IAE Academy

FOR CORE AND IT JOBS · CHAPTER 2 OF 7

Static CMOS logic gates

This chapter shows how to turn a Boolean expression into a static CMOS gate, how to size its transistors so rise and fall take equally long, and how transmission gates and pass transistors pass signals. You need it to read or draw a gate at transistor level and to judge its speed and area.

Updated 30 Oct 2026

What you will be able to do

You will turn a Boolean expression into a table of series and parallel nMOS groups, then write the pMOS network as its dual. You will count transistors and size them so the gate pulls up and down through the same resistance as a reference inverter. You will build a 2:1 multiplexer from transmission gates and say what goes wrong with a single pass transistor.

Before you start

  1. Chapter 1, The MOSFET as a switch and the CMOS inverter

  2. Boolean algebra and De Morgan's laws

  3. Resistors in series and in parallel

  4. Charging a capacitor through a resistor (time constant RC)

Building a gate from an expression

From Chapter 1, an nMOS conducts when its gate is high and a pMOS conducts when its gate is low. A static gate has an nMOS pull-down network between the output and ground, and a pMOS pull-up network between the output and . For every input pattern exactly one network must conduct. Two nMOS in series conduct only if both inputs are high, which is an AND. Two in parallel conduct if either is high, which is an OR.

OUTPUT OF A STATIC GATE

For a single-stage gate, with written as AND and OR of the uncomplemented inputs . The pull-down builds (series for AND, parallel for OR) and pulls to 0 exactly when . The pull-up, its dual with series and parallel swapped, pulls to exactly when .

DUALITY OF THE TWO NETWORKS

The pull-up must conduct exactly when , and a pMOS with input conducts when . These laws write in complemented inputs with every AND and OR swapped. So, at every level, a series pair of nMOS becomes a parallel pair of pMOS and a parallel pair a series pair. are logic inputs.

Gate

Y

nMOS groups

pMOS groups

Transistors

NAND2

A series B

A parallel B

4

NOR2

A parallel B

A series B

4

AOI21

(A series B) parallel C

(A parallel B) series C

6

OAI21

(A parallel B) series C

(A series B) parallel C

6

AOI22

(A series B) parallel (C series D)

(A parallel B) series (C parallel D)

8

When each input appears once in the expression, it drives one nMOS and one pMOS, so a gate with inputs has transistors. AOI stands for AND-OR-INVERT and OAI for OR-AND-INVERT. AOI21 needs 6 transistors in one stage. From separate gates, a NAND2 (4) for , an inverter (2) for and a NOR2 (4) of and , it needs 10 in three stages.

WATCH OUT FOR

Copying the nMOS arrangement into the pMOS side. Test: in a NAND2 set , ; must be 1. With series pMOS, B's pMOS is off (its gate is high) and A's nMOS leaves the pull-down open, so connects to neither supply.

WORKED EXAMPLE

Build the gate for . Count its transistors and test two input patterns.

Solution
  1. The pull-down conducts exactly when , so build it level by level: B parallel C, then in series with that pair, then in parallel with the whole branch.

  2. The pull-up is the dual, so swap at every level: in series, in parallel, and in series.

  3. Each of the 4 inputs appears once, so there are transistors.

  4. Test : the nMOS of and give a path to ground. In the pull-up, 's pMOS is off and the B-C branch has 's off, so it is open. , matching .

  5. Test : 's and 's nMOS are off, so no path reaches ground. 's and 's pMOS are on and in series, so , matching .

Answer: Pull-down: A in series with (B parallel C), all in parallel with D. Pull-up: A in parallel with (B series C), all in series with D. Eight transistors.

Sizing for equal rise and fall

You size a gate against a reference inverter, the inverter whose rise and fall delays the gate must match. The output's load capacitance (in fF: the inputs it drives plus wiring) charges or discharges through the on transistors. So its delay scales with the time constant , where is the resistance to the supply it is pulled to.

For small , Chapter 1's linear-region current , with (carrier mobility , oxide capacitance per area ), the width-to-length ratio and the threshold, loses its squared term: . So is a resistance.

ON RESISTANCE OF A TRANSISTOR

Holds for an on device with (pMOS: ) and . It is from the line above. Use , for nMOS and , for pMOS. in A/V², voltages in V, in Ω.

WIDTH RATIO FOR EQUAL RISE AND FALL

Set in the reference inverter. With , the last step holds when and both types have the same and , giving . , are the electron and hole mobilities.

Electrons are more mobile than holes, so is above 1, often 2 to 3 in silicon, and a pMOS must be that much wider to carry the same current. Call this ratio . In this RC model, equal resistance gives equal rise and fall times. A gate sized as below matches the inverter's resistance, but its wider transistors add drain capacitance at the output, so its delays are a little longer than the inverter's.

WIDTH OF A TRANSISTOR IN A SERIES STACK

At fixed , , so transistors in series, each times wider, have resistance . is the number of series devices on the longest path to the supply through the device. is the reference or , in µm.

A device shared by several paths is sized for the longest path through it, so a shorter path comes out stronger than needed, never weaker. Parallel devices stay at the reference width, because in the worst case only one of them conducts. So in a NAND with inputs each nMOS is and each pMOS ; in a NOR with inputs each nMOS is and each pMOS .

