Professional Documents
Culture Documents
When logic 1 is connected to input A, the output Q is at logic 0. The logic state
is input to second NOT gate and its output Q will be at logic 1, which is the
same state as the input A. In this manner, the output Q and Q would stay at its
respective logic state even if the logic 1 at input A is removed.
- 173 -
When logic 0 is connected to input A, the output Q will be at logic 1. The logic
state is input to second NOT gate and its output Q will be at logic 0, which is
the same state as the input A. In this manner, the output Q and Q would remain
at its respective logic state even if the logic 0 at input A is removed. Combining
both conditions of logic states, the bi-stable element forms the basic memory
bit. The layout of the bi-stable element is shown in Fig. 7.3.
The bi-state element has two stable states and one unstable state. The unstable
state occurs at the mid-point voltage. At this point all transistors are in
saturation mode and also at the highest potential energy.
The SR flip-flop is shown in Fig. 7.4. The output Q is Q = S CLK + Q and
Q = R CLK + Q . Using DeMorgans theorem, Q is also equal to Q =
(S + CLK ) Q , which forms the p-MOS transistor circuit of the output Q . Output
Q is also equal to (R + CLK ) Q , which forms the p-MOS transistor circuit of
output Q.
Another compact way to design a D flip-flop is shown in Fig. 7.9. The output at
node X is D Load + D Load , which is equal to D. Thus, data D is latched when
Load is equal to logic 1.
In order to avoid wrong data being latch into the D flip-flop, the D flip-flop can
be designed with master/slave operation that utilizing transmission gate and
NOT gates. Figure 7.10 shows the CMOS design of a master/slave D flip-flop.
Transmission gate B and D are used to prevent the output of master flip-flop
- 177 -
and slave flip-flop from driving the output of transmission gate A and C. Unless
the output of transmission gate A and C is able to sink or source sufficient
current to overcome the output drive from output of master flip-flop and slave
flip-flop, wrong data latch would occur. Besides having transmission gate B and
D, the aspect ratio W/L of transmission gate A and C can be designed sufficient
large as compare with the aspect ratio of other transistor.
At node X, the logic function is Load D and the logic function at node Y is
Load D . The logic function at node Y is Load D Load = (Load D) + Load and
the logic function at node Z is (Load D) + Load = D, which is the data D.
- 178 -
Based on the output equations mentioned above, the CMOS circuit design of the
JK flip-flop is shown in Fig. 7.12.
= Q T CLK + Q = ( Q + T + CLK ) Q .
- 179 -
Based on the output equations, the CMOS circuit design of the T flip-flop is
shown in Fig. 7.14.
Based on the design for a basic D-flip-flop, the design of a master-slave D flipflop is shown in Fig. 7.16. This design is used to avoid latching of wrong logic.
This memory device has 128 row addresses and 8 column addresses. The
memory has 8 matrix blocks and each block has 128x8 cells. The other main
parts of the memory are the sense amplifier, control unit, input/output data
control, output data control, address bus, and data bus.
We shall discuss the approaches used to design memory cell static and
dynamic cell, the sense amplifier, and the address decoders row and column
decoders, and I/O data control circuits.
During the write cycle, the desired logics are placed on bit line and BIT line.
When the WORD line is asserted, the desired data will be latched into the bistable memory element. For an example, to write logic 1 into the memory, the
BIT line is set at logic 1, whilst the BIT line is at logic 0.
However, due to high pack density of the memory cell whereby many
column memory cells are connected in the same bit line, the total drain-bulk
capacitance of the pass-transistors is sufficiently large that the charging and
discharging of the bit lines would take long time. Thus, during read cycle, the
BIT and BIT lines are pre-charged to the pre-defined level, which is usually 0.5
of VDD voltage level. These lines are then allowed to float. When the WORD
line is asserted, the BIT line and BIT line begin to charge or discharge that
reflect the logic level stored in memory cell. The small change in voltage level
is passed to the sense amplifier for output user. The read cycle is a destructive
cycle whereby the data stored in the memory can be erased. Therefore, it is
necessary to refresh the memory. Other mean to prevent the bit data being
erased is to design the pass-transistor to have large width and length. But this is
not desired because in the modern design, scale down is necessary to save cost
and fast access time.
- 183 -
The BIT value is logic 0 then the gate voltage shall be 0V. If the BIT value is at
logic 1 then the gate voltage will be at logic 1 that has voltage (VWrite Vtn(M2)).
