"Metal Oxide Semiconductor Field Effect Transistors
"Learn about MOSFET's & the CD4007: how they are made, how
they work, and how to put them to work in practical circuits."
Rewritten and modified by Tony van Roon
The Metal-Oxide-Semiconductor Field-Effect Transistor, or MOSFET for short, is similar in many ways to the junction
FET (JFET). Both are voltage-driven unipolar devices that depend on either electron or hole movement--but not both, as
does the bipolar transistor. However, there is a fundamental structural difference between the two field-effect
transistors: The JFET has three layers while the MOSFET has two. The MOSFET's simpler construction has give it a
performance edge over the JFET, and made it the world's most popular transistor style.
Transistor Tutorial, Part 10: Working with MOSFET's
Earlier articles in this series stated that the MOSFET's controlling gate voltage is applied directly to its channel
region across a thin layer of insulating oxide, as shown in Fig. 1-a. This
geometry contrasts with that of the JFET, which is controlled by switching an internal PN junction. The MOSFET will
work from lower power that the JFET, and its simpler design is reflected in lower production costs. That is why it
has become the basis for all CMOS digital logic IC families.
Part 9 and Part 11 in this series discussed FET's. The basic principles of JFET's and MOSFET's were explained in
the first article of (Electronics Now, February 1993) and the words that describe them were defined. The second
article (Electronics Now, March 1993) focused on JFET's, and included practical JFET circuit schematics.
This article concentrates on the enhancement-mode MOSFET, and it includes practical MOSFET circuit schematics base
upon small-signal MOS transistors available in a low-cost CMOS integrated circuit. You might wish to review the first
two articles to refresh your general knowledge of FET's before reading this article.
MOSFET Basics:
There are both N- and P-channel MOSFET's just as there are both N- and P-channel JFET's.
In the cross-section view of an N-type enhancement-mode transistor, Fig. 1a, you can see the thin layer of
silicon dioxide (glass) that electrically isolates the metal gate from the channel between the N-doped source and
drain regions. The presence of that insulated gate is why the MOSFET has also been called and IGFET (for insulated
gate FET). However, that term is now considered obsolete.
As shown in Fig. 1a, the channel between the N-type source and drain of an N-channel, enhancement-mode MOSFET
is the substrate P-type material. This MOSFET can be turned on so that current flows between source and drain only
when a positive forward bias is placed on the gate. As a result, the enhancement-mode MOSFET is said to be "normally
off". Its operation depends on the electron flow.
Recall, from Part 9's article, that all JFET's are depletion-mode or "normally on" devices. They are turned 'off' by
applying reverse bias. Depletion-mode MOSFET's are being made today for high-frequency radio applications.
A P-channel, enhancement-mode MOSFET has a cross section that is identical to that shown in Fig. 1-a except that
the substrate is N-type material and the source and drain regions are P-type material. A negative forward bias is
needed to turn a P-channel, enhancement-mode MOSFET ON. Its operation depends on the movement
of holes, which have lower mobility than electrons. This means that an N-channel MOSFET can switch faster than
a P-channel MOSFET.
As in the JFET, signal voltages or biases applied between the gate and source terminals of the MOSFET control the
magnitudes of signal currents flowing between the drain and source terminals.
N- and P-channel MOSFET's are said to be complementary because the doping of their substrate, source, and
drain materials as well as their forward bias polarities are opposite.
However, as will be seen, advantage is taken of those characteristics in complementary MOS (CMOS) logic families and
some of the circuit is discussed here.
Figure 1-b is the schematic symbol for an enhancement-mode N-channel MOSFET. The
dotted vertical line between the drain and the source represents a "normally-off" channel. (The symbol for
complementary P-channel device is similar except that its arrow points outwards.)
Figure 2-a shows a cross section of a monolithic CMOS IC with both N- and P-channel
MOSFET's integrated on the same substrate. The drains of both FET'S are connected. Fig. 1-b shows the
schematic symbols for N- and P-channel MOSFET's that are integrated into the CMOS chip shown in Fig. 1-a. A CMOS IC
provides the small-signal transistors needed for the experiments described in this article.
In passing, it is worth stating that, although they are organized in a similar manner, the most advanced CMOS logic
families have polysilicon rather than metal electrodes. Polysilicon, a pure form of silicon, is a conductor.
Figure 3 shows typical drain-to-source output characteristic curves for an N-channel,
enhancement-mode MOSFET. Drain current (ID) increases with increases in positive
gate-to-source bias.
These curves were obtained from a MOSFET transistor within a CMOS circuit that will be discussed in detail later.
