How a circuit remembers: a bit stored in a loop, the difference between level and edge, and the four flip-flop types.
Interactive·Flip-Flop Simulator
Set inputs, then hit Clock ↑ to see the flip-flop capture on the rising edge — just like posedge clk in Verilog.
Inputs
State Register
Q
0
Q̄
1
Memory from feedback
Every circuit so far forgets instantly: change the inputs and the old output is gone. To remember, a circuit needs an output that feeds back into its own input. Two NOT gates in a loop can hold a bit forever, as long as there's power. The only trouble is that you can't change it. Replace the NOTs with NORs, and you get a way in.
The SR latch
The SR latch has two inputs: S (set) makes the stored bit 1, and R (reset) makes it 0. With both at 0, the loop holds whatever it had.
> Loading the animation…
S
R
Next Q
What happens
0
0
Q
Hold
1
0
1
Set
0
1
0
Reset
1
1
—
Forbidden: Q and Q' stop being opposites
Watch outTry S = R = 1, then press Release both. Both gates change together, and the result depends on which gate is faster, so the outcome is unpredictable. Designs must never do this.
A NAND version works the same way with active-low inputs, so its forbidden input is S' = R' = 0.
Level and edge
A D latch removes the forbidden input by adding a data line D and an enable. While the enable is high, Q copies D. When it goes low, Q holds its last value. It's called transparent, because the input passes straight through while enabled.
That transparency is a problem. If a latch's output feeds back through logic to its own input, the signal can run round the loop many times while the clock is high. What we want is a device that changes only at one instant: a flip-flop, which responds to a clock edge.
> Loading the animation…
A flip-flop can be built from two latches back to back, called master and slave. While the clock is low, the master follows D and the slave holds. When the clock rises, the master freezes and the slave copies it. So the output only changes at the rising edge, and there's never a straight path from input to output.
A rising-edge D flip-flop. Q takes the value D had just before each rising edge. Glitches on D between edges don't matter.
The four flip-flop types
Each type differs only in how its inputs decide the next state. Q⁺ means the value of Q after the next clock edge.
Type
Next state
Used for
D
Q⁺ = D
Storing data. Nearly everything in modern design
T
Q⁺ = T ⊕ Q
Toggling, so counters
JK
Q⁺ = JQ' + K'Q
General purpose; J = K = 1 toggles
SR
Q⁺ = S + R'Q, with SR = 0
Mostly historic
> Loading the circuit…
Three flip-flops on one clock. Set the inputs, then press Clock edge.
Turn on t, and press Clock edge a few times. q_t toggles each time.
Set j and k both on: q_jk toggles too. Now try only j, then only k.
Change d without clocking. q_d doesn't move until the edge.
To design with flip-flops you need the reverse question: to go from this Q to that Q⁺, what must the inputs be? That's the excitation table, and an X means either value works.
Q
Q⁺
D
T
J
K
S
R
0
0
0
0
0
X
0
X
0
1
1
1
1
X
1
0
1
0
0
1
X
1
0
1
1
1
1
0
X
0
X
0
TipAll those X's are why JK flip-flops give small logic, and they feed K-maps as don't-cares. D is simplest to think about, since D is just the next state.
Practice
A JK flip-flop has Q = 1, J = 1 and K = 1. What's Q after three clock edges?
Converting between types
To turn a flip-flop of type X into type Y, work out what X's inputs must be, given what Y's inputs ask for. The method is the same each time.
Write Y's characteristic table: Y's inputs and Q, giving Q⁺.
Using X's excitation table, add the X input that produces each Q to Q⁺ change.
Minimise X's input as a function of Y's inputs and Q.
Asynchronous reset forces Q to 0 immediately, whatever the clock is doing.
Synchronous reset does it at the next clock edge, so it's just more logic in front of D.
Clock enable holds the value unless enabled, which is a MUX in front of D. The clock itself is left alone, because gating a clock with ordinary logic can create glitches that clock flip-flops by accident.
// A D flip-flop with asynchronous reset
always @(posedge clk or posedge rst)
if (rst) q <= 1'b0;
else q <= d;
Use <= (nonblocking) for flip-flops, so that every flip-flop in the design sees the old values at the edge. Chapter 20 explains why.
Flip-flops have timing rules too. D must be stable for a short setup time before the edge and a hold time after it, and Q changes a little after the edge. Chapter 17 is about what happens when those rules are broken.
Exam pointsGATE: fine to skip at first
Race-around: in a level-triggered JK latch with J = K = 1, the output toggles repeatedly for as long as the clock stays high. Master-slave and edge-triggered flip-flops solve this, because the output changes only once per cycle.
A T flip-flop with T = 1 divides its clock by 2. Chain n of them and you divide by 2ⁿ.
Practice · GATE-style
What's the next-state equation of a flip-flop made by connecting J = K = T?
Practice · GATE-style
To change Q from 1 to 0 in a JK flip-flop, what must J and K be?
Practice · GATE-style
A 1 MHz clock drives a chain of 4 toggling flip-flops. What frequency comes out of the last?
Practice · GATE-style
To make a T flip-flop from a D flip-flop, D should be…