bit column address.
• Once a memory cell or a group of memory cells are selected in this
fashion, a data read and/or a data write operation may be performed
on the selected single bit or multiple bits on a particular row.
• The column decoder circuit serves the double duties of selecting the
particular columns and routing the corresponding data content in a
selected row to the output.
• We can see from this simple discussion that individual memory cells
can be accessed for data read and/or data write operations in random
order, independent of their physical locations in the memory array.
• Thus, the array organization examined here is called a Random
Access Memory (RAM) structure.
• Notice that this organization can be used for both read-write memory
arrays and read-only memory arrays.
Dynamic Read-Write Memory (DRAM) Circuits
• As the trend for high-density RAM arrays forces the memory cell size
to shrink, alternative data storage concepts must be considered to
accommodate these demands.
• In a dynamic RAM cell, binary data is stored simply as charge in a
capacitor, where the presence or absence of stored charge
determines the value of the stored bit.
• Note that the data stored as charge in a capacitor cannot be retained
indefinitely, because the leakage currents eventually remove or
modify the stored charge.
• Thus, all dynamic memory cells require a periodic refreshing of the
stored data, so that unwanted modifications due to leakage are
prevented before they occur.
• The use of a capacitor as the primary storage device generally enables
the DRAM to be realized on a much smaller silicon area compared to
the typical SRAM cell.
• Notice that even as the binary data is stored as charge in a capacitor,
the DRAM cell must have access devices, or switches, which can be
activated externally for "read" and "write“ operations.
• Note that a DRAM array also requires additional peripheral circuitry
for scheduling and performing the periodic data refresh operations.
• Following circuits shows some of the steps in the historical evolution
of the DRAM cell.
The four-transistor cell
• 4T DRAM is the simplest and one of the earliest dynamic memory
cells.
• This cell is derived from the six-transistor static RAM cell by removing
the load devices.
• The cell has in fact two storage nodes, i.e., the parasitic oxide and
diffusion capacitances of the nodes indicated in the circuit diagram.
• Since no current path is provided to the storage nodes for restoring
the charge being lost to leakage, the cell must be refreshed
periodically.
The three-transistor DRAM cell
• 3T DRAM cell is the first widely used dynamic memory cell.
• It utilizes a single transistor as the storage device (where the
transistor is turned on or off depending on the charge stored on its
gate capacitance)
• And one transistor each for the "read" and "write" access switches.
• The cell has two control and two I/O lines. Its separate read and write
select lines make it relatively fast, but the four lines with their
additional contacts tend to increase the cell area.
The one-transistor DRAM cell
• 1T DRAM cell has become the industry standard dynamic RAM cell in high-
density DRAM arrays. With only one transistor and one capacitor, it has the
lowest component count and, hence, the smallest silicon area of all the
dynamic memory cells.
• The cell has one read-write control line (word line) and one I/O line (bit
line).
• unlike in the other dynamic memory cells, the storage capacitance of the
one transistor DRAM cell is explicit.
• This means that a separate capacitor must be manufactured for each
storage cell, instead of relying on the parasitic oxide and diffusion
capacitances of the transistors for data storage.
• The word line of the one-transistor DRAM cell is controlled by the row
address decoder. Once the selected transistor is turned on, the charge
stored in the capacitor can be detected and/or modified through the bit
line.
Three-Transistor DRAM Cell Operation
• The circuit diagram of a typical three-transistor dynamic RAM cell is
shown in Fig. as well as the column pull-up (precharge) transistors
and the column read/write circuitry.
• Here, the binary information is stored in the form of charge in the
parasitic node capacitance C1. The storage transistor M2 is turned on
or off depending on the charge stored in C1, and the pass transistors
M1 and M3 act as access switches for data read and write operations.
• The cell has two separate bit lines for "data read" and "data write,"
and two separate word lines to control the access transistors.
• The operation of the three-transistor DRAM cell and its peripheral
circuitry is based on a two-phase non-overlapping clock scheme.
• The precharge events are driven by φ1,whereas the "read" and
"write" events are driven by φ2.
• Every "data read" and "data write“ operation is preceded by a
precharge cycle, which is initiated with the precharge signal PC going
high.
• During the precharge cycle, the column pull-up transistors are
activated, and the corresponding column capacitances C2 and C3 are
charged up to logic-high level.
• With typical enhancement type nMOS pull-up transistors (Vt= 1V) and
a power supply voltage of 5 V, the voltage level of both columns after
the precharge is approximately equal to 3.5 V.
• All "data read" and "data write" operations are performed during the
active φ2 phase, i.e., when PC is low.
Typical voltage waveforms associated with the 3-T DRAM cell during four consecutive operations:
write "1," read "1," write "0," and read “0."
Column capacitances C2 and C3 are being charged-up through MP1
and MP2 during the precharge cycle.
• For the write "1" operation, the inverse data input is at the logic-low
level, because the data to be written onto the DRAM cell is logic "1."
• Consequently, the "data write“ transistor MD is turned off, and the
voltage level on column Din remains high.
• Now, the"write select" signal WS is pulled high during the active
phase of φ2.
• As a result, the write access transistor M1 is turned on. With M1
conducting, the charge on C2 is now shared with C1.
• Since the capacitance C2 is very large compared to C1, the storage
node capacitance C1 attains approximately the same logic-high level
as the column capacitance C2 at the end of the charge-sharing
process.
Charge sharing between C2 and Cl during the write " 1 "
sequence.
• After the write "1" operation is completed, the write access transistor
M1 is turned off. With the storage capacitance C1 charged-up to a
logic-high level, transistor M2 is now conducting.
• In order to read this stored "1," the "read select" signal RS must be
pulled high during the active phase of φ2, following a precharge cycle.
• As the read access transistor M3 turns on, M2 and M3 create a
conducting path between the "data read“ column capacitance C3 and
the ground.
