Domino Logic Gates and its Advantages

       A problem with the dynamic logic is its limitation in cascading several stages. Domino logic, a modification of the dynamic logic, can be used to cascade several stages. The configuration of a domino-logic multiple-inverter gate is shown in Fig. 3.36. It can be seen from Fig. 3.36 that the circuit is the same as that of the dynamic logic gate with the addition of a CMOS inverter at the output. The addition of the inverter makes the domino logic to produce double inversion (i.e., one by the main transistor block, and the other by the inverter). Hence, only non-inverting gates are available in the domino logic system.

      The working of the circuit shown in Fig. 3.36 can be explained as follows. As in the case of the dynamic-logic NAND gate, here also, when the C = 0, both the NMOS inverter blocks NG1 and NG2 are in the disabled conditions. At this time, both the PMOS transistors TP1 and TP2  are in the ON condition and the outputs Z1 and X2, respectively, of NG1 and NG2 are at +VDD (≡ logic 1). These outputs Y1 and Y2, will be inverted by the respective CMOS inverters of I1 and I2 to give  their outputs Z1 = Z2 = 0.

      Now, when the C = 1, NMOS transistors TN2 and TN4 get turned-on, and PMOS transistors TP1 and TP2  get turned-off. The outputs Y1 and Y2 are now dependent on the input A only. If A = 1, then Y1 = A' = 0 and Z1 = A = 1. But, if A = 0, then the output Y1 = 1, and Z1 = 0. This produces double-inversion operation by gate NG1. It may be observed that this operation is repeated by gate NG2 to give its output Z2 = A. We thus notice that both the gates produce double inversions.

       The main aspect in the operation of the domino gate is that, initially as in the case of the dynamic logic, the load capacitor CL is pre-charged to +VDD through the PMOS transistor. When transition occurs due to the pulse input A, the output Y changes first. The CMOS inverter will then invert this to produce transition in the output Z. The inverter thus ensures that only one transition occurs for single-trigger inputs at A in the case of a singe-gate system.

      Now, suppose we cascade two such domino-logic inverter gates, as shown in Fig. 3.36. With a trigger input, Gate NG1 produces the required transition, and then after the stipulated propagation delay, Gate NG2 produces the final transition.

      If several such stages are cascaded in this fashion, the transition takes place one after another, just like thin wooden planks or playing cards fall in a domino. It may be noted that a domino is a special arrangement of several playing cards. In this arrangement, when we give a small push to the first card to the right (or left, depending on the domino arrangement), it falls onto the second card which in turn falls onto the third card and so on in that order. This is called the falling of the domino.

Advantages of Domino Logic Gates

Domino logic has the following advantages:

1.    Domino logic structure has much smaller chip area than CMOS structure.
2.    Since each gate has an inverter at its output, only one transition will take place for each triggering, as stated before. This avoids glitches (sudden unwanted transitions).
3.    Operating speeds are increased due to reduced value of load capacitance.


Dynamic NMOS (d-NMOS) Logic Gates

Dynamic NMOS (d-NMOS) Logic Gates:

We know that the number of components required to construct a static circuit can be reduced by converting it into a dynamic circuit. However, to keep the circuit dynamic, we require additional circuitry in the form of memory and clock circuits for which additional expenditure is required. Figure 3.35(a) shows a dynamic NMOS (d-NMOS) NAND gate.

The working of the d-NAND gate can be explained using the equivalent circuits shown in Figs. 3.35(b) and (c), respectively. Let initially the clock C = 0. Then the PMOS transistor Tp gets turned-on (represented by the ON-switch) and the NMOS transistor T3N gets turned-off [represented by an OFF-switch in Fig. 3.35(b)]. In this condition, the NAND operation is disabled. We also notice that in this condition the parasitic load capacitance CL gets charged from +VDD through Tp; this charging operation of CL is known as pre-charging. This is required for speeding up the operation of the gate.

Now, let C = 1. Then Tp gets turned-off and T3N gets turned-on, as shown in Fig. 3.35(c). In this condition, since CL has already been charged, the circuit will not take any additional time to charge it and its speed of operation increases; it also will start to act as a conventional NAND gate.

Advantages of Dynamic Logic Gates:

1. Reduce the static power dissipation of the pseudo-logic gates and other similar gates to zero by allowing the circuit to work only when clock pulses are present.
2. It can be seen that the number of active devices used in a given dynamic logic gate will be less than that of the corresponding CMOS gate. Therefore, the chip area required it for constructing the dynamic logic gate will be less than that of the corresponding CMOS gate.

Disadvantages of Dynamic Logic Gates:

1.   More complex circuitry is required for dynamic operation.
2.  Clock pulses are required for the operation of the circuit.
3.  Due to loading problems, it is difficult to cascade several stages.



