Converters
Prudhvi Raj Thota and Ashis Kumar Mal
Department of Electronics and Communication Engineering
National Institute of Technology, Durgapur-713209, India
prudhvi691@gmail.com akmal@ece.nitdgp.ac.in
I. I NTRODUCTION
Analog-to-digital converters (ADCs) demand a
synchronous counter that is capable of operating efficiently
for wide range of frequencies[1,2] . Besides its necessity
in ADCs, it is found to be the basic block for various
applications in the field of VLSI design, communication
and measuring systems. Thus, the extensive studies are
going on the design of high speed low power counters.
Synchronous counters for low to moderate frequency Fig. 2. TSPC based Positive-edge Triggered Divide-by-two circuit
applications, are usually build using stage by stage, each
containing combinational logic and T Flip-Flop (TFF) or
D Flip-Flop (DFF). Although a common clock triggeres all To minimize the tclk , various techniques have been
TFFs simultaneously as shown in Fig. 1, the decision of proposed. The tplogic being the main factor due to
toggling depends on the input T, which is driven by the its increase in proportion with the number of stages,
combinational logic. The combinational logic is usually the several efficient counters have been proposed based on
carry propogation chain having the inputs from the outputs Manchester carry chain, carry-look-ahead and binary tree
of previous stages. carry propagation to decrease tplogic , but they occupy
larger area and dissipate more power.
The counting rate of synchronous counter is bounded
by the delay of combinational logic (tplogic ), setup time Even Transmission Gate (TG) based TFF does not
(tsetup ) and the propagation delay (tpT F F ) of TFF. Thus, minimize propogation delay (tpT F F ) to the greater extent.
the clock period of a conventional counter (tclk ) can be Yuan[3] proposed a true single phase clock (TSPC) based
expressed as [6] counter capable of operating at 250 MHz with transistor
sizing in a 3 m CMOS process. However, the speed
tclk tplogic + tsetup + tpT F F (1) improvement allowed by the technolgy scaling has made
978-1-4673-6621-2/16/$31.00
c 2016 IEEE this counter (transistors sized in scale with the sizes in [3])
Fig. 3. TSPC based Positive-edge Triggered Toggle Flip-Flop
to operate at near about 4.44 GHz in 0.18 m CMOS At the positive edge of the clock transition, the
process. With a small modification in the Yuans counter N-dynamic block is in evaluation phase.
design[3] , this paper proposes an efficient synchronous If the evaluation result is at logic 1, the NMOS
counter that can be operated upto a maximum frequency of (M7 ) of the N-C2 MOS stage is ON and makes the
4.54 GHz. output as logic 0. This is same as the precharge
phase, so no change occurs at the following nega-
The rest of the paper is organized as follows. Section II tive edge of the clock transition.
gives the review on the basic building block i.e., TFF If the evaluation result is at logic 0, the PMOS
facilitating the asynchronous inputs like preset and clear. (M9 ) of the N-C2 MOS stage is ON and makes
Section III introduces the proposed counters including the the output as logic 1. When the N-dynamic
way of embedding the combinational carry chain logic along block starts to precharge, at the subsequent clock
with the TFF. Simulation results are discussed in Section IV, transition, the clocked NMOS (M8 ) is OFF. So, the
followed by a conclusion in Section V. ON result is latched at the output of the N-C2 MOS
stage.
II. BASIC B UILDING B LOCK : T OGGLE F LIP -F LOP An extra unit inverter load (M10 and M11 ) increases the
Fig. 2 shows a basic nine-transistor (M1 to M9 ) TSPC current driving ability and gives out the non-inverting output
based T Flip-Flop proposed by Yuan[3] . It follows the with level restored.
single-phase precharge logic, which is an alternate se-
A. Csk as the Input of Toggle Flip-Flop
quence of P-units (comprising of P-dynamic block fol-
lowed by P-C2 MOS latch stage) and N-units (comprising Placing the transistors M12 and M13 as shown in Fig. 3
of N-dynamic block followed by N-C2 MOS latch stage). provides the input to the flip-flop. Yuan[3] suggested that this
It comprises of three stages : P-C2 MOS stage for latching, transistor is to be inserted at a lower position to the clocked
N-dynamic block for precharging and evaluating the logic, transistor in the last stage, so that no charge sharing takes
followed by N-C2 MOS stage for latching. The output of place at the output node.
last (N-C2 MOS) stage is fed back as the input to first If Csk is at logic 1, it operates in toggle mode as
(P-C2 MOS) stage, still satisfying the rules of single-phase shown in Fig. 2.
precharge logic. The operation of the circuit is as follows: If Csk is at logic 0, the N-stage is completely OFF
The first stage latches the input data by every positive and thus remains in the previous state.
edge of the clock (Clk) signal.