• The capacitance C3 discharges through M2 and M3, and the falling
column voltage is interpreted by the "data read" circuitry as a stored
logic "1."
• The active portion of the DRAM cell during the read "1" cycle is
shown in Figure below
The column capacitance C3 is discharged through the
transistors M2 and M3 during the read "1" operation
Note that the 3-T DRAM cell may be read repeatedly in this fashion without disturbing the charge
stored in C1.
• For the write "0" operation, the inverse data input is at the logic-high
level, because the data to be written onto the DRAM cell is a logic "0.“
• Consequently, the data write transistor is turned on, and the voltage
level on column Din is pulled to logic "0."
• Now, the "write select" signal WS is pulled high during the active
phase of φ2.
• As a result, the write access transistor M1 is turned on.
• The voltage level on C2, as well as that on the storage node C1, is
pulled to logic "0" through M1 and the data write transistor, as shown
in Figure below.
Both C1 and C2 are discharged via M1 and the data
write transistor during the write "0" sequence.
• Thus, at the end of the write "0" sequence, the storage capacitance
C1 contains a very low charge, and the transistor M2 is turned off
since its gate voltage is approximately equal to zero.
• In order to read this stored "0," the "read select" signal RS must be
pulled high during the active phase of φ2, following a precharge cycle.
• The read access transistor M3 turns on, but since M2 is off, there is
no conducting path between the column capacitance C3 and the
ground (as shown in Fig. below)
• Consequently, C3 does not discharge, and the logic-high level on the
Dout column is interpreted by the data read circuitry as a stored "0"
bit.
The column capacitance C3 cannot discharge during the read
"0" cycle
Operation of One-Transistor DRAM Cell
• The circuit diagram of the one-transistor (1-T) DRAM cell consisting of
one explicit storage capacitor and one access transistor is shown in
Fig. above
• Here, C1 represents the storage capacitor and Capacitor C2
represents the much larger parasitic column capacitance associated
with the word line.
• Charge sharing between this large capacitance and the very small
storage capacitance plays a very important role in the operation of
the 1-T DRAM cell.
• For the write "1"operation, the bit line (D) is raised to logic " 1 " by
the write circuitry, while the selected word line is pulled high by the
row address decoder.
• The access transistor M1 turns on, allowing the storage capacitor C1
to charge up to a logic-high level.
• For the write “0“ operation, the bit line (D) is pulled to logic “0" and
the word line is pulled high by the row address decoder.
• In this case, the storage capacitor C1 discharges through the access
transistor, resulting in a stored “0" bit.
• In order to read stored data out of a 1-T DRAM cell, on the other hand, we
have to build a fairly elaborate read-refresh circuit.
• The reason for this is the fact that the "data read" operation on the one-
transistor DRAM cell is by necessity a "destructive readout."
• This means that the stored data must be destroyed or lost during the read
operation.
• Typically, the read operation starts with precharging the column
capacitance C2.
• Then, the word line is pulled high in order to activate the access transistor
M1.
• Charge sharing between C1 and C2 occurs and, depending on the amount
of stored charge on C1, the column voltage either increases or decreases
slightly.
• Note that charge sharing inevitably destroys the stored charge on C1.
Hence, we also have to refresh data every time we perform a "data read"
operation.
1T DRAM cell array with control circuits
• A typical memory structure consisting of one transistor DRAM cell
array and the control circuits is shown in the above figure
• In recent DRAM architectures, the bit lines are folded and precharged
to ½ VDD to improve noise immunity and reduce power consumption
• In addition, the amplifier to sense the perturbed signal at the bit line
is shared by adjacent blocks
• Bit lines (BL and BLB) and sensing nodes (SA and SAB) are set to a
precharge level, half-VDD through bit and sensing line
equalizers,respectively
• Timing diagram of DRAM read operation for data “ 1” is shown,
• Before the read operation,the signals to precharge the bit lines(PEQ)
and bit line sense amplifier (PSAEQ) are disabled.
• The bit line amplifier is shared by two adjacent memory array blocks
to reduce the chip size.
• Thus, the memory array select signal (PISOi) is set to boosted
voltage(Vpp). Vpp is on chip boosted voltage
• A word line in the memory array block is selected according to the
row address
• When the word line is enabled,the charge at the cell capacitor Cs is
shared with bit line capacitance.since the bit line is precharged to ½
VDD a small voltage differenceis developed at the bit line
• The control signals(PSA and PSAB) are activated sequentially to
reduce the charge injection and short circuit current due to
simultaneous activation of N and P latches in the bit line sense
amplifiers
• Hence the BLB node is discharged to the ground and signal is
amplified
• The level of BLB and BL nodes eventually reaches the ground and the
operating voltage respectively and the voltage of storage node is
recovered, this is called restoring operation of the cell data
• The data on the bit lines are read by transferring the voltage levels to
the secondary data lines BL_IO and BL_IOB.
• After the voltage difference on the bit lines are amplified,the column
switch transistors are enabled by activating a column select line by
the column decoder.
• To sense the small signal difference, a latch amplifier is used.
• The sense nodes (SA and SAB) are connected together in the same
block and the source transistors(MNS and MPS) are placed per
memory array segment
• Typically, BL_IO and BL_IOB lines are precharged at VDD or VDD-Vtn,
since BLB is discharged to the ground the level of BL_IOB is slowly
discharged due to its large capacitance e,g., 10 times CBL through
MNC2.
• The level of BLB is affected slightly by the BL_IOB line. The voltage
difference on secondary data line is amplified by the read amplifier to
the full CMOS output level and transferred to the Memory interface
circuit drive the off chip load.
Static Read-Write Memory (SRAM) Circuits
INTRODUCTION:
A failure is said to have occurred in a circuit or system if it deviates from its specified
behavior.A fault, on the other hand, is a physical defect that may or may not cause a
failure.
