Chapter 3 - Implementation Technology
Chapter 3 - Implementation Technology
MOSFET Technology:
• By now, you should be familiar with the metal-oxide-
silicon-field-effect-transistor (MOSFET).
• There are two
types of
MOSFET: NMOS
and PMOS.
• An NMOS
operates in
saturation
mode when
𝑽𝑫𝑺 ≥ 𝑽𝑮𝑺 − 𝑽𝒕
and in cut-off
mode when
𝑽𝑫𝑺 < 𝑽𝒕 .
• The PMOS is
inverted.
CMOS Implementation:
• Complementary MOS using NMOS in conjunction
with PMOS.
• In the circuit shown on the right, the two transistors
operate in opposite modes. They forma NOT gate.
• The NAND gate is shown on the left.
• To examine the result, you must check
each of the 4 transistors.
CMOS Implementation:
• Below are the NAND and NOR gates implemented
in CMOS technology:
CMOS Implementation:
• The AND gate
requires two more
transistors.
• This is the main
reason why NAND-
only and NOR-only
logic is preferred.
• Can you derive the
truth table for it?
𝒙𝟏 𝒙𝟐 𝑻𝟏 𝑻𝟐 𝑻𝟑 𝑻𝟒 𝑻𝟓 𝑻𝟔 𝒇
0 0 …… …… …… …… …… …… …
0 1 …… …… …… …… …… …… …
1 0 …… …… …… …… …… …… …
1 1 …… …… …… …… …… …… …
Programming of Connections:
• The connections to the AND/OR planes are simply
switches that can be open or closed.
• The designer usually specifies the desired circuit
using a CAD tool, which automatically produces
the necessary information for programming each
switch in the PLD (programming file / fuse map).
• The PLD is usually placed within a programmer with
which the CAD interfaces through a cable.
• Usually, the PLD is packaged in a plastic-leaded
chip carrier (PLCC).
• Here is an example:
Custom Chips:
• Although PLDs are cheap and provide reasonable
flexibility to the designer, the programmable
connections they feature suffer from two main
drawbacks:
– Low speed.
– Increased cost.
• When PLDs do not meet the requirements of the
desired circuit, the designer must resort to custom
designed chips, where the layout of each transistor
(and wire) must be specified.
• Custom chips have a huge overhead, and thus are
only feasible when a sufficient number of circuits is
to be sold.
• A middle solution can be:
– Application specific integrated circuits (ASIC).
– Gate Arrays.
Gate Array:
• In a gate array, parts of the chip are pre-fabricated
and parts are customized.
• The manufacturer create a gate array template.
• The template is basically a silicon wafer of partially
complete chips.
• NAND-only
templates are
common.
• The
connection
wires are
manufactured
after receiving
the circuit
requirements
from the user.
An Example:
• Consider the function:
𝑭 𝒙𝟏 , 𝒙𝟐 , 𝒙𝟑 = 𝟎, 𝟐, 𝟒, 𝟓, 𝟔
An Example:
• The previous code is synthesized by the CAD tool
according to the devices to be programmed.
• The CAD optimizer produces: 𝑭 = 𝒙′𝟑 + 𝒙𝟏 𝒙′𝟐 .
• Below are two different implementations of the
function:
– Left: PAL-like implementation.
– Right: 4-input LUT implementation (d = don’t care).