Transistor types and circuits
In December 1947, researchers at Bell Laboratories demonstrated a ‘PNP point-contact germanium transistor’, acting as a speech amplifier with a power gain of 18. This event is widely regarded as marking the birth date of the transistor.
Since then, transistors have achieved their ubiquitous presence through their use in both switching and amplification functions, and because they are available in a vast range of power capabilities, switching speeds and many other parameters. While this gives great choice to electronics designers, it can also create a barrier: How do I choose the best transistor for my new project or upgrade?
This article endeavours to provide some guidance on this, by classifying transistors and their parameters. It shows how to use parameters in assessing the suitability of transistors for different applications. The discussion also covers one specific issue that frequently arises when engineers are seeking to connect process sensors to programmable logic controllers (PLC) inputs; whether to use NPN or PNP devices.
While the article is mostly based on bipolar junction transistors (BJTs) and field effect transistors (FETs), more specialist technologies – insulated gate bipolar transistors (IGBTs) and Gallium Nitride (GaN) high-electron-mobility transistors (HEMTs) – are also introduced. An example FET common source amplifier is also described.
We then look at how transistors can efficiently be matched to applications by using parametric searches such as those available on the Farnell website.
Transistors are solid state devices built with semiconductor materials, typically silicon, germanium and gallium-arsenide. They usually have three terminals –one terminal common to the input and output signals, and a signal on one of the remaining terminals controls the current in the other, as shown in Fig.1.
The tree diagram in Fig.2 summarises the different ways in which this basic three-terminal semiconductor concept can be implemented.
This diagram shows that transistors fundamentally split into BJTs and FET types. However, the most basic question of all when choosing a transistor is not whether it's a BJT or an FET but its polarity – in use is its output terminal positive or negative with respect to its common terminal? If the answer is positive we need an NPN BJT or an N-channel FET, otherwise we need a PNP or a P-channel.
An article by James Bryant, published on Analog Devices’ Wiki on May 19, 2014, provides much valuable information and guidance on choosing transistors for applications; some of this is reproduced below.
Most general-purpose transistor applications need devices which are non-conducting with zero bias on the control input (base or gate). Such devices are BJTs or enhancement mode MOSFETs. Depletion-mode FETs are much less common, although valuable for some applications. The choice of NPN/N-channel or PNP/P-channel is dictated by whether the supply is positive or negative, but do we need a BJT or a MOSFET?
In many cases it does not matter. Discrete MOSFETs are perhaps ten or twenty percent more expensive than BJTs, but they do not need base resistors which cost money and expensive board area. They are slightly more vulnerable to electrostatic damage (ESD) during handling, but they do not draw base current and load circuits at DC (since they have relatively large input capacitance they may give rise to capacitive loading issues in higher frequency circuits).
At one time the gate threshold voltage (the value of Vgs at which a MOSFET starts to conduct) was several volts, so they could not be used with very low supply voltages, but today the threshold voltages of many devices are comparable to the 0.7V base turn-on voltage of a silicon BJT. So, BJTs and MOSFETs will now function equally well in both amplifier and switching applications.
However, the input of a BJT is a silicon diode. We can use its thermal properties to measure temperature, and its high current when over-driven to act as a clamp or limiting circuit, so there are some circuits where a BJT is essential.
For some twenty years the magazine Elektor has published circuits designed around transistors which it calls TUNs and TUPs (“Transistor Universal NPN” and “Transistor Universal PNP”). These transistors are silicon planar BJTs and any transistor which exceeds the following specification qualifies:
Most cheap small-signal silicon transistors do qualify. The list could also include MUNs and MUPs (“MOSFET universal N-channel” and “MOSFET universal P-channel”) – and most cheap small MOSFETs qualify for this specification:
There are a number of parameters – in addition to obvious factors like power rating – to be considered when assessing a transistor as a candidate for your circuit. We consider these next.
Maximum collector/drain voltage. BVceo or BVds: If the maximum supply voltage is less than BVceo or BVds and there is no inductive circuitry in the collector/drain which might produce higher voltage transients, and there is no external signal source which might apply higher voltages, then we need not worry about this specification.
However, there are many circuits where a transistor may be expected to work with high values of Vce or Vds, either steady state or as transients, and it is essential that where this is the case the correct maximum is chosen.