Pseudo NMOS Logic Circuit

Even though CMOS logic gates have very low power dissipation, they have the following limitations:

1.    They occupy larger area than NMOS gates.
2.    Due to the larger area, they have larger capacitance.
3.    Larger capacitance leads to longer delay in switching.

These limitations of the CMOS gates can be reduced by several alternative structures discussed below. These structures resemble the CMOS structure in some way; yet, they are able to reduce the chip area and hence the capacitive delay. Pseudo-NMOS logic, dynamic NMOS logic, and domino logic are some of these special CMOS structures.


Pseudo-NMOS (p-NMOS) Logic Gates


Figure 3.32 shows a pseudo-NMOS inverter (p-NMOS NOT) gate, Fig. 3.33 shows a pseudo-NMOS NAND (p-NMOS NAND) gate, and Fig. 3.34 shows a pseudo-NMOS NOR (p-NMOS NOR) gate. As shown in all these figures, there is a block of NMOS FETs, which will contain one or more NMOS transistors, as required by the structure of the gate. However, there will be only one PMOS transistor in any pseudo-NMOS logic, and this will be always grounded.

The p-NMOS circuit is a modification of NMOS circuits with DMOS loads. In p-NMOS circuits, we use a PMOS transistor, instead of the DMOS transistor, as its load. The advantages of using a PMOS load are:

·    The circuit retains its basic CMOS structure. Hence, the chip area is minimum compared with the conventional CMOS structure.
·       The circuit becomes compatible with CMOS devices.
·       The channel resistance of the pseudo-NMOS devices is higher than that of the NMOS devices. Hence, power dissipation is lower for the pseudo devices.
·       Pseudo-NMOS circuits are useful in applications where the output remains in logic-1 state most of the time.

          However, the pseudo circuits have the disadvantage that they possess greater propagation delay than the NMOS devices. Pseudo NMOS Logic Circuit is shown below.




Advantages and Disadvantages of CMOS Circuit

The following are the advantages and disadvantages of CMOS circuit are as follows.

Advantages of CMOS Logic Gates

1.    Extremely large fan-out capability (>50).
2.    Lowest power dissipation of all gates (a few nW).
3.    Very high noise-immunity and noise-margin (typically, VDD/2) 
4.    Lower propagation delay than NMOS.
5.    Higher speed than NMOS. Currently, computer chips operating at (or more than) 4 GHz are available in the open market.
6.    Large logic swing (=VDD).
7.    Only a single power supply (+ VDD) is required.
8.    Directly compatible with TTL gates.
9.    Temperature stability is excellent.
10.    Low-voltage (1.5 V) chips are now available.

Disadvantages of CMOS Logic Gates

1.    Increased cost due to additional processing steps. But, this is being rectified.
2.    Packing density less than NMOS. Using Pass-Transistor logic structure, packing density comparable to or more than that of NMOS gate is possible.
3.  MOS chips must be protected from acquiring static charges by keeping the leads shorted. Static charges acquired in leads will destroy the chip. At present this problem has been rectified by using built-in protective devices or circuits.


Voltage Transfer Characteristic of CMOS Inverter

Voltage-Transfer Characteristic of CMOS Inverter

Figure 3.32(a) shows an experimental set-up to plot the input-output voltage-transfer characteristic of a CMOS inverter. For plotting the characteristic, CMOS inverter gates themselves can be used, or CMOS NAND/NOR gates converted into inverters (by short-circuiting their input terminals) can be used. Figure 3.31(a) shows such a connection.

The experiment is conducted by varying the input voltage from 0 to 15 V in steps of, say, 0.1 V each and measuring the corresponding output voltage. The input-output voltage transfer characteristic is then plotted, as shown in Fig. 3.31(b).

In Fig. 3.31 (b), we notice that, when the input voltage Vi VDD/2, the output voltage remains at VDD. This is because, when Vi VDD/2, the NMOS FET will not conduct, but the PMOS FET will conduct and this keeps Vo = +VDD.

When Vi  VDD/2, NMOS starts conducting and PMOS gets turned-off. The input voltage level for transition in CMOS is thus considered to be Vi = VDD/2. This is due to the fact that both the FETs (NMOS and PMOS) are complementary-identical transistors (that is, identical in all respects, except that they are complementary devices). Hence, the turn-on voltage of one device must be equal to the turn-off voltage of the other, which means that this should be half of the voltage swing. As CMOS devices are perfect switches with small ON-resistance, very low drain currents, and fast switching, we find that:

1.      Vo = +VDD, when Vi £ VDD/2 
2.      Vo = 0 V, when Vi ³ VDD/2
3.      Transition from +VDD to 0 volt occurs in zero time, showing a vertical edge in the transfer curve from the OFF-state to the ON-state.