The characteristic curves of a P-channel MOSFET are similar except that drain current increases as its bias becomes
more negative.
Figure 4 illustrates the gate-to-source transfer characteristics for the same
enhancement-mode, N-channel MOSFET shown in Fig. 3. It shows how drain current (ID)
increases directly (and almost linearly) with positive gate-to-source voltage (VGS),
while the DC supply voltage (VDD) remains constant at 15 volts. Note that no
significant drain current flows until the gate voltage rises to a threshold (VTH)
value of a few volts. (This can also be seen if Fig. 3).
Before proceeding with this discussion of MOSFET's, it is important to point out that protective precautions must be
taken when handling all MOSFET devices. All MOSFET's are susceptible to damage from the discharge of
electrostatic energy between any two pins.
Th extremely high input impedance of these devices lends itself readily to the buildup of electrostatic charges.
Because the oxide that insulates the gate of a MOSFET typically breaks down with the application of about 80 volts,
damage or destruction of the devices can be caused by higher levels of ElectroStatic Discharge (ESD).
Protective circuitry has been built into many discrete MOSFET's and protective networks are now included in most
(if not all) current manufactured CMOS IC's. However, the high electrostatic charge generated simply by scuffing your
shoes on a carpet and then touching the pins of a "protected" device can overwhelm those defenses and damage or
destroy the part.
Ideally, all MOSFET parts handling should be done only at a workstation organized to protect against ESD. In the
event that is it is not practical, at the very least a grounded wrist strap should be worn, and all CMOS parts handling
should be done on a grounded conductive work surface.
Complementary Pair:
A small-scale, industry-standard CMOS IC that contains a number of accessible N- and P-channel enhancement-mode
transistors is an excellent source of MOSFET's for experiments. Figure 5 shows the
schematic of a CD4007UB, a dual complementary pair plus inverter. It has six accessible MOSFET transistors: two pairs
are unconnected and the third pair is connected as a CMOS inverter or NOT gate.
There are many sources for the CD4007UB, know generically as the 4007. Prefixes identify the manufacturer. The
CD4007UB, for example, is made by Harris Semiconductor: it is pin-for-pin compatible with the Motorola MC14007UB.
These versatile parts are useful as digital logic parts as well as for linear applications in amplifiers,
pulse-shapers, and crystal oscillators.
The suffix "UB" indicates a CMOS series whose gates and inverters are constructed with a single inverting stage between
input and output, which results in decreased gain. However, this characteristic is useful when these normally digital
logic products are operated in the "linear" regions of their characteristic curves.
Figure 5 is the schematic for the 4007UB in a 14-pin DIP package. The MOSFET's have been labeled Q1 to Q6.
Transistors Q1, Q3, and Q5 are P-channel devices, while Q2, Q4, and Q6 are N-channel devices. Pin access is given to
all three terminals of transistors Q1 to Q4, but transistors Q5 and Q6 are permanently configures as an inverter.
The typical output and transfer characteristics of Figs. 3 and 4 were obtained from an N-channel MOS transistor Q2 of
a 4007UB.
Figure 6 is the pin assignment diagram for the 4007UB. It has been supplemented with
labels that relate pin numbers to the function they perform in the schematic. Fig. 5. Table-1 presents the outstanding
characteristics of the 4007UB.
Figure 7 shows the protection network for the CD4007UB. Input diode D2 is a distributed
resistor-diode network that appears as two diodes to (VDD)
Application Rules:
Here are some simple rules to keep in mind when working with the 4007UB:
In any application, all unused elements of the device must be disabled.
Complementary pairs of MOSFET's can be disabled by connecting them as standard
CMOS inverters, as shown in Fig. 8 (gate-to-gate and source-to-source), and
grounding their inputs. Refer to Fig. 5 to interpret the pin numbers given in
the figure. (The triangle symbol used here to designate a complementary pair of
transistors is the digital logic symbol for an inverter.)
Individual MOSFET's can be disabled by connecting their sources to their
substrates and leaving their drains open-circuited.
Input terminals must not be allowed to rise above the supply voltage
(VDD) or fall below zero volts (VSS).
To use an N-channel MOSFET, the source must be tied directly to (VSS) or
through current-limiting resistor. Similarly, to use a P-channel MOSFET, the
source must be tied directly to (VDD) or through a
current-limiting resistor.
Linear Operation:
Figure 9 shows how to connect Q2, an N-channel MOSFET in the 4007UB as a linear inverting
(common-source) amplifier. Resistor R1 is the drain load of Q2, and series resistors R2 and
(RX) form a voltage divider that biases the gate so that Q2 operates in the linear
region.