WORKED EXAMPLE

The reference inverter has and with equal channel length. Size a NAND3 and a NOR3 for the same rise and fall delay as it, and compare their total width.

Solution
  1. Here .

  2. NAND3: the three nMOS are in series, so each is . The pMOS are in parallel and in the worst case one conducts, so each stays .

  3. NAND3 total: .

  4. NOR3: the parallel nMOS stay each. The three series pMOS are each .

  5. NOR3 total: . The NOR stacks the slow, wide pMOS, so it costs more area for the same delay.

Answer: NAND3: nMOS 3 µm, pMOS 2 µm, total 15 µm. NOR3: nMOS 1 µm, pMOS 6 µm, total 21 µm.

Transmission gates and pass transistors

A pass transistor connects a signal input to a node, not an output to a supply. It conducts both ways: its source is the lower-voltage channel end for an nMOS and the higher one for a pMOS, and Chapter 1's is taken from that end. An nMOS passing a high starts with the output low, so the output is the source and shrinks as it rises. At it turns off, short of . A pMOS passing a low stops short of 0 the same way.

LEVELS A SINGLE TRANSISTOR PASSES

The output drives only a capacitance, starting at the other level. nMOS, gate and input at : conducts while . pMOS, gate and input at 0: conducts while . Voltages in V. The threshold rises when the source moves away from the body (the substrate, at ground for nMOS and for pMOS), the body effect; these numbers neglect it, so the real drop is larger.

A transmission gate puts an nMOS and a pMOS in parallel between input and output. The nMOS gate is driven by and the pMOS gate by . With the nMOS passes a low level fully and the pMOS a high level, so the full swing passes. With both are off. The cost is two transistors and two control signals.

2:1 MULTIPLEXER FROM TRANSMISSION GATES

, are the data inputs and the select. M1 and M2 pass when , M3 and M4 pass when , and their outputs join at . A pass transistor conducts both ways, so the table lists its channel ends as End 1 and End 2. M5 (pMOS) and M6 (nMOS) form the select inverter, input , output : six transistors in all.

Transistor

Type

End 1

Gate

End 2

M1

nMOS

A

Y

M2

pMOS

A

S

Y

M3

nMOS

B

S

Y

M4

pMOS

B

Y

Pass-transistor logic builds functions from nMOS switches alone, which saves transistors. A 2:1 multiplexer needs 4: two pass nMOS (gate passing to , gate passing to ) and the pMOS and nMOS of the select inverter, an ordinary inverter.

The price is the threshold drop: a high passed by an nMOS reaches only . Let drive an inverter, the next stage. Its pMOS has , so it stays on when . Its nMOS has , above when , so it is on too. Both on give a static current from to ground. The fix is one weak pMOS: source at , drain at , gate at the inverter's output. With high the output is low, so it turns on and pulls to ; being weak, it lets the pass nMOS still pull low.

Check yourself

CHECK YOURSELF 1

A static CMOS gate has . Which describes its pMOS pull-up network?

  1. (A)

    A in parallel with the series pair B, C

  2. (B)

    A in series with the parallel pair B, C

  3. (C)

    All three in series

  4. (D)

    All three in parallel

Answer and solution

Answer: (B)

  1. The pull-down network conducts when : A in parallel with (B series C).

  2. Dualise: parallel becomes series and series becomes parallel.

  3. So the pull-up is A in series with (B parallel C).

CHECK YOURSELF 2

A 4:1 multiplexer is a tree of three 2:1 transmission-gate multiplexers: two on the first level share the select , and one on the second level uses . Each select line has one inverter making its complement, shared by every multiplexer on that line. How many transistors does it have?

  1. (A)

    12

  2. (B)

    14

  3. (C)

    16

  4. (D)

    18

Answer and solution

Answer: (C)

  1. Each 2:1 multiplexer has two transmission gates of 2 transistors each, so 4. Three of them give 12.

  2. Two select lines need two inverters of 2 transistors each: 4.

  3. Total: .

CHECK YOURSELF 3

With V and V (body effect neglected), an nMOS pass transistor has its gate at 1.8 V and its input at 1.8 V. The output node starts at 0 V and drives only a capacitance. What is the final output voltage in V?

Answer and solution

Answer: (accepted range )

  1. The output node is the source, so the nMOS conducts while .

  2. It stops when .

  3. V.

CHECK YOURSELF 4

A reference inverter has . Take , , and equal and channel length for both types. An AOI21 gate, , is sized by this chapter's series-stack rule for the same rise and fall delay as the inverter. What is its total transistor width in µm?

Answer and solution

Answer: (accepted range )

  1. Reference pMOS: .

  2. Pull-down: (A series B) parallel C. A and B lie on a 2-transistor path, so each is . C is alone on its path and stays 0.4 µm. nMOS total: 2.0 µm.

  3. Pull-up: (A parallel B) series C. Every path to has 2 pMOS in series, so each is . pMOS total: 6.0 µm.

  4. Total: .

Used in these GATE topics