BJTs and MOSFETs with breakdown voltages of more than 500V are inexpensive and readily available, although the current gain, ß, of high voltage BJTs is more often in the range 40-100 rather than the =100 of the TUN/TUP. Similarly, the gate threshold voltage of a high voltage MOSFET is more likely to be in the range 2-5V rather than 500-2000mV of the MUN/MUP.
Absolute maximum collector/drain current. Ic(max) or Id(max): The maximum current expected in the collector/drain must not exceed the absolute maximum current rating of the device. Given that the TUN/etc value for this is 100 mA this is unlikely for small-signal circuits, but if the transistor is required to provide power to a load the maximum current must be checked.
The absolute maximum current rating of some devices may be divided into a DC (or perhaps mean) current rating and a higher transient rating for short pulses. It is important to ensure that peak transient currents are within their rated limits.
Most small-signal transistors have Imax ratings greater than 100 mA – usually 300-1000 mA – and many devices which meet the TUN/etc specification will actually have such a rating and may be used when such medium currents are needed. If higher currents are required TUN/etc devices will be inadequate and a power device must be chosen. At higher currents it is important to comply with power ratings as well as current ratings, packages will probably be larger, and a heat sink may be necessary. BJTs with higher maximum currents may have lower values of ß at high currents.
Packages & Power: There are innumerable different transistor packages from near microscopic surface mount ones to large plastic and metal packages capable of handling several kW with adequate cooling. Choose the one which is most convenient for your application – surface mount for mass production, leaded for prototyping and small-scale production where ease of hand soldering is helpful, and whatever power package is appropriate when dissipation and heat sinks need to be considered.
Collector/drain leakage current Ice0 or Idss0: (Sometimes called the “cutoff current”.) This is the small leakage current which flows from collector to emitter or drain to source when the transistor is turned off. It is usually in the order of tens of nA but data sheets sometimes set rather larger worst case maximum values to reduce testing costs. Transistors used as very low-level switches or amplifiers should be chosen for leakage below 50 nA but for most applications 200 nA or even more is quite satisfactory.
The low power inverter shown in Figure 3 is an example of a circuit requiring very low collector/drain leakage. Drain leakage of 100 nA gives a voltage drop of 1V and an output voltage of 2V, only just on the threshold of permitted logic 1 levels, so practical designs should use an MOSFET having drain/source leakage = 50 nA.
Note that although this inverter is very low power [300 nA = 0.9 µW when the transistor is on] it is also very slow – assuming a transistor output capacitance plus track capacitance plus next stage input capacitance of 20 pF, which is not unreasonable, it has a rise time of some 0.2 msec – this is acceptable for DC applications, but not for even medium speed switching circuits.
Current gain ß or hfe: The current gain of a BJT is the ratio of the collector current to the base current when the device is not in saturation, i.e. the collector/base voltage is positive [for an NPN device]. ß is usually fairly constant over a wide range of currents, but it may be slightly lower at very low base currents and will almost certainly start to fall as the collector current approaches its absolute maximum value. Since it is a ratio it is a dimensionless value.
TUNs and TUPs have ß = 100, but high current and high voltage BJTs may have slightly lower (=40 or 50) minimum specified values.
An emitter follower/source follower output stage, illustrated in Figure 4, is equally accurate with a BJT or an MOSFET. In simple emitter followers it is assumed that the base/emitter or gate/source voltages Vbe or Vgs remain constant, giving a fixed offset between the input and the load voltage, but in more accurate circuits feedback may be taken from the emitter (source)/load connection.
Since some of the emitter current must flow in the base, the collector and emitter currents of a BJT are not identical, which means that the current output stage should be made with a MOSFET rather than a BJT since MOSFETs have virtually zero gate current.
Forward transconductance gfs: The forward transconductance of an FET is the ratio of ΔIds/ΔVgs when the device is turned on and the drain circuit is not current-limited. It is measured in siemens (S). Small-signal FETs and MOSFETs may have gfs as low as a few mS but larger ones can have gains of large fractions of a siemens to several siemens or more.
In general, a few volts change of gate voltage is sufficient to change the drain current from minimum (off) to its absolute maximum value. It is also important to know at what gate voltage conduction starts (see below).
Gate threshold voltage Vgs(th): The gate threshold voltage of a MOSFET is the gate/source voltage at which the correctly biased drain starts to draw current. The definition of “starts” will be specified on the data sheet and may be as low as a few µA, but is more likely to be defined as 1 mA, or even more with a high power MOSFET. Above this threshold drain current will rise very quickly with small increases of gate voltage.