This voltage is hold on as long as the Read transistor M3 is not switched on.
During the read cycle, transistor M3 is switched on and if the BIT value is
logic 1 then the BIT line would turn logic 0. Likewise, if the BIT value is logic
0 then upon reading the BIT line would turn logic 1 that has maximum value
(VRead Vtn(M3)).
- 184 -
Read cycle is a destructive operation. Thus, the data must be re-written into the
memory capacitor CM.
If logic 1 is stored in the memory, during the read cycle, BIT voltage shall be
larger than BIT voltage. The output of MOS transistor M2 is
g m2
ro 4 || r02 (VBIT VBIT ) , whilst the output of MOS transistor M6 is
2
g m 2 g m6
(ro 4 || r02 )r07 (VBIT VBIT ) . This voltage value is closed to 5V. Therefore, the
2
voltage value is closed to 0V, thus, the output of the inverter shall be at logic 1.
The typical sense amplifier circuit for the dynamic RAM cell is shown in
Fig. 7.22. Timing is used to charge the capacitor C2 to approximately 5.0V
and the MOS transistor M3 is used during pre-charging the BIT line. The clamp
- 186 -
voltage set the gate voltage of the MOS transistor M2 to be greater than the precharged value of the BIT line.
If the stored bit is a 1, during the read cycle, upon the WORD line is
asserted, the stored charge in memory capacitor CM would charge the CBIT to
higher voltage. This set the MOS transistor M2 to be remained off. As the result
the output is at logic 0.
When the stored bit is a 0, upon the WORD line is asserted, the pre-charge BIT
lines capacitor CBIT would charge the memory capacitor CM. As the result, the
voltage of BIT line capacitor CBIT is lower than the pre-charged value. This
causes the MOS transistor M2 to switch on and capacitor C2 would transfer
charge to BIT line capacitor CBIT. This causes the voltage at input of inverter to
drop very fast due to the value of C2 is smaller than the value of CBIT. As the
result the output is at logic 1.
A cascode voltage switching logic CVSL sense amplifier is shown in Fig.
7.23. During the read cycle, the bit line BL and BL are directed to Output and
Output . Upon enable the sense amplifier SE, the logic of the bit line is sent to
output and Output .
- 187 -
7.3.3 Decoder
A 2-address line dynamic NOR row decoder is shown in Fig. 7.24. There are
two stages for the operation of this decoder. They are the pre-charge stage and
evaluate stage.
The pre-charge input is asserted low that switches on all p-channel MOS
transistors and all outputs WL 0 , WL1 , WL 2 , and WL 3 of the decoder go to logic
1. The decoder should not be read at this instant, since all the outputs are at VDD
logic 1.
The next stage after pre-charge stage is the evaluation stage. Once the precharge is done, the p-channel MOS transistors are switched off. The outputs will
stay at logic 1 because the charge will be stored on the capacitors. The inputs
are applied on two address lines. The corresponding n-channel MOS transistor
will be switched on and the charged capacitor on that line will be discharged to
- 188 -
ground. All lines except one will go to logic 0. The line that remains logic 1 will
be the decoded line that will drive all the n-channel MOS transistors on that
line. Thus, the right data will be available on the data line. This data will be
given to the output data block, which will return the correct data to the outside
world when the output enable signal is asserted.
A 2-address line dynamic NAND row decoder is shown in Fig. 7.25. There are
two stages for the operation of this decoder. They are the pre-charge stage and
evaluate stage.
The pre-charge input is asserted low that switches on all p-channel MOS
transistors and all outputs WL 0 , WL1 , WL 2 , and WL 3 of the decoder go to logic
1. The decoder should not be read at this instant, since all the outputs are at VDD
logic 1.
The next stage after pre-charge stage is evaluation stage. Once the precharge is done, the p-channel MOS transistors are switched off. The outputs will
stay at logic 1 because the charge will be stored on the capacitors. The inputs
- 189 -
are applied on two address lines. The corresponding n-channel MOS transistor
will be switched on and the charged capacitor on that line will be discharged to
ground. All lines remain logic 1 except one will go to ground logic 0. The line
that remains logic 0 will be the decoded line. The right data will be available on
the data line. This data will be given to the output data block, which will return
the correct data to the outside world when the output enable signal is asserted.