A fault is characterized by its nature, value, extent, and duration. The nature of a fault
can be classified as logical or non logical. A logical fault causes the logic value at a point
in a circuit to become opposite to the specified value.Non logical faults include the rest
of the faults, such as the malfunction of the clock signal, power failure, and so forth.
The value of a logical fault at a point in the circuit indicates whether the fault creates
fixed or varying erroneous logical values.The extent of a fault specifies whether the
effect of the fault is localized or distributed.
A local fault affects only a single variable, whereas a distributed fault affects more than
on
When testing digital circuits, two distinct philosophies are commonly used:
Terminology:
Faults in a circuit may occur due to defective components, breaks in signal lines, lines
shortened to ground or power supply, short-circuiting of signal lines, ex- cessive delays, and
so forth.Besides errors or ambiguities in design specifications, design rule violations, among
other things, also result in faults.
Poor designs may also result in hazards, races, or metastable flip-flop behavior in a circuit;
such faults manifest themselves as "intermittents" through- out the life of the circuit.
In general, the effect of a fault is represented by means of a model, which represents the
change the fault produces in circuit signals.
1. STUCK AT FAULTS:
Stuck-at Fault (SAF) is a type of fault analysis in which a node/pin in a digital circuit is
assumed to be fixed at logic 0 or logic 1.
Stuck-at fault model is also called a permanent fault model because the faulty effect is
assumed to be permanent.
Stuck-at fault occurs in logic gates which results in one of the inputs or the output being
fixed to either a logic 0 (stuck-at 0) or logic 1 (stuck-at 1).
LOGIC 1 ( STUCK AT 1)
EXPLAINATION:
The most common model used for logical faults is the single stuck-at fault. It assumes
that a fault in a logic gate results in one of its inputs or the output being fixed to either a
logic 0 (stuck-at-0) or a logic 1 (stuck-at-1).
Stuck-at-0 and stuck- at-1 faults are often abbreviated to s-a-0 and s-a-1, respectively,
and these abbreviations will be adopted here.
Let us consider a NAND gate with input A s-a-1. The NAND gate perceives the input A
as a 1 irrespective of the logic value placed on the input.
The output of the NAND gate in is 0 for the input pattern shown above, when the s-a-1
fault is present. The fault-free gate has an output of 1. Therefore, the pattern shown in
can be used as a test for the A input s-a-1, because there is a difference between the
output of the fault-free and the faulty gate.
The stuck-at model, often referred to as classical fault model, offers good representation
for the most common types of failures, for example, short-circuits (shorts) and open
circuits (opens) in many technologies.
Unintended shorts between the lines form a class of permanent faults, known as bridging
faults.
It has been observed that physical defects in MOS (Metal Oxide Semiconductor) circuits
are manifested as bridging faults more than as any other type of fault.
An input bridging fault corresponds to the shorting of a certain number of primary input
lines.
Feedback bridging fault occurs if there is a short between an output line to an input line.
A non feedback bridging fault identifies a bridging fault that does not belong to either of
the two previous categories.
From these definitions, it will be clear that the probability of two lines getting bridged is
higher if they are physically close to each other.
The presence of a feedback bridging fault can cause a circuit to oscillate or convert it into
a sequential circuit.
The probable bridging fault in this circuit (or in any other CMOS circuit) can be grouped
into four categories:
On the other hand, if two transistors with floating gates are permanently conducting, one
of them can be considered as stuck on.
If a transistor with a floating gate remains in a nonconducting state due to a signal line
break, the circuit will behave in a similar fashion as it does in the presence of the
integrate break b₂.
3.STUCK-ON AND STUCK-OPEN FAULTS
STUCK-ON FAULTS:
The faulty gate output is difficult to predict because the stuck-on transistor competes
with its complementary transistors for control of the output.
Sometimes this competition doesn’t lead to catastrophic failure, but it can disrupt normal
circuit behavior.
STUCK-OPEN FAULTS
The faulty gate output remains open (disconnected) even when it should be driving a
logic value.
Stuck-open faults are modeled as additional non-classical faults alongside the standard
stuck-at faults.
4.DELAY FAULTS
Delay faults occur when a circuit is too slow to propagate signals through certain paths or
gates.
Even if the logic operations within the circuit are correct, the output may not be reached
within the allotted time due to delays.
Types:
Gate Delay Faults (GDF)
Path Delay Faults (PDF)
Transition Delay Faults
Transition delay faults occur during signal transitions (rising or falling edges).
These faults affect the timing between input and output transitions.
Detecting and mitigating transition delay faults is essential for maintaining correct circuit
behavior.
5.TEMPORARY FAULTS:
● Temporary faults are the faults or errors that occur unpredictably and temporarily affect
the VLSI or electronic circuits.
● Temporary faults are often referred to as intermittent or transient faults with the same
meaning.
● Transient faults are non-recurring temporary faults.
● They are not repairable because there is no physical damage to the hardware.
VTU QUESTIONS
MODULE 4 VDT
Module 4
Part 2
1
MODULE 4 VDT
When all test patterns, done detected faults F used to compute fault coverage
(Fc).
Fc=F / x
2
MODULE 4 VDT
• The path is said to be sentisized if we apply inputs to the gates along that
particular path under consideration.
Example 1:
STEP 1:
Set X5=G2=1
X3=0
3
MODULE 4 VDT
There are two ways of propagation from G5 that is G7G9 and G8G9
Set G4 = G8 =1
Similarly if G8=1 then the ckt output depends on G7 only. This is called
forward trace.
In backward trace the gate conditions to propagate the fault along the
sentisised path is established.