The value of (RX) is selected to give the desired quiescent drain voltage. It is
normally in the range 18,000 to 100,000 ohms.
Figure 10 shows how the Fig. 9 circuit is modified to give it very high input impedance.
The 10-megoHm isolating resistor R3 is placed between the function of resistors R2 and (RX)
and the gate of Q2.
Figure 11 show how to connect Q2 as a unity-gain, non-inverting (common-drain) amplifier
or source-follower. If the gate of Q2 is biased at half-supply voltage by the voltage divider made up of R2
and R3, the source pin assumes a quiescent value that is slightly more than the threshold voltage
VTH below the gate value. The circuit has an input impedance equal to the values
of resistors R2 and R3 in parallel (50,000), but this value can easily be increased to greater than 10 megohms by
inserting resistor R4 as shown in the figure.
Alternatively, the input impedance of the circuit in Fig. 11 can be raised to several hundred megoHms with the
"bootstrapped" source-follower configuration shown in fig. 12. The output signal from Q2 is fed back to the junction
through capacitor C1. As a result, near-identical input signals appear at each end of resistor R4, which, in turn,
passes near-zero signal current and appears (to the input signal) as a near-infinite impedance.
It can be seen from the previous descriptions that an enhancement-mode MOSFET acts like a bipolar transistor, except
that:
It exhibits very high input impedance.
It has a self-limiting drain-to-source current.
It has a substantially larger input-off-set voltage than a bipolar transistor.
The base-to-emitter offset of a bipolar transistor is typically 600 millivolts, while the gate-to-source offset of a
MOSFET is typically 2 volts. If one allows for those differences, a small-signal, enhancement-mode MOSFET can replace
a small-signal bipolar transistor in many kinds of bipolar transistor circuits.
The CMOS Inverter:
The most basic CMOS circuit is a complementary pair of N- and P-channel MOSFET's connected in series to form an
inverter. The inverter shown in Fig. 13-a was specifically intended for digital
circuitry where it performs the NOT operation. Fig. 2 is a section view of N- and P-channel
MOSFET's integrated on the same chip with a common drain connection. This device can be converted to the CMOS inverter
shown in Fig. 13-a by connecting the gates of P-channel MOSFET Q1 and N-channel MOSFET Q2 to form input
terminal VIN, and taking the output VOUT from the
common drain. The source of Q1 is connected to the positive power supply VDD, and
the source of Q2 is grounded at VSS.
Consider the P-channel MOSFET Q1 to be the driver and the N-channel MOSFET Q2 to be the load. Recall that an N-channel
MOSFET conducts with a positive gate bias, and a P-channel enhancement-mode MOSFET conducts with a negative gate bias.
When the voltage at the input terminal VIN is low (logic zero), the voltage on the
gate of Q1 is negative, causing it to conduct and short the supply voltage VDD to the
output terminal VOUT because Q2 is OFF (its gate voltage zero),
a high-impedance path exists between VOUT to ground VSS.
As a result the voltage at VOUT is VDD.
Alternately, when the input voltage is high (logic 1), the situation is reversed: Q1 is cut OFF,
forming a high-impedance path between VDD and VOUTS
and Q2 conducts, forming a low-impedance path from VOUT to ground VSS,
causing the output voltage to fall to zero.
This response makes the circuit a logic inverter or NOT gate. As can be seen in the truth table,
Fig. 13-b, a low (logic 0) input results in a high (logic 1) output; conversely, a
high (logic 1) input results in a low (logic 0) output.
In either logic state one enhancement-mode MOSFET is ON while the other is OFF.
Because of this, the quiescent current of a CMOS inverter is extremely low. It is this quality that gives the CMOS digital
logic IC families their many advantages.
Figure 13-c is the accepted logic symbol for a NOT gate.
(This symbol was used in Fig. 8 to simplify the discussion of disabling unused complementary pairs).
Although the CMOS digital inverter consumes zero quiescent current, it can source (feed) or sink (absorb)
significant current into or out from external loads. When the input is at logic-0, the output is effectively shorted
by Q1 to the positive power supply, so substantial current can feed through Q1 into a load connected to its output.
When the input to the digital inverter is at logic-1, the output is effectively shorted by Q2 to ground, so significant
current can be drawn through Q2 from a load connected between the output and the positive supply. This is another very
important feature of the CMOS digital inverter circuit.
A CMOS inverter can become a linear inverting amplifier by biasing its input terminal VIN
at a value intermediate between the logic-0 and logic-1 levels. In this situation Q1 and Q2 are both partly biased
ON, so the inverter passes significant quiescent current.