- 190 -
The main advantage of this type of decoder is the number of transistors used is
drastically reduced. The major disadvantage is time delay, which increases four
times with the number of sections increases. Thus, this type of design is
prohibited for designing large decoder.
- 191 -
The clock pre-charge circuit is shown in Fig. 7.29. Falling edge clock will
charge the capacitor of the p-channel MOS transistors VDD voltage. During
evaluation cycle the stored charge can be charged or hold depending on the
access condition. The center transistor is the equalization transistor. It is used to
speed up equalization of the two bit lines by allowing the capacitance and pullup device of the non-discharged bit line to assist in pre-charging the discharged
line.
During a write cycle, the write enable line WR is low asserted, which switches
on all pass-transistors. Likewise, during the read cycle, all pass-transistors are
switched off preventing data from entering the memory cell array.
The two inverters before the pass-transistors are basically buffers used to
increase the current handling capacity of the input data control circuit especially
when passing data speedily to the pass-transistor.
One of the other approaches is to use transmission gate replacing passtransistors for passing strong logic 0 and logic 1. One may replace the entire
pass-transistor circuit each with an inverter and a tri-state inverter.
- 193 -
Diode ROM has default logic 0 at bit line. When the word line of ROM is
selected, it provides logic 1 at bit line due low forward voltage of the diode.
The MOS ROM 1 has default logic 0 at bit line. When the word line is selected,
the n-MOS transistor switches on and provides logic 1 at bit line. The MOS
ROM 2 has default logic 1. When the word line is selected, it provides logic 0 at
bit line.
An example of a 4x4-bit NOR ROM is shown in Fig. 7.33. The pull-up pMOS transistor and the n-MOS bit-transistor forms a pseudo n-MOS NOR gate.
ROM is designed with background logic 1. When the WORD line is asserted
the cell that has n-MOS transistor would be switched on to provide logic 0 at the
BIT line upon switched on by CLK signal on p-MOS transistor. Normally the
clock CLK is pre-charged to half of VDD for enable fast access of the memory
cell. The WORD line and BIT line matrix that does not contain any n-MOS
transistor would provide logic 1 upon clock line is asserted. The one that
contains an n-MOS transistor would provide logic 0 because the n-MOS
transistor is a high-asserted low device. When it is selected, the output would be
at logic 0 simply because the transistor is switched on.
The ROM has data value WORD 0 = 0001, WORD 1 = 1000, WORD 2 =
0010, and WORD 3 = 0100.
- 195 -
Instead of having the pull-up p-MOS transistor, ROM design with pull-down nMOS transistor and n-MOS transistor cell tied to VDD is called OR ROM. Figure
7.34 shows a 4x4 bit OR ROM.
The background bit of the OR ROM is logic 0. When a word line is selected, the
bit that has an n-MOS transistor would provide logic 1 at the bit line.
The 4x4 MOS NAND ROM is shown in Fig. 7.35. The background of the
ROM is logic 1. All the n-MOS transistors in the column have to be switched on
in order logic 0 is shown in bit line.
ROM also can be designed using pseudo n-MOS gate both for the decoder and
the memory cell design, which is basically the n-MOS design technique. The
design example is shown in exercise 7.9. The logic to row or word is provided
by the pseudo n-MOS transistor, which act as decoder.
There are two types of flash memory namely the NOR flash and NAND flash.
In NOR gate flash, each cell has one end connected directly to ground and the
other end connected directly to a bit line as shown in Fig. 7.37. This
arrangement is called NOR flash because it acts like a NOR gate. When one of
the word lines is brought high, the corresponding storage transistor acts to pull
the output bit line low.
NAND flash also uses floating-gate transistors, but they are connected in a way
that resembles a NAND gate as shown in Fig. 7.3. Several transistors are
- 198 -
connected in series and only if all word lines are pulled high, the bit line is
pulled low.
Exercises
7.1.
7.2.
7.3.
7.4.
7.5.
Describe how a logic 1 stored in the SRAM cell is read and interpreted by
the sense amplifier.
7.6.
7.7.
7.8.
7.9.
7.10. Determine the data stored in NAND ROM shown in Fig. 7.35.
7.11. A 4x4 ROM designed using pseudo n-MOS transistor concept is shown
in the figure below.
- 199 -
(i).
(ii).
Derive the logic functions for the rows and output D3, D2, D1, and
D0.
Write down the content of the memory.
Bibliography
1.
2.
3.
4.
5.
- 200 -