Q451. How multiple vias are used to reduce crosstalk?
As multiple vias are introduced, the resistances will be in parallel, which will reduce the RC delay and further decrease the crosstalk accordingly.
Q452. What is the difference between crosstalk noise and crosstalk delay?
If two signals are close enough, they can cause crosstalk due to coupling
capacitance.
Crosstalk delay:
When the one net is switching at a faster rate and the other is switching at a
slower speed, due to crosstalk, the speedy net will boost-up the slower net.
This is known as crosstalk delay, which is due to timing errors of signals.
Crosstalk noise:
In the case of crosstalk noise, one net is idle (either at logic 1 or logic
0) and the other net is in transition mode (switching from 0 to 1 or vice
versa). There may be the introduction of unwanted signal transition due to
the coupling capacitor. The reason for crosstalk noise is charge storage effect,
power supply or substrate noise. The crosstalk noise analysis tool determines
the worst-case glitch on the idle net. The commands for noise analysis are
report_noise, check_noise, and update_noise.
Q453. Elaborate on the concept of OCV (on-chip variation)?
All devices along a chip should run at a specific speed and interconnects
should be either at the worst-case or best-case corner. But due to some
variations at the manufacturing level, the speed is not uniform throughout
the chip. There is variation in effective channel length and width of transistors. Due to complexities and variations in submicron technologies, the
devices with the same size may have different width as compared to the idle
condition.
The major on-chip variations are:
- Variation in the channel length
- Variation in temperature
- IR drop variation
- Variation in transistor width
- Variation in threshold voltage
- Variation in interconnects
Q454. Enlist any two sources of on-chip variation (OCV)?
- Etching
- Photolithography
- Chemical mechanical planarization
Q455. What do you understand by the clock generator and clock distributor?
A clock is a signal that oscillates between low to a high state and vice
versa. A network that distributes clock to all clocked elements, for example, buffer and metal network, is known as clock distributor. Whereas,
the clock generator is an electronic circuit that produces timing signals.
The clock generator is used in the synchronization of a circuit. The
basic components of the clock generator are the amplifier and resonant
circuit.
Q.456. What are global chip-to-chip variation and local on-chipvariation?
The performance difference between the die is known as global chip-tochip variation. It is modeled as operating corners. Within the same die, if
there is a performance difference between the transistors, it is known as
local on-chip variations. It is modeled as an added derating factor to skew
calculations.
The variation constraints may be:
- The thickness of the oxide layer
- Transistor channel dimensions (length and width)
- The number of doping atoms
Q.457. What are the two main clock distribution styles in VLSI?
There are two clock distribution systems:
- Clock tree, also known as Clock Tree Synthesis (CTS). It is placing and routing clock tree elements.
- Clock mesh or Clock grid distribution system
Q.458. What is the clock grid distribution system?
The main goal of the clock grid distribution system is to provide uniform
delay from the source of the clock to the end receiver i.e. flip-flop, latches,
etc. It minimizes the clock skew by providing the concept of stages. The
number of stages is dependent on the size of the chip or process technology.
It is desirable to have a minimum number of stages. The clock distribution
system targets to have the same delay in all stages.
There are two distribution stages; one stage is from the clock source
to the boundaries of the block and another stage is a distribution from
block boundaries to block. The clock grid distribution system is shown as
in Figure.
In stage two, the clock distribution is done inside the block through a fixed
number of stages through the mesh/grid of clock buffers. The clock buffers
should be symmetric.
Q.459. Explain the concept of the clock mesh distribution system?
In the clock mesh distribution system, the main clock signal is divided into
parallel paths using drivers. The array of buffers is cross-connected in a
metallic mesh. The driver feeds these buffers. It routes the path to clock sinks.
A resonant structure is created using mesh cross-links. The delay of buffers
is terminated because of the resonant structure. The clock mesh distribution
system is shown in Figure
This kind of clock distribution system is used in high-speed microprocessors. Usually, clock routes are shielded to reduce the coupling effect and
variations due to coupling. Due to the variation tolerance nature of the clock
mesh system, the problem of clock skew is reduced to a great extent.
Q.460. What is a clock tree distribution system?
In the clock tree distribution system, the clock is distributed to all receivers
i.e. flip-flop, counter, etc. In clock tree synthesis, the number of stages is
optimal. It is not necessary to have an equal number of stages throughout the
distribution system. As the number of stages is less, so the power dissipation
is also low. The clock tree distribution system is shown in Figure.
For slow clock designs, the clock tree distribution is best suited