Transcript MOS Sequential logic III

Digital Integrated
Circuits
A Design Perspective
Jan M. Rabaey
Anantha Chandrakasan
Borivoje Nikolic
Designing Sequential
Logic Circuits
Revised from Digital Integrated Circuits, © Jan M. Rabaey el al
© Digital Integrated Circuits2nd
Sequential Circuits
Other Latches/Registers: TSPC
VDD
VDD
VDD
VDD
Out
In
CLK
CLK
In
CLK
CLK
Out
Negative latch
Positive latch
The main advantage is the use of single clock, but 12
transistors for the register.
Also, output node might be floating (dynamic storage). It
needs either a static inverter or to be made pseudo-static.
© Digital Integrated Circuits2nd
Sequential Circuits
Embedding Logic in true single phase clock
VDD
VDD
VDD
In1
PUN
VDD
In2
Q
In
CLK
CLK
PDN
Q
CLK
CLK
In1
In2
Example: logic inside the latch
AND latch: implements
AND + latch
The setup time of the embedding TSPC is smaller compared
to the conventional approach, resulting in a smaller clock
period, thus improves the system performance.
This approach is extensively used in EV4 DEC Alpha and other processors
© Digital Integrated Circuits2nd
Sequential Circuits
Simplified TSPC: split output
VDD
VDD
VDD
A
Out
In
CLK
VDD
In
CLK
A
Positive latch
Out
Negative latch
Advantage: reduced complexity, reduced clock load
Disadvantage: not all voltage nodes have full logic swing,
e.g., node A for positive latch is VDD-VT when In=0, which
results in a reduced drive for the following NMOS.
© Digital Integrated Circuits2nd
Sequential Circuits
A specialized TSPC Register
VDD
VDD
VDD
Hold time: 1 inv
M3
CLK
M6
M9
Y
D
CLK
X
M2
M1
CLK
M5
M4
Q
PD time: 3 inv
Q
CLK
Setup:
1 inv
M8
M7
When CLK=0, X=D, second inverter is in precharge and Y=1.
Third inverter is in hold mode.
On rising edge of CLK, second inverter evaluates X and third
inverter is on and passes Y to Q (X can make only 1->0
change during2ndthis time)
© Digital Integrated Circuits
Sequential Circuits
An Alternative Approach to register:
Pulse-Triggered Latches
 Up to now, we have been constructing registers using
master-slave configuration.
 A fundamentally different approach is to use pulse
signals, which is to construct a short pulse around the
rising/falling edge of the clock. The pulse acts as the clock
input to the latch, sampling the input only in a short
window time.
 Race condition can be avoided by keeping the opening
time of the latch very small.
© Digital Integrated Circuits2nd
Sequential Circuits
Pulse-Triggered Latches
Ways to design an edge-triggered register cell:
Master-Slave
registers
L1
Data
Pulse-Triggered
Latch
L2
D Q
D Q
Clk
Clk
L
Data
Clk
D Q
Clk
Clk
More used in AMD processors
© Digital Integrated Circuits2nd
Sequential Circuits
Pulsed Latches
VDD
VDD
M3
M6
CLK
VDD
Q
D
CLKG
M2
CLKG
M1
MP
M5
CLKG
X
MN
M4
(a) register
(b) glitch generation
CLK
CLKG
© Digital Integrated
Circuits2nd
(c) glitch clock
Sequential Circuits
Pulsed Latches
 When CLK=0, node X is charged up to VDD (transistor
MN is off now), CLKG=0
 When CLK=1, there is a short period of time when
both inputs to the AND gate are high, causing CLKG=1.
So, there is initial delay between the rising edge of the
clock and that of the glitch pulse signal.
 When CLKG=1, MN is activated, pulling X to low and
eventually CLKG=0. The delay time is the MN delay, AND
and two Inverter gates delay, which also determines the
pulse width.
© Digital Integrated Circuits2nd
Sequential Circuits
Pulsed Latches
 Setup time=0, hold time=length of the pulse,
propagation delay=two inverter gate delays.
 Advantage of the approach is reduced clock load and
smaller number of transistors (pulse generation circuitry
can be amortized over multiple registers).
 Disadvantage is a substantial verification effort (need
to simulate all corners to ensure the clock pulse always
reliable).
© Digital Integrated Circuits2nd
Sequential Circuits
Another Pulsed Latch
CLK
VDD
VDD
P1
P3
x
M6
M3
D
Q
P2
M2
M5
VDD
M1
CLKD
M4
Hybrid Latch – Flip-flop (HLFF) used in AMD K-6 and K-7
© Digital Integrated Circuits2nd
Sequential Circuits
Another Pulsed Latch
 When CLK=0, P1 on, X precharged to 1, Q is decoupled
from X and is in hold mode.