To set G4=1,
X1=G1=1,
To set G2=1
X4=0
To set G8 to be 1
G6=0
NOTE:
2. If the output of the circuit is 0 then it is fault free circuit, if the output
4
MODULE 4 VDT
Example 2
STEP 1: X2=X3=0,
Then G2=1
STEP 2:X4=0
Then G6=0
STEP 3:
Set G4=G5=G7=0
X4=X2=0
5
MODULE 4 VDT
Then G3 =1
Makes G4=1
BOOLEAN DIFFERENCE
The basic principle of the Boolean Difference is to derive two Boolean expressions- one of
which represents normal fault-free behaviour of the circuit, and the other, the logical
behaviour under assumed single s-a-1 or s-a-0 fault condition.
Let F(X) F(x1,....,xn) be a logic function of a variables. If one of the inputs to the logic
function, for example, input x, is faulty, then the output would be F(x1,...,xi,...,xn). The
Boolean difference of F(X) with respect to xi is defined as
The function dF(X)/dxi is called the Boolean difference of F(X) with respect to xi.
6
MODULE 4 VDT
7
MODULE 4 VDT
8
MODULE 4 VDT
9
MODULE 4 VDT
10
MODULE 4 VDT
D-A LGORITHM
The D-algorithm is the first algorithmic method for generating tests
for nonre- dundant combinational circuits [2.7]. If a test exists for
detecting a fault, the D-algorithm is guaranteed to find this test.
Before the D-algorithm can be dis- cussed in detail, certain new
terms must be defined.
SINGULAR COVER
The singular cover of a logic gate is basically a compact version of
the truth table. Figure 2.6 shows the singular cover for a two-input
NOR gate; Xs or blanks are used to denote that the position may be
either 0 or 1. Each row in the singular cover is termed a singular
cube. The singular cover of a network is just the set of singular
covers of each of its gates on separate rows in the table. This is
illustrated by the example in Fig. 2.7.
11
MODULE 4 VDT
PROPAGATION OF D CUBES
Thus, all Ds in a circuit imply the same value (0 or 1), and all Ds
will have the opposite value. The propagation D-cubes of a gate are
those that cause the output of the gate to depend only on one or
more of its specified inputs (and hence to propagate fault on these
inputs to the output). The propagation D-cubes for a two-input
NOR gate are
12
MODULE 4 VDT
The primitive D-cube of any fault in a gate can be obtained from the
singular covers of the normal and the faulty gates in the following manner:
13
MODULE 4 VDT
Form the singular covers of the fault-free and the faulty gate. Let alpha_{0}
and be sets of cubes in the singular covers of the fault-free gate the output
alpha_{1}
Test Generation for Combinational Logic Circuits coordinates of which are
0 and 1, respectively, and let beta_{0} and beta_{1} be the corresponding
sets in the singular covers of the faulty gate. 2. Intersect members of
alpha_{1} with members of beta_{0} and members of alpha_{0} with
members of The intersection rules are similar to those u beta_{1} gation
D-cubes.
D INTERSECTION
Finally, we need to consider the concept of D-intersection, which provides
the tool for building sensitized paths. This is first explained by a simple
example. Consider the simple circuit shown in Fig. 2.9. We attempt to
generate a test for the 2s - a + 0 fault, described by the D-cube of the fault:
14
MODULE 4 VDT
Problem 1:
15
MODULE 4 VDT
https://www.youtube.com/watch?v=mVoxhnDWM_w&t=180s
PROBLEM 2
Step 1:
16
MODULE 4 VDT
17
MODULE 4 VDT
Step 2:
Step 3:
Step 4:
18
MODULE 4 VDT
1. The signal line (on which a stuck-at fault is assumed to be present) has
the same logic value as the stuck-at value.
In the first case, the fault remains masked in the presence of the assigned
input values, whereas in the second case the input patter cannot be a test,
because D or D cannot be propagated to the output. Therefore, only when
none of the foregoing situations occurs is a test possible with the current
assignment of Pls
Box 5 If all primary inputs have been assigned values and a test patter is
still not found, it is checked whether an untried combination values at the
inputs might generate a test or not.
19
MODULE 4 VDT
It is clear from the preceding discussion that the decision tree is an ordered ' list
of nodes (Fig. 2.13) having the following features:
2. The ordering reflects the sequence in which the current assignments have been
made.
11
0
MODULE 4 VDT
20
MODULE 4 VDT
21
2
MODULE 4 VDT
21
3
MODULE 4 VDT
21
4
MODULE 4 VDT
21
5
MODULE 4 VDT
21
6
MODULE 4 VDT
For example, an x-input logic gate can have 2x + 2 possible faults; however,
for certain input faults, a gate output would be forced into a state that is
indistinguishable from one of the s-a-0/s-a-1 output faults. Thus, for an
AND gate any input s-a-0 fault is indistinguishable from the output s-a-0
fault, and for an OR gate any input s-a-1 fault is indistinguishable for the
output s-a-1 fault. Such faults are said to be equivalent. For a NAND
(NOR) gate, the set of input s-a-0 (s-a-1) faults and the set of outputs faults
s-a-1 (s-a-0) are equivalent. Thus, an x-input gate has to be tested for x + 2
logically distinct faults.
A systematic approach that reduces the number of faults that have to be
considered in test generation is the process of fault folding (2.21). The
central idea behind the process is to form fault equivalence classes for a
given circuit by folding faults toward the primary inputs. For non
reconvergent fan-out circuits, the folding operation produces a set of faults
on primary inputs, and this set test covers all faults in the circuit. For
reconvergent fan-out circuits, the set of faults at the primary inputs, fan-
out origins, and fan-out branches test cover all faults in the circuit.
Another approach that results in a significant reduction in the number of
faults to be tested uses the concept of prime faults (2.22). The set of prime
faults for a network can be generated by the following procedure: 1. Assign
a fault to every gate input line if that is a primary input line or a fan-out
branch line. The fault is s-a-1 for AND/NAND gate inputs, und 6-a-0 for
OR/NOR gate inputs. Treat an inverter as a single input NANI/ NOR gate
if its output is a primary output; otherwise, no fault value should be
assigned to an inverter input line.