Figure 14 shows the typical drain-current transfer characteristics of the linear inverting
amplifier under this intermediate condition. Drain current (ID) is effectively zero
when the input voltage (VIN) is either at zero or full supply volts. However, drain
current rises to its maximum value when the input voltage is approximately half the supply voltage.
Three different supply voltage (VDD) conditions are shown in Fig. 14: 5, 10, and 15
volts. These result in drain currents of 0.5 milliamperes, 4 milliamperes, and 10.5 milliamperes, respectively. Under
these conditions both inverter MOSFET's are biased ON equally.
Figure 15 shows typical input-to-output voltage-transfer characteristic's for a CMOS
inverter at three different power supply voltage VDD values: 15, 10, and 5 volts.
With a 15-volt supply, for example, the output voltage changes only a small amount when the input voltage is shifted
between the VDD and zero-volt levels.
However, when VIN is biased a roughly half the supply voltage, a small change of input
voltage causes a large change of output voltage. In the half-supply condition, the inverter typically provides a
voltage gain of about 30 dB when used with a 15-volt supply, or 40 dB with a 5-volt supply.
Figure 16 shows a CMOS inverting amplifier. This circuit is biased automatically by
connecting a 10-megohm resistor R1 between the input and output terminals. As a result, the output self-biases at
approximately half the supply voltage.
Figure 17 shows typical voltage gain and frequency characteristic curves for a CMOS
inverter stage when it is powered at three different levels: 15, 10, and 5 volts. These curves were obtained when the
amplifier output fed into the high impedance of a 10-megohm, 15-picoFarad oscilloscope probe. Under this condition,
the circuit has a bandwidth of 2.5 MHz when operated from a 15-volt supply.
As might be expected from the voltage transfer curves in Fig. 15, the distortion characteristics of the CMOS linear
amplifier are not very good. The linearity is acceptable with small-amplitude signals whose output amplitudes reach
3 volts peak-to-peak with a 15-volt supply. However, distortion increases progressively as the output approaches the
upper and lower power supply limits. Unlike a bipolar transistor circuit, the CMOS amplifier does not "clip" excessive
sinewave signals; it progressively rounds off their peaks.
Figure 18 shows the typical drain-current vs. supply-voltage characteristics of the CMOS
linear amplifier. The drain current (ID) typically swings from 0.5 milliampere at 5
volts (VDD) to 12.5 milliamperes at 15 volts (VDD).
In many applications the quiescent supply current of the 4007UB CMOS amplifier can be reduced with a penalty of reduced
amplifier bandwidth by placing external resistors in series with the source terminals of the two MOSFET's of the CMOS
stage. This is illustrated as the micropower circuit of Fig. 19.
Table 2 shows the measured results of placing different values of resistor in the source circuits of transistors Q5 and
Q6.
With changing values of both R1 and R2, a constant supply voltage (VDD of 15 volts,
and the output loaded by a 10-megohm, 15-picofarad oscilloscope probe, the results can be read across the table.
They are drain current (ID), voltage gain, and upper 3dB bandwidth.
The additional resistors shown in the circuit of Fig. 19 increase the output impedance of the amplifier. (The output
impedance is roughly equal to the R1-voltage gain product.) This impedance and the external load resistance and
capacitance has a significant effect on the overall gain and bandwidth of the circuit.
When the values of R1 and R2 are 10,000 ohms, it can be seen that if the load capacitance is increased from
15-picofarads to 50 picofards, the bandwith fall to abot 4 kHz. However, if th ecapacitance is reduced to 5
picofarads, the bandwidth is increased to 45 kHz. Similarly, if the resistive load is reduced from 10 megohms to
10,000 ohms, th evoltage gain falls to unity. thus, to obtain significant gain, the load resistance must be large
relative to the output impedance of the amplifier.
The basic (unbiased) CMOS inverter stage has an input capacitance of about 5 picofarads and an input resistance of
near infinity. Thus, if the output of the circuit in fig. 19 is fed directly to such load, it will show a voltage gain
of about 30 and a bandwidth of 3 kHz when R1 and R2 are both 1 megohm. The amplifier will work when R1/R2 value is 10
megohm, but it will consume a quiescent current of only 0.4 microampere.
Practical CMOS Circuits:
A CMOS linear amplifier will function in either its standard or micropower forms ot provided a wide range of
fixed-gain amplifiers, mixers, integrators, active filters, and oscillators.
Figures 20 to 24 are examples of some of the possible circuits derived from the
amplifier.