 On rising edge, M3 and M6 are on and M1/M4 are on for
a short period determined by the delay of three inverters.
Now, D is sampled by the latch.
 Once CLKD goes to low, X is decoupled from D and is
either held or precharged to 1 through P2.
CLK
CLKD
Overlapping clock pairs
© Digital Integrated Circuits2nd
Sequential Circuits
Hybrid Latch-FF Timing
© Digital Integrated Circuits2nd
Sequential Circuits
CLK
Reference data path
b
CLK
CLK
REG
REG
CLK
REG
CLK
Out
REG
b
log
REG
CLK
REG
a
REG
a
REG
Pipelining
log
CLK
CLK
Pipelined data path
Simplifying the analysis, assume each logic module in the
reference datapath has equal delay, then each is active for
only one-third of the clock period.
Pipelining is a popular design technique used to accelerate
the operation of datapath in microprocessor. It improves
resource utilization and increase the functional throughput.
© Digital Integrated Circuits2nd
Sequential Circuits
Out
CLK
Reference
© Digital Integrated Circuits2nd
b
CLK
CLK
log
REG
REG
CLK
REG
CLK
Out
REG
log
REG
b
REG
CLK
a
REG
a
REG
Pipelining
CLK
CLK
Pipelined
Sequential Circuits
Out
Latch-Based Pipeline
H
E
Operation of the pipeline:
When CLK high, E/G
writing C1/C3, while
C2/C4 hold
Computed F
sampled onto C2
Computed G
sampled onto C3
When CLK low, F/H
writing C2/C4, while
C1/C3 hold
Disadvantage: Potential race condition with clock overlap. (e.g., if
input sampled onto C1 propagates to C2 before CLK goes to low)
© Digital Integrated Circuits2nd
Sequential Circuits
Non-Bistable Sequential Circuits─
Schmitt Trigger
V OH
Vou t
In
Out
•VTC with hysteresis
V OL
•Fast response time
•Employs positive feedback
© Digital Integrated Circuits2nd
VM–
VM+
Vi n
Sequential Circuits
Noise Suppression using
Schmitt Trigger
© Digital Integrated Circuits2nd
Sequential Circuits
CMOS Schmitt Trigger
Switching threshold of an inverter is determined by the kn/kp ratio
of NMOS/PMOS. Increasing the ratio raises Vm, while decreasing
the ratio lowers Vm.
VDD
M4
M2
Vin
If Vin=0 so that
Vout=0, M4 on, M3
off (M4,M2
together with M1).
Vout
X
M1
M3
When it switch,
M4 off, M3 on,
this speeds up the
transition.
Similarly with
Vin=1, Vout=1.
Moves switching threshold of the first inverter
© Digital Integrated Circuits2nd
Sequential Circuits
Schmitt Trigger Simulated VTC
2.5
2.5
2.0
2.0
VM1
1.5
1.0
1.5
1.0
VM2
k=1
k=3
0.5
0.0
0.0
k=2
0.5
0.5
1.0
1.5
Vin (V)
2.0
2.5
Voltage-transfer characteristics with hysteresis.
© Digital Integrated Circuits2nd
k=4
0.0
0.0
0.5
1.0
1.5
Vin (V)
2.0
2.5
The effect of varying the ratio of the
PMOS device M4. The width is k* 0.5m m.
Sequential Circuits
CMOS Schmitt Trigger (2)
In=0, Out=1, M5 on so that
when In changes from 0 to
1, it has to be larger to
discharge M5 (through VDD
to M1).
VDD
M4
M6
M3
In
Out
M2
X
M1
© Digital Integrated Circuits2nd
M5
VDD
On the other hand, once Out
switches, M6 on, so positive
feedback to allow fast
transition (M6 diverts current
to GND so that M3, M4
charges less to Out).
Sequential Circuits
Multivibrator Circuits
R
S
Bistable Multivibrator
flip-flop, Schmitt Trigger
T
Monostable Multivibrator
one-shot
Astable Multivibrator
oscillator
© Digital Integrated Circuits2nd
Sequential Circuits
Transition-Triggered Monostable
In
DELAY
td
© Digital Integrated Circuits2nd
Out
td
Sequential Circuits
Astable Multivibrators (Oscillators)
0
1
2
N-1
Ring Oscillator
simulated response of 5-stage oscillator
© Digital Integrated Circuits2nd
Sequential Circuits
PLL Block Diagram
Reference
clock
Up
Phase
detector
Local
clock
Charge
pump
Loop
filter
vcont
VCO
Down
Divide by
N
System
Clock
 PLL is an analog circuit, which is sensitive to all sources
of noise (it might need to be guarded by careful layout).
 When stable, the system clock is N times the reference
clock frequency and ideally 0 skew from the reference clock
© Digital Integrated Circuits2nd
Sequential Circuits