2. Identify every gate that has faults assigned to all its input lines as a prime
gate. Assign a fault to the output line of every prime gate that does not fan
out. The fault is s-a-0 for AND/NOR gate outputs, and s-a-1 for OR/NAND
gate outputs.
21
7
MODULE 4 VDT
30
MODULE 4 VDT
1. For the circuit shown in Fig.2 using Boolean difference (i) detect
s@0 and s@1 at x2, (ii) determine partial Boolean difference for
x2-l-n-p-F.
3. Find the test pattern for line 6 s@0 for the circuit shown in Fig.3
using D Algorithm.
31
MODULE 4 VDT
32
Module -5 Part a
Test generation for sequential circuit
EX:1 Homing sequence
Distinguish sequence:
Distinguishing tree:
Sequence is “ 10”
Module-05 (part 2)
Design of Testable Sequential Circuits
TOPICS
1. Controllability and Observability
2. Adhoc Design Rules
3. Diagnosable Sequential Circuits
4. Scan Path Technique
5. LSSD
6. Random Access Scan
7. Partial Scan
Module-05 (part 2)
Design of Testable Sequential Circuits
There are two key concepts in designing for testability: controllability and observability.
Controllability refers to the ability to apply test patterns to the inputs of a subcircuit
via the primary inputs of the circuit. For example, in Fig. 5.1(a) if the output of the
equality checker circuit is always in the state of equal, it is not possible to test whether
the equality checker is operating correctly or not. If a control gate is added to the circuit
(Fig. 5.1(b)), the input of the equality checker and hence, the operation of the circuit
can be controlled. Therefore, to enhance the controllability of a circuit, the state that
cannot be controlled from its primary inputs has to be reduced.
Observability refers to the ability to observe the response of a subcircuit via the
primary outputs of the circuit or at some other output points. For example, in Fig. 5.2
the outputs of all three AND gates are connected to the inputs of the OR gate. A stuck-
at-0 fault at the output of the AND gate 3 is not detectable because the effect of the fault
is masked and cannot be observed at the primary output. To enhance the observability,
we must observe the output of the gate Separately as shown.
Ad hoc rules are used to improve testability of specific circuits. One of the simplest ways of
achieving this is to incorporate additional control and observation points in a circuit. For
example, the fault o stuck-at-1 in the circuit of Fig. 5,3(a)is undetectable at the circuit output.
The addition of an extra-output line in the circuit makes the fault detectable (Fig, 5.3(b)).
The usefulness of inserting a control point can be understood from the circuit shown in Fig.
5.4(a). The output of the NOR gate is always 0; therefore, it is not possible to determine whether
the gate is functioning correctly or not. If a control point is added to the circuit as shown in Fig.
5.4(b), the NOR gate can be easily tested for single stuck-at fault.
A different way of achieving access to internal points is to use tristate drivers as shown in Fig.
5.6. A test mode signal could be used to put the driver into the high impedance state. In this
mode, the internal point could be used as a control point. When the driver is activated, the
internal point becomes a test point.
Another approach to improve, testability is to permit access to a subset of the logic as shown
in Fig. 5.7 [5.1,5.2]. Module 8 is physically embedded between the two modules A and C. A
set of gates G and H is inserted into each of the inputs and outputs, respectively, of module B.
In normal operation, the test control signal is such that modules A, B, and C are connected and
the complete network performs its desired function. In the test mode, the test control input is
changed;
module B is connected to the primary inputs and outputs of the board. In this mode, the control
signal also causes the Outputs of module C to assume a high impedance state, and hence C
does not interfere with the test results generated by B, Basically, this approach is similar to the
previously discussed technique of using multiplexers to improve testability.
The test mode signals required by the 4 hardware such as multiplexers, tristate drivers, and so
forth cannot always be applied via the edge pins, because there may not be enough of them. To
overcome this problem, a “test state register” may be incorporated in the design. This could in
fact be a shift register that is loaded and controlled by just a few signals, The various testability
hardware in the circuit can then be controlled by the parallel outputs of the shift register.
Frequently, flip-flops, counter, shift registers, and other memory elements assume
unpredictable states when power is applied, and they must be set to known states before testing
can begin, Ideally, all memory elements should be reset from the external pins of the circuit,
whereas in some cases additional logic may be required (Fig, 5.8), With complex circuits it
may be desirable to set memory elements in several known states, This not only allows
independent initialization, it also simplifies generation of certain internal states required to test
the board adequately.
A long counter chain presents another practical test problem. For example, the counter chain
shown in Fig. 5.9 requires thousands of clock pulses to go through all the states, One way to
avoid this problem is to break up the long chain into smaller ones by using a multiplexer. When
the control input c of the multiplexer is al logic 0, the counter functions normally, When c is at
logic 1, the counter is partitioned into two smaller counters.
A feedback loop is difficult to test, because it hides the source of a fault. The Source can be
located by breaking the loop physically and bringing both lines to external pins that can be
short-circuited for normal operation. When not short-circuited, the separated lines provide a
control point and a test point. An alternative way of breaking a feedback loop, rather than using
more costly test/control points, is to add to the feedback path a gate that can be interrupted by
a signal from the tester (Fig. 5.10).
The use of checking experiments to determine whether a sequential circuit represents the
behavior specified by its state table yields good results provided that
For circuits that do not have any distinguishing sequences, the checking experiments are very
long and consequently hard to apply in any practical situation. One approach to this problem
is to modify a given circuit by adding extra outputs so that the modified circuit has a
distinguishing sequence. A sequential circuit which possesses one or more distinguishing
sequences is said to be diagnosable.
