Title: Chap 8 - 1
1Chapter 8MOS Memory and Storage Circuits
- Microelectronic Circuit Design
- Richard C. JaegerTravis N. Blalock
2Chapter Goals
- Overall memory chip organization
- Static memory circuits using the six-transistor
cell - Dynamic memory circuits
- Sense amplifier circuits used to read data from
memory cells - Learn about row and address decoders
- Implementation of CPU registers via flip-flops
- Pass Transistor Logic (PTL)
- Read Only Memory (ROM)
3Random Access Memory
- Random Access Memory (RAM) refers to memory in a
digital system that has both read and write
capabilities - Static RAM (SRAM) is able to store its
information as long as power is applied, and it
does not lose the data during a read cycle - Dynamic RAM (DRAM) uses a capacitor to
temporarily store data which must be refreshed
periodically to prevent information loss, and the
data is lost in most DRAMs during the read cycle - SRAM takes approximately four times the silicon
area of DRAM
4A 256-Mbit Memory Chip
- The figure shows the block structure of a 256-Mb
memory - There are sets of column and row decoders that
are used for memory array selection - The column decoder splits the memory into upper
and lower halves - The row decoder and wordline drivers bisect each
32-Mb subarray
Note that the basic building block for this
memory is a 128Kb cell
5A 256-Mbit Memory Chip
- The memory block diagram contains 2MN storage
locations - When a bit has been selected, the set of sense
amplifiers are used to read/write to the memory
location - Horizontal rows are referred to as wordlines,
whereas the vertical lines are called bitlines
6Static Memory Cells
- Inverters configured as shown in the above figure
form the basic static storage building block - These cross-coupled inverters are often referred
to as a latch - The circuit uses positive feedback
7Static Memory Cells VTC
- The previous latch has only two stable states and
is termed bistable - However, it is possible for it to be held at an
unstable equilibrium point where slight changes
in the voltage will cause it to latch in one of
the stable states
8The 6-T Cell
- With the addition of two control transistors it
is possible to create the 6-T cell which stores
both the true and complemented values of the data
98.2.2 The Read Operation of a 6-T Cell
Initial state of the 6-T cell storing a 0 with
the bitlines initial conditions assumed to VDD/2
Conditions after the WL transistors have been
turned on
10The Read Operation of a 6-T Cell
Final read state condition of the 6-T cell
11The Read Operation of a 6-T Cell
Waveforms of the 6-T cell read operation
Wordline capacitive coupling effect
12The Read Operation of a 6-T Cell
- Reading a 6-T cell that is storing a 1 follows
the same concept as before, except that the
sources and drains of the WL transistors are
switched - Note that the delay is approximately 20ns for
this particular cell
138.2.3 The Write Operation of a 6-T Cell
It can be seen that not much happens while
writing a 0 into a cell that already stores a
0
14The Write Operation of a 6-T Cell
While writing a 0 to a cell that is storing a
1, the bitlines must be able to overpower the
output drive of the latch inverters to force it
to store the new condition
158.3 Dynamic Memory Cells
- The 1-T cell uses a capacitor for its storage
element (data is represented as either a presence
or absence of a charge) - Due to leakage currents of MA, the data will
eventually be corrupted, hence it needs to be
refreshed
16Data Storage in a 1-T Cell
Storing a 0
Storing a 1
17(No Transcript)
18(No Transcript)
19(No Transcript)
208.3.2 Data Storage in a 1-T Cell
- Notice that the voltage stored on the storage
capacitor on the previous slide does not reach
VDD - It instead is determined by the following
21Data Storage in a 1-T Cell
- To read a DRAM cell, the bitline is precharged to
either VDD or VDD/2, and then MA is turned on
22Data Storage in a 1-T Cell
- The charge stored on CC will be shared with CBL
through the process of charge sharing, where the
read voltage varies slightly - Normally CBL gtgt CC, and the charging time
constant is
23The Four-Transistor (4-T) Cell
- Since the 6-T SRAM provides a large signal
current drive to the sense amplifier, it
generally has shorter access time as compared to
a DRAM - The 4-T DRAM cell is an alternative that
increases access time, and automatically
refreshes itself
24- Node D charges up to 3 VTN 1.9 V through
access transistor MA1. - If BL, BL and the wordline are all forced high,
the two access transistors temporarily at as load
devices for the 4-T cell, and the cell levels are
automatically refreshed.