When the clock is at logic 0, the N-dynamic block is B. Asynchronous Inputs
precharged and its N-C2 MOS stage is OFF, as PMOS Two asynchronous inputs (Preset and Clear) on a flip-flop
(M7 ) and one NMOS (M8 ) are OFF. The output is have control over the outputs regardless of clock and input
therefore stable. status i.e., they are given higher priority than any other
TABLE I
T RUTH TABLE OF T OGGLE F LIP -F LOP
Csk qk
0 qk1
1 q k1
C. Toggle State with Static locking loop Fig. 5. 2:1 TG based Multiplexer facilitating the counter with U p/Down
mode control
The dynamic condition is possibly broken at the Most
Significant Bit (MSB) or at low clock frequencies, due to
the problems like charge sharing and leakage. This problem Least significat bit can be generated using either a simple
can be solved by placing a weak PMOS (M16 ) and weak divide-by-two circuit (Fig. 2) or TSPC based TFF (Fig. 3)
NMOS (M17 ) as shown in Fig. 3. producing Cs0 from E-TSPC based carry stage (Fig. 4).
When the clock is at logic 0, these transistors together Since each stage is enabled only when the carry is high,
with the first three transistors (M1 , M2 and M3 ) forms the power consumption of the most significant bit stages
a static dual-inverter locking loop, making the output are substantially reduced.
unable to change and remain in the previous state.
When the clock is at logic 1, the loop will be broken TABLE II
and hence allowing the change in the output. F UNCTION TABLE OF P ROPOSED C OUNTER
As weak devices have slower performance, they does not Enable Preset Clear Comment on output
effect the operation at high frequencies. If the circuit is 0 0 0 Resets all bits to logic 0
desired to be operated always at high frequencies, these 0 0 1 Undetermined
transistors (M16 and M17 ) could be removed. 0 1 0 Counts up by one
0 1 1 Sets all bits to logic 1
III. P ROPOSED C OUNTER D ESIGNS
1 0 0 Resets all bits to logic 0
A. Synchronous Counter 1 0 1 Undetermined
Synchronous counter is designed using TFFs having 1 1 0 Holds its previous state
Csk as one of the input to its N-unit. The carry logic 1 1 1 Sets all bits to logic 1
that drives Csk should be P-unit with the inputs from
N-units (previous stages), so as to make it single-phase 1) Down Counter: The positive edge triggered
precharge logic. The carry stage used by Yuan[3] is synchronous down counter can be obtained by providing
the P-unit consisting of P-dynamic block followed by the carry stage with non-inverting inputs (qk=0,1,...(n1) )
P-C2 MOS stage. An Extended True Single-Phase Clock instead of inverting inputs (q k=0,1,...(n1) ).
(E-TSPC) circuit technique: NMOS like P-latch block[7]
allow building P-unit with speed comparable to NMOS. 2) Up / Down Counter: Up / Down count feature can be
So, NMOS like P-latch block based combinational carry added using 2:1 Transmission (TG) based multiplexer with
unit is designed to make Csk equal to one whenever q 0 , qk and q k as data inputs and U p/Down as selection input
q 1 , ... and q n1 are logic 0s. as shown in Fig. 5. The output bk=0,1,...(n2) is fed as the
input to carry stage instead of feeding in qk=0,1,...(n2) or
NMOS like blocks may suffer from charge-sharing q k=0,1,...(n2) directly.
problem[8] , but in this case this problem does not exist as
there are no dynamic nodes. Further, NMOS like blocks can 3) Negative-edge Triggered Counter: Negative edge
also be used to bulid divide-by-2 circuit, but the problem triggered counter can be realized on holding the same
of charge sharing does not allow to provide Csk as input single-phase precharge logic using TSPC based TFF
and hence cannot be used in synchronous counter design. (N-C2 MOS stage, P-dynamic block, P-C2 MOS stage)
shown in Fig. 6 and E-TSPC based carry stage (NMOS
Fig. 6. TSPC based Negative-edge Triggered Toggle Flip-Flop
a Resimulated result