A procedure for modifying a sequential circuit to possess a distinguishing
sequence if it does not already do so has been presented by Kohavi and Lavelle [5.3]. Let us
explain the procedure by considering the state table of Circuit M shown in Fig. 5.11(a); Circuit
M does not have a distinguishing sequence. The procedure begins with the construction of the
testing table of the circuit; the testing table for Circuit M is shown in Fig. 5.11(b). The column
headings consist of all input/output combinations, where the pair X/Z corresponds to input X
and output Z. The entries of the table are the ‘‘next state.’’ For example, from state A under
input the circuit goes to state 8 with an output of 0. This is denoted by entering B in column
1/0 and a dash (—) in column 1/1. In a similar manner, the next states of A are entered in the
upper half of the table.
The lower half of the table is derived in a straightforward manner from the upper half. If the
entries in rows Si, and Sj column Xk/Z1 of the upper half are Sp and Sq respectively, the entry
in row SiSj, column Xk/Z1, of the lower half, is SrSq. For example, because the entries in rows
A and B, column 1/0, B and C, respectively, the corresponding entry in row AB, column 1/0,
is BC and so on. If for a pair Si and Sj either one or both corresponding entries in some column
Xk/Z1 are dashes, the corresponding entry in row SiSj, column Xk/Z1, is a dash. For example,
the entry in row BD, column 0/0, is a dash, because the, entry in row D, column 0/0, is a dash.
Whenever an entry in the testing table consists of a repeated state (e.g., AA in row AB), that
entry is circled. A circle around AA implies that both states A and B are merged under input 0
into state A and hence are indistinguishable by any experiment Starting with an input 0.
The next step of the procedure is to form the testing graph of the circuit. The testing graph is a
directed graph with each node corresponding to a row in the lower half of the testing table. A
directed edge labeled Xk/Z1 is drawn from node SiSj to node SpSq, where p not equal q, if
there exists an entry in row SiSj column Xk/Z1 of the testing table. Figure 5.12 shows the
testing graph for Circuit M.
A circuit is definitely diagnosable if and only if its testing graph has no loops and there
are no repeated states, that is, no circled entries in its testing table. Circuit M is therefore not
definitely diagnosable, because AA exists in its testing table and its testing graph contains tow
loops: AB-BC-CD-AD-AB and AC-BD-AC.
To make the circuit definitely diagnosable, additional output variables are required to
eliminate all repeated entries from its testing table and to open all loops in its testing graph.
The maximum number of extra output terminals re- quired to make a 2^k state circuit definitely
diagnosable is k; however, the addition of one output terminal is sufficient to make Circuit M
definitely diagnosable. The modified state table of Circuit M is shown in Fig. 5.13; this version
possesses the distinguishing sequences 0 and 11. The checking experiment for a definitely
diagnosable circuit can be derived as follows:
1. Apply a homing sequence, followed by a transfer sequence (Si, So) is necessary, to bring the
circuit into an initial state So.
2. Choose a distinguishing sequence so that it is the shorter one of the sequences of all 0s or all
1s.
3. Apply the distinguishing sequence followed by a 1. (If the all-0s sequence has been chosen,
apply a 0 instead of a 1.)
4. If S01, that is, the 1-successor of S0, is different from S0, apply another 1 to check the
transition from S01, under a 1 input. Similarly, if S011 not equal to S01 and S011 not equal to
S0, apply another 1.Continue to apply 1 inputs in the same manner as long as new transitions
are c
5. When an additional 1 input does not yield any new transition, apply an input of 0 followed
by the distinguishing sequence.
6. Apply inputs of 1s as long as new transitions can be checked. Repeat steps 5 and 6 when no
new transitions can be checked.
7. When steps 5 and 6 do not yield any new transitions and the circuit, which is in state Si, is
not yet completely checked, apply the transfer sequence T(Si, Sk), where Sk is a state the
transition of which has not been checked, such that T(Si, Sk) passes through checked transitions
only,
8. Repeat the last three steps until all transitions have been checked.
The checking experiment for the definitely diagnosable circuit of Fig. 5.13 has been designed
using the foregoing procedure. It required only 23 symbols and is illustrated here:
The testing of sequential circuits is complicated because of the difficulties in setting and
checking the states of the memory elements. These problems can be overcome by modifying
the design of a general sequential circuit so that it will have the following two properties [5.4]:
1. The circuit can easily be set to any desired internal state.
2. It is easy to find a sequence of input patterns such that the resulting out- put sequence will
indicate the internal state of the circuit. In other words: the circuit has a distinguishing
sequence.
The basic idea is to add an extra input c to the memory excitation logic in order to control the
mode of a circuit, When c = 0, the circuit operates in its normal mode, but when c = 1, the
circuit enters into a mode in which the elements are connected together to form a shift register.
This facility is incorporated by inserting a double-throw switch in each input lead of every
memory element.
All these switches are grouped together, and the circuit can operate either in its normal mode
or shift register mode. Figure 5.14 shows a sequential circuit using D flip-flops; the circuit is
modified as shown in Fig. 5.15, Each of the double throw switches may be realized as in Fig
5.16.
eae : eae ‘ ee
Dept of ECE, AIT Page 12
Module 5 VLSI Design (21EC63)
In the shift register mode, the first flip-flop can be set directly from the primary inputs
(scan-in inputs) and the output of the last flip-flop can be directly monitored on the primary
output (scan-out output). This means that the circuit can be set to any desired state via the scan-
in inputs, and that the internal state can be determined via the scan-out output. The procedure
for testing the circuit is as follows:
The main advantage of the scan-path approach is that a sequential circuit can be transformed
into a combinational circuit, thus making test generation for the circuit relatively easy. Besides,
very few extra gates or pins are required for this transformation.
Another implementation of the scan-path technique has been described by Funatsu et al. [5.5].