Figure 20 shows the practical circuitof a X10 inverting amplifier. The CMOS stage is
biased by feedback resistor R2, and the voltage gain is set at X10 by the ratio of resistor R2 to resistor R1. The
input impedance of the circuit is 1 megohm, and that equals the value of resistor R1.
Figure 21 shows how the circuit in fig. 20 can be modified to become an audio mixer or
analog voltage adder. The circuit has four input pins, and the voltage gain between each input and the output pin is
held at unity by the relative values of the 1-megohm feedback resistor. Figure 22 shows
how the basic CMOS amplifier is organized as a simple integrator.
Figure 23 shows how a linear CMOS amplifier can function as a crystal oscillator. The
CMOS amplifier is biased into the linear region by resistor R1, which provides a 180° phase shift. The pi-type
crystal network formed by Rx, C1 and C2, and XTAL1, provides the additional 180° of phase shift at the resonant
frequency to cause the circuit to oscillate.
If you only want the circuit to oscilate at a frequency accurate within about 0.1%, resistor Rx can be replaced with a
shorting wire and both capacitors C1 and C2 can be omitted. For ultra-high accuracy, however, the correct values of
Rx, C1, and C2 must be individually determined.
Figure 24 shows the schematic for a micropower version of the CMOS crystal oscillator.
In this circuit, Rx is included in the amplifier. The output of this oscillator can be fed directly to the input of
another CMOS inverter stage if you want a more precise waveform shape and higher amplitude.
The Metal-Oxide-Semiconductor Field-Effect Transistor, or MOSFET for short, is similar in many ways to the junction
FET (JFET). Both are voltage-driven unipolar devices that depend on either electron or hole movement--but not both, as
does the bipolar transistor. However, there is a fundamental structural difference between the two field-effect
transistors: The JFET has three layers while the MOSFET has two. The MOSFET's simpler construction has give it a
performance edge over the JFET, and made it the world's most popular transistor style.
As shown in Fig. 1a, the channel between the N-type source and drain of an N-channel, enhancement-mode MOSFET
is the substrate P-type material. This MOSFET can be turned on so that current flows between source and drain only
when a positive forward bias is placed on the gate. As a result, the enhancement-mode MOSFET is said to be "normally
off". Its operation depends on the electron flow.
Before proceeding with this discussion of MOSFET's, it is important to point out that protective precautions must be
taken when handling all MOSFET devices. All MOSFET's are susceptible to damage from the discharge of
electrostatic energy between any two pins.
Application Rules:
Linear Operation:
This response makes the circuit a logic inverter or NOT gate. As can be seen in the truth table,
Fig. 13-b, a low (logic 0) input results in a high (logic 1) output; conversely, a
high (logic 1) input results in a low (logic 0) output.
However, when VIN is biased a roughly half the supply voltage, a small change of input
voltage causes a large change of output voltage. In the half-supply condition, the inverter typically provides a
voltage gain of about 30 dB when used with a 15-volt supply, or 40 dB with a 5-volt supply.
Figure 17 shows typical voltage gain and frequency characteristic curves for a CMOS
inverter stage when it is powered at three different levels: 15, 10, and 5 volts. These curves were obtained when the
amplifier output fed into the high impedance of a 10-megohm, 15-picoFarad oscilloscope probe. Under this condition,
the circuit has a bandwidth of 2.5 MHz when operated from a 15-volt supply.
Table 2 shows the measured results of placing different values of resistor in the source circuits of transistors Q5 and
Q6.
Practical CMOS Circuits:
Figure 20 shows the practical circuitof a X10 inverting amplifier. The CMOS stage is
biased by feedback resistor R2, and the voltage gain is set at X10 by the ratio of resistor R2 to resistor R1. The
input impedance of the circuit is 1 megohm, and that equals the value of resistor R1.
Figure 23 shows how a linear CMOS amplifier can function as a crystal oscillator. The
CMOS amplifier is biased into the linear region by resistor R1, which provides a 180° phase shift. The pi-type
crystal network formed by Rx, C1 and C2, and XTAL1, provides the additional 180° of phase shift at the resonant
frequency to cause the circuit to oscillate.
If you only want the circuit to oscilate at a frequency accurate within about 0.1%, resistor Rx can be replaced with a
shorting wire and both capacitors C1 and C2 can be omitted. For ultra-high accuracy, however, the correct values of
Rx, C1, and C2 must be individually determined.
Figure 24 shows the schematic for a micropower version of the CMOS crystal oscillator.
In this circuit, Rx is included in the amplifier. The output of this oscillator can be fed directly to the input of
another CMOS inverter stage if you want a more precise waveform shape and higher amplitude.