25Reading and Writing to the 4-T DRAM Cell
268.4 Sense Amplifiers
- Sense amplifiers are used to detect the small
currents that flow through the access transistors
or the small voltage differences that occur
during charge sharing
27A Sense Amplifier for the 6-T Cell
- MPC is the precharge transistor whose main
purpose is to force the latch to operate at the
unstable point previously mentioned
28Sense Amplifier Example
- For the figure on the previous slide, find the
currents in the latch transistors when MPC is
turned on under the following conditions
29Sense Amplifier Example
- Since the output voltage should equal on both
sides of the latch when MPC is on, it is known
that VGSn VDSn for the latch NMOS devices and
VSGp VSDp for the latch PMOS devices. Therefore
these transistors are saturated. - Due to the symmetry of the situation, the drain
currents are equal giving the following
30Sense Amplifier Example
- The drain currents are then found by
- Note that the PMOS and NMOS drain currents are
equal - The power dissipation is given by
31A Sense Amplifier for the 1-T Cell
- The same sense amplifier used in the 6-T cell can
be used for the 1-T cell in manner shown in the
figure
32Sense Amplifier for the 1-T Cell
- The sense amplifier works the same as it did for
the 6-T cell, but takes longer to reach steady
state after precharge
33The Boosted Wordline Circuit
- Obviously it is desired to have a fast access in
many DRAM applications. - By driving the wordline to a higher voltage
(referred to as a boosted wordline), say 5V
instead of 3V, it is possible to increase the
amount of current supplied to the storage
capacitors
34Clocked CMOS Sense Amplifiers
- The sense amplifier can definitely be a major
source of power dissipation, but by using a
clocking scheme, it is possible to reduce the
power dissipated, Dummy Cell (DC), Precharge
(PC), Latch clock (LC)
35Clocked CMOS Sense Amplifiers
- Clocking the previous circuit in the manner shown
in the figure will eliminate static currents in
the latch during the precharge state, and only
transient currents will appear
36Address Decoders
- The following figures are examples of commonly
used decoders for row and column address decoding
NMOS NOR Decoder
NMOS NAND Decoder
37Address Decoders
Complete 3-bit domino CMOS NAND decoder
38Address Decoders
3-bit column data selector using pass-transistor
logic
39Data Transmission through the Pass-Transistor
Decoder
40Read-Only Memory (ROM)
- ROM is often needed in digital systems such as
- Holding the instruction set for a microprocessor
- Firmware
- Calculator plug-in modules
- Cartridge style video games
41Read-Only Memory (ROM)
- The basic structure of the NMOS static ROM is
shown in the figure - The existence of an NMOS transistor means a 0
is stored at that address otherwise a 1 is
stored - Power dissipation is large
42Read-Only Memory (ROM)
- The domino CMOS ROM is one technique used to
lower the amount of power dissipation
43Read-Only Memory (ROM)
- Another ROM option is the NAND array ROM which
can be directly used with a NAND decoder
44Read-Only Memory (ROM)
- The main problem with these previous ROMs is that
they must be designed at the mask level, meaning
that it is not a versatile product. - To solve this problem, the programmable ROM
(PROM) was introduced - The standard PROM cannot be erased, so the
erasable ROM (EPROM), and later, electrically
erasable ROM (EEPROM) were introduced - High density flash memories allow for electrical
erasure and reprogramming of memory cells
458.7 Flip-Flops
46(No Transcript)
47(No Transcript)
48RS Flip-Flop
- The reset-set (RS) flip-flop can be easily
realized by using either two cross-coupled NOR or
NAND gates - The RSFF has the following truth tables
NOR RSFF
NAND RSFF
R S Q Q
0 0 Q Q
0 1 1 0
1 0 0 1
1 1 0 0
R S Q Q
0 0 Q Q
0 1 0 1
1 0 1 0
1 1 1 1
49RS Flip-Flop
50RS Flip-Flop
51D-Latch using T-Gates
- A very important circuit of digital systems is
the D-Latch which is used for a D Flip-Flop - Whenever clock C goes high in the D-Latch, the
data on D is passed through to Q
52Master-Slave D Flip-Flop
- By using series D-Latches that latch the data on
opposite clock phases, a master-slave D flip-flop
can be realized
53