The basic memory element used in this approach is known as a raceless D-type flip-flop with
scan path [5.6] Figure 5.17(a) shows such a memory element, which consists of two latches L/
and L2. The two clock signals Cl and C2 operate exclusively, During normal operation, C2
remains at logic 1 and C/ is set to logic 0 for sufficient time to latch up the data at the data input
D1. The output of L1 is latched into L2 when C1 returns to logic 1.
Scan-in operation is realized by clocking the test input value at D2 into the latch L1 by
setting C2 to logic 0. The output of the L1 latch is clocked into L2 when C2 returns to logic 1.
The configuration of the scan-path approach used at logic card level is shown in Fig.
5.17(b). All the flip-flops on a logic card are connected as a shift register, such that for each
card there is one scan path. In addition, there is provision for selecting a specified card in a
subsystem with many cards by X-Y address signals (Fig. 5.17(b)). If a card is not selected, its
output is blocked; thus, a number of card outputs in a subsystem can be put together with only
a particular card having control of the test output for that subsystem. The Nippon Electric
Company in Japan has adopted this version of the scan path approach to improve the testability
of their FLT-700 processor system.
One of the best known and the most widely practiced methods for synthesizing testable
sequential circuits is the IBM LSSD (level-sensitive scan design).
The clock signal C will normally occur (change from 0 to 1) after the data signal D has
become stable at either a 1 or a 0. The output of the latch is set to the new value of the data
signal at the time the clock signal occurs. The correct changing of the latch does not depend on
the rise or fall time of the clock signal, only on the clock signal being 1 for a period equal to or
greater than the time Required for the data signal to propagate through the latch and stabilize.
A shift register latch (SRL) can be formed by adding a clocked input to the polarity-
hold latch L1 and including a second latch L2 to act as intermediate storage during shifting
(Fig. 5.19). As long as the clock signals A and B are both 0, the L1 latch operates exactly like
a polarity-hold latch. Terminal 1 is the scan-in input for the shift register latch and +L2 is the
output. The logic implementation of the SRL is shown in Fig. 5.20.
When the latch is operating as a shift register, data from the preceding Stage are gated
into the polarity-hold switch via l, through a change of the clock A from 0 to 1. After A has
changed back to 0, clock B gates the data in the latch L1 into the output latch L2 Clearly, A
and B can never both be 1 at the same time if the shift register latch is to operate properly.
The SRLs can be interconnected to form a shift register as shown in Fig. 5.21. The input
l and the output +L2 are strung together in a loop, and the clocks A and B are connected in
parallel.
A specific set of design rules has been defined to provide level-sensitive logic subsystems with
a scannable design that would aid testing [5.8).
Rule 3: It must be possible to identify a set of clock primary inputs from which the clock
inputs to SRLs are controlled either through simple powering trees or through logic that is
gated by SRLs and/or non-clock primary inputs (see Fig. 5.23). Given this structure, the
following rules must hold:
(a) All clock inputs to all SRLs must be at their OFF states when all clock primary inputs
are held to their OFF states (bee Fig. 5.23(a)).
(b) The clock signal that appears at any clock input of any SRL must be controllable from
one or more clock primary inputs, such that it is possible to set the clock input of the SRL to
an ON state by turning any one of the corresponding clock primary inputs to its ON state and
also setting the required gating condition from SRLs and/or non-clock primary inputs.
(c) No clock can be ANDed with either the true value or the complement value of another
clock (see Fig. 5.23(b)).
Rule 4: Clock primary inputs may not feed the data inputs to latches either directly or through
combinational logic, but they may only feed the clock input the latches or the primary outputs.
A sequential logic network designed in accordance with Rules 1-4 would be level-
sensitive. To simplify testing and minimize the primary inputs and outputs, it must also be
possible to shift data into and out of the latches in the system. Therefore, two more rules must
be observed:
Rule 5: All SRLs must be interconnected into one or more shift registers, each of which has
an input, an output, and clocks available at the terminals of the module.
Rule 6: There must exist some primary input sensitizing condition (referred to as the scan state)
such that
(a) Each SRL or scan-out primary output is a function of only the preceding SRL or scan-
in primary input in its shift register during the shifting operation.
(b) All clocks except the shift clocks are held OFF at the SRL inputs.
(c) Any shift clock to an SRL may be turned ON and OFF by changing the corresponding
clock primary input for each clock.
A sequential logic network that is level-sensitive and also has the scan Capability as per
Rules 1 to 6 is called a level-sensitive scan design (LSSD). Figure 5.24(a) depicts a general
structure for an LSSD system in which all system Output are taken from the L2 latch; hence, it
is called a double-latch design. In the double-latch configuration, each SRL operates in a
master-slave mode. Data transfer occurs under system clock and scan clock B during normal
operation, and under scan clock A and scan clock B during scan-path operation. Both latches
are therefore required during system operation. In the single-latch configuration, the
combinational logic is partitioned into two disjoint sets, Comb1 and Comb2 (Fig. 5.24(b)). The
system clocks used for SRLs in. Comb1 and Comb2 are denoted by Clock1 and Clock2,
respectively; they are nonoverlapping.
The outputs of the SRLs in Comb1 are fed back as secondary variable inputs to Comb2, and
vice versa. This configuration uses this output of latch L1 as the system output;
the £2
The L2 latch is used only for shifting. In other words, the L2 latches are redundant and
represent an overhead for testability. However, the basic SRL design can be modified to reduce
the overhead. The modified latch galled the L1/L2* SRL is shown in Fig. 5.25. The main
difference between the basic SRL and the L1/L2* SRL is that L2* has an alternative system
data input D2 clocked in by a separate system clock C2. The original data input D in Fig. 5.19
is also available and is now identified as D1 in Fig. 5.25. D1 is clocked in by the original system
clock, which is now called C1. Clock signals C1 and C2 are nonoverlapping. The single-latch
configuration of Fig. 5.25(b) can now be modified to the configuration of Fig. 5.26, in which
the system output can be taken from either the L1 output or the L2* output. In other words,
both L1 and L2* are utilized, which means fewer latches are required in the system. As a result,
there is a significant reduction in the silicon cost when L1/L2* SRLs are used to implement the
LSSD. Although both latches in L1/L2* SRLs can be used for system functions, it is absolutely
essential, as in conventional LSSD, that both L1 and L2* outputs do not feed the same
combinational logic.