Usually, CMOS gates operate on supply voltages ranging from +5 volts to +15 volts.

Noise Margin of CMOS Gates

Since the logic-1 level and logic-0 level of CMOS gates are +VDD and 0 volt, respectively, and the transition from 0-to-1 or 1-to-0 occurs at VDD/2, we conclude that CMOS gates can tolerate noise level up to VDD/2. Therefore:

                                      Noise margin of CMOS gatesVDD/2

Low Power Dissipation in CMOS Logic Gates

Low-Power Dissipation in CMOS Logic Gates

CMOS logic gates are currently the most popular forms of logic gates because of their advantages over other logic families. One of the main features of CMOS gates is its low-power dissipation capability. This feature is explained below.

CMOS gates are constructed, as explained above, with NMOS upon PMOS or vice versa. For example, consider the CMOS inverter shown in Fig. 3.26. This is redrawn here as shown in Fig. 3.30(a) for illustration. Figure 3.30(b) shows an approximate equivalent inverter network using mechanical switches. In this, we have represented the PMOS transistor using a closed switch and the NMOS transistor using an open switch. We also find that the two switches are ganged together by a push/pull lever, which can be operated to open the PMOS switch and close the NMOS switch or vice versa. Assuming push logic 0 and pull logic 1, we find that the circuit shown in Fig. 3.30(b) performs inversion.

It can be seen from Fig. 3.30(b) that only one of the switches remains closed and the other remains open at any instant under consideration,. This means that, no current will flow through the circuit at any instant of time, and hence no power loss will occur in the circuit. However this is too ambitious a statement to be realized in practice
In actual practice, a very small amount of leakage current will flow through the device, especially when the transistors are turned-on or turned-off. For example, let initially the PMOS be ON and the NMOS be OFF. We assume that the drain current ID = 0. Now, let a trigger applied to the gate of NMOS turns it ON suddenly. At this time, the PMOS, which was conducting previously, still continues to conduct for a little more time before it turns OFF. So, we find that both the transistors are conducting simultaneously for a very small interval of time. During this period some power will be drained from the battery and dissipated in the transistors. But this is usually negligibly small. 

Figure 3.31 shows two situations when both the transistors remain in the ON-state simultaneously; this occurs at transitions points TP1 and TP2, as shown. TP1 is the instant at which T1 jumps up from its ON-state to OFF-state and T2 jumps down from its OFF-state to ON-state. Similarly, TP2 is the instant at which T2 jumps up from its ON-state to OFF-state and T1 jumps down from its OFF-state to ON-state. 

From data manuals of CMOS gates, it is found that the drain current of CMOS transistors is in the range of nA to pA (10−9 to 10−12). Power dissipation of COMOS gates is about a few pW. With the advent of modern technology, packing density of CMOS gates has been found to be extremely high and the speed of operation has reached the GHz range.



Advantages and Disadvantages of NMOS Gates

The following are the advantages and disadvantages of NMOS gates.

Advantages of NMOS gates

1. The number of diffusions steps required is the lowest. Hence, cost per gate is the lowest.
2. Very low power dissipation (on the order of a few nW).
3. Power supply can be varied from 5 V to 15 V possible.
4. Only NMOS transistors are used, no isolation-islands are required. Hence, highest  packing density of all logic families. Suitable for VSLI, ULSI etc.
5. Large fan out capability (20 gates)
6. Very high noise-margin. Suitable for use in industrial atmosphere.
7. MOS circuits are used as capacitors. They can also be used as resistors.
8. MOS devices are used in charge-coupled devices (CCDs).

Disadvantages of NMOS gates

1. MOS transistors are also capacitors. Hence, speed of operation is lowest due to very high capacitive loading.
2. Large propagation delay per gate.
3. Higher power dissipation than CMOS gates.
4. This technology has not acquired as much popularity as the CMOS technology. CMOS, because of its lower power dissipation and higher speed of operation has replaced NMOS almost completely.


NMOS AND-OR-Invert Gate Circuit

NMOS AND-OR-Invert Gate Circuit
NMOS AOI Gate Circuit
Figure 3.25 shows a multiple-logic NMOS gate, called the AND-OR-INVERT (AOI) gate. This is a combination of two NMOS AND gates producing the functions of AB and CD. These two AND gates are connected in parallel so that they perform an ORing operation. This AND-OR function is then NOTed to produce the function Z = (AB+CD)′. As shown in Fig. 3.25, transistors T1 and T2 form one AND operation and T3 and T4 form the other. The parallel combination of these two pairs will produce the desired AOI output.