The LSSD approach is very similar to the scan-path approach used by the NEC, except that it
has the level-sensitive attribute and requires two separate clocks to operate latches LJ and L2.
The use of LSSD alleviates the testing problems in the following ways:
1. The correct operation of the logic network is independent of a.c. characteristics such as
clock edge rise time and fall time.
2. Network is combinational in nature as far as test generation and testing is concerned.
3. The elimination of all hazards and races greatly simplifies both test gen eration and fault
simulation.
Any desired pattern of 1s and 0s can be shifted into the polarity-hold latches inputs to the
combinational network. For example, the combinational network of Fig. 5.24(a) is tested by
shifting part of each required pattern into the SRLs, with the remainder applied through the
primary inputs. Then the system clock is turned on for one cycle, the test pattern is propagated
through the combinational logic, and the result of the test is captured in the register and at the
primary outputs. The result of the test captured in the register is then scanned out and compared
with the expected response. The shift register must also be tested, and accomplished by shifting
a short sequence of 1s and 0s through the shifted latches.
The design methods discussed in Secs. 5.4 and 5.5 use sequential access scan-in techniques to
improve testability; that is, all flip-flops are connected in testing to form a shift register or
registers. In an alternative approach, known as random access scan each flip-flop in a logic
network is selected in an address for control and observation of its state [5.11]. The basic
memory element in a random access scan-in/scan-out network is an addressable latch. The
circuit diagram of an addressable latch is shown in Fig. 5.27. A latch Y address signals, the
state of which can then be controlled and scan-in/scan-out lines. When a latch is selected and
its scan 0 to 1, the scan data input is transferred through the network to output, where the
inverted value of the scan data can be observed. DATA line is transferred to the latch output Q
during the negative transition (1 to 0) of the clock. The scan data out lines from all latches are
produce the chip scan-out signal, the scan-out line of a latch remains at logic 1 unless the latch
is selected by the X-Y signals.
A different type of addressable latch—the set/reset type—is shown in Fig. 5.28. the “clear”
signal clears the latch during its negative transition. Prior to scan-in operation, all latches are
cleared. Then, a latch is addressed by the X-Y lines and the preset signal is applied to set the
latch state.
The basic model of a sequential circuit with random access scan-in/scan-out network is shown
in Fig. 5.29. The X- and Y-address decoders are used to access an addressable latch—like a
cell in random access memory. A tree of AND gates is used to combine all scan-out signals.
Clear input of all latches are tied together to form a master reset signal. Preset inputs of all
latches receive the same scan- in signal gated by the scan clock; however, only the latch
accessed by the X-Y addresses is affected
.
The test procedure of a network with random access scan-in/scan-out network is us follows:
1. Extra logic in the form of two address gates for each memory element, plus the address
decoders and output AND trees, result in 3—4 gates overhead per memory element.
2. Scan control, data, and address pins add up to 10-20 extra pins. By using a serially loadable
address counter, the number of pins can be reduced to around 6.
3. Some constraints are imposed on the logic design, such as the exclusion of asynchronous
latch operation.
In full scan, all flip-flops in a circuit are connected into one or more shift registers: thus, the
states of a circuit can be controlled and observed via the primary inputs and outputs,
respectively. In partial scan, only a subset of the circuit flip-flops are included in the scan chain
in order to reduce the overhead associated with full scan design [5.12]. Figure 5.30 shows a
structure of partial scan design. This structure has two separate clocks: a system clock and a
scan clock. The scan clock controls only the scan flip-flops. Note that the scan clock is derived
by gating the system clock with the scan-enable signal; no external clock is necessary. During
the normal mode of operation, namely, scan-enable signal at logic 0, both scan and non-scan
flip-flops update their states when the system clock is applied. In the scan mode operation, only
the state of the shift register (constructed from the scan flip-flops) is shifted one bit with the
application of the scan Nip-flop; the non-scan flip-flops do not change their states.
The disadvantage of two-clock partial scan is that the routing of two separate clocks
with small skews is very difficult to achieve. Also, the use of a separate scan clock docs not
allow the testing of the circuit at its normal Operating speed. Cheng [5.13] proposed a partial
scan scheme, shown in Fig. 5.31, in which the system clock is also used as the scan clock. Both
scan and non-scan flip-flops move to their next states when the system clock is applied.
A test sequence is derived by shifting data into the scan flip-flops. This data together with
the contents of non-scan flip-flops constitute the Starting state of the test sequence, The other
patterns in the sequence are obtained by single-bit shifting of the contents of scan flip-flops,
which form part of the required circuit states. The remaining bits of the states, that is, the
contents of non-scan flip-flops, are determined by the functional logic. Note that this form of
partial scan scheme allows only a limited methods. Trischler [5.14] has used testability analysis
to show that the fault coverage in a circuit can be significantly increased by including 15-25%
of the flip flops in the partial scan. Agrawal et al. [5.15] have shown that the fault coverage
can be increased to as high as 95% by including less than 65% of the flip-flops in the partial
scan.
Full and partial scan techniques improve the controllability and observability of flip-flops in
sequential circuit, and therefore the test generation for such circuits is considerably simplified.
However, a scan-based circuit cannot be tested at its
IMPORTANT QUESTIONS:
2) With a neat logic diagram, explain clocked hazard free latches used in
LSSD Technique.
normal speed, because test data have to be shifted in and out via the scan path.