NMOS NOR Gate Circuit

The  NMOS NOR Gate Circuit:

Figure 3.24(a) shows a two-input NOR gate using NMOS FETs replacing the mechanical switches of the two-input NOR gate shown in Fig. 3.24(b).


As in the previous cases, switching transistors T1 and T2 are of the enhancement type and T3, which acts as the load resistance, is of the depletion type. If A = 0 and B = 0 (i.e., both A and B have 0-volt inputs), transistors T1 and T2 will remain OFF. Then, the output across the load Z = +VDD (logic 1). However, when A = 0 and B = 1, B = 0 and A = 1, or A = B = 1, either T1 or T2 or both T1 and T2 will conduct as the case may be and the output Z = 0 volt (logic 0). These conditions are summarized as given in Table 3.5. The logic relations given in Table 3.5 that the logic gate shown in Fig. 3.24(a) performs the positive-logic NOR operation. 

Table 3.5 NMOS NOR

      A
      B
      Z
      0
      0
      1
      1
      0
      1
      0
      1
      1
      0
      0
      0

NMOS OR Gate Circuit

The NMOS OR Gate Circuit 

Figure 3.23(a) shows a two-input OR gate using NMOS FETs and 3.23(b) shows its equivalent using switches. It may be noticed in this case also that NMOS transistors are used to replace the mechanical switches and the load resistor.  We find that, if A = 0 and B = 0 (i.e., A and B have 0-volt inputs), transistors T1and T2 will remain OFF. Then, output Z = 0 volt (logic 0). However, when A = 0 and B = 1, or B = 0 and A = 1, or A = B = 1, either one or both of transistors T1 and T2 conduct and develop +VDD, across the load transistor T3. This means that the output Z = logic 1. This may be summarized as given in Table 3.4. The logic relations given in Table 3.4 are seen to be that of the OR operation. So, we conclude that the logic gate shown in Fig. 3.23(a) performs the positive-logic OR function.
Table 3.4 NMOS OR

      A
      B
      Z
      0
      0
      1
      1
      0
      1
      0
      1
      0
      1
      1
      1

NMOS NAND Gate Circuit

The NMOS NAND Gate Circuit

Figure 3.22(a) shows a two-input NMOS NAND gate circuit. This circuit is a modification of the NAND gate using mechanical switches shown in Fig. 3.22(b). The mechanical switches of Fig. 3.22(b) are replaced with NMOS transistors in Fig. 3.22(a).  NMOS transistors T2 and T3 are of the enhancement type and T1, which acts as the load resistance, is of the depletion type. Note that this load is located on top of the switching transistors T2 and T3 to produce inversion. If A = 0 and B = 0 (i.e., A and B have 0-volt inputs), transistors T2 and T3 will remain OFF. Then, the output Z = +VDD (logic 1). The same situation prevails, when A =  0 and B = 1, or B = 0 and A = 1. Now, if A = B = 1, then both T2 and T3 will conduct and the output Z = 0. These logic operations are summarized as given in Table 3.3, which suggests that the gate shown in Fig. 3.22(a) is a positive-logic NAND gate. To get a negative-logic NAND gate, we replace the NMOS transistors with PMOS transistors.


Table 3.3 NMOS NAND

      A
      B
      Z
      0
      0
      1
      1
      0
      1
      0
      1
      1
      1
      1
      0

NMOS AND Gate Circuit

The NMOS AND Gate Circuit

Figure 3.21(a) shows a two-input AND gate using NMOS field-effect transistors. This diagram is a modification of the AND gate using switches shown in Fig. 3.21(b). The AND gate constructed using mechanical switches was discussed in Section 24 using Fig. 2.4. Figure 3.21(b) is a reproduction of Fig. 2.4 for reference. The mechanical switches of 3.21(b) are replaced with NMOS transistors in Fig. 3.21(a). We find that in Fig. 3.21(a), if A = 0 and B = 0 (i.e., A and B have 0-volt inputs), transistors T1 and T2 will remain in the OFF-state. Then, output Z = 0 volt (logic 0). The same situation prevails, when A = 0 and B = 1, or B = 0 and A = 1 (here logic 0 = 0 volt). Now, if A = B = 1, then both T1 and T2 conduct and develop +VDD, across the load transistor T3. This means that the output Z = logic 1. These may be summarized as given Table 3.2. The entries given in Table 3.2 are found to be that of the AND operation. So, we conclude that the logic gate shown in Fig. 3.21(a)-  performs the positive-logic AND operation.


Table 3.2 NMOS AND

      A
      B
      Z
      0
      0
      1
      1
      0
      1
      0
      1
      0
      0
      0
      1