If a MOSFET is to be driven by logic it is important that its threshold voltage be above the worst-case value of logic 0 over the temperature range of the circuit, which is likely to be at least several hundreds of mV, as otherwise it may start to turn on when it is supposed to be turned off.
Saturation Voltage Vce(sat): When a BJT is turned on hard enough to make the voltage drop in its collector load sufficient to bring the collector potential below the base potential (in other words the base-collector junction is forward biased) it is said to be saturated. This saturation voltage is not proportional to the collector current, so the model of a saturated transistor is not just a resistance between its collector and emitter.
Two examples of the importance of a low saturation voltage are:
[A] In classic TTL logic each input sources 1.6 mA into a logic 0 output driving it. With a full fan-out of 10 this means that a TTL output transistor may be called upon to sink some 16 mA with a saturation voltage of no more than 400 mV.
[B] When a power BJT is used to switch high current loads its dissipation, for a given load current, is proportional to its saturation voltage. The lower the saturation voltage, the less heat must be removed from the transistor.
Note that when you remove the input drive from a saturated transistor there is a delay (usually nsecs or tens of nsecs, but it can be more) before it starts to turn off. This is its saturation recovery time and may be specified, under well-defined conditions, on its data sheet.
On Resistance Ron: MOSFETs do not saturate because they are majority carrier devices. When they are turned hard on with a gate voltage well above their gate threshold voltage they behave as low value resistors and their on resistance is specified on their data sheet. Ohm's law applies – the voltage drop is proportional to the current and the on resistance, and their dissipation is I2R.
Noise Figure NF: The majority of transistor applications are relatively high-level, and noise is not an issue. Where it is an issue, though, it is critically important. Many transistors, both BJTs and FETs, have their noise figure specified and guaranteed by their manufacturers. When comparing the noise figures of different devices, it is essential that they have been measured with the same source impedance. If the transistors are intended for use in radio systems it is likely that their NF will have been measured at 50Ω and so comparison is simple, but it is meaningless to compare the NFs of two devices whose NFs were measured at different impedances.
Transition Frequency ft: The ft of a BJT is the frequency at which the current gain, with a short circuit (at HF) output, is unity. ft is the most widely used figure of merit for comparing the frequency response of BJTs. Most TUNs and TUPs will have ft well over the 100 MHz minimum but high power and high voltage transistors will often have rather lower values.
FETs are transconductance devices with infinitesimal DC input current, so it is incorrect to consider their DC current gain. But since they have input capacitance (Cgs) of pF to hundreds of pF their capacitive input impedance is relatively low at HF and so their HF input current may be measured and their ft derived. Occasionally an FET or MOSFET data sheet will contain a value of ft derived in this way and it is certainly valid to use it, if available, to evaluate FET frequency response, but usually the speed of FETs is specified in terms of switching times.
Switching Times t(on) & t(off): Most FETs, and many BJTs, have switching time specifications, defined as the time taken, under specified conditions, for the output current to rise from zero to a specified value, or to return to zero, respectively. The switching signal is either assumed to be instantaneous (a legal fiction) or defined as a few nsec. Comparing switching times is a reliable way of comparing the relative speeds of transistors, provided they are tested under similar conditions.
Capacitances C??: There are three capacitances associated with a transistor – the input capacitance Cin, the output capacitance Cout and the Miller (or feedback) capacitance Cfb. Different manufacturers use different names (therefore the C?? in the heading) but which is which should be perfectly clear from Figure 5.
FETs, especially power MOSFETs, may have values of Cin as large as 1 nF or even more, although small-signal MOSFETs will have much smaller values, probably in the range of 15-50 pF. It is important, though, when designing circuits where such capacitance may affect rise times or circuit stability to ensure that the design takes account of such values and that devices are chosen to have capacitances which the circuit design can tolerate.
NPN vs PNP devices
When deploying programmable logic controller (PLC) systems, their digital inputs must be matched to their connected sensors, not only in terms of voltage levels, which are usually 24 VDC, but also for polarity – NPN or PNP. PNP sensors are current-sourcing devices, while NPN types are current sinks. A current-sourcing sensor must be connected to a current-sinking input and vice versa.
In choosing, there are a couple of arguments to favour PNP. Firstly, PNP sensors are easier to understand and troubleshoot by technicians, since the sensor will give a high-level voltage signal when the output is active. Secondly, in an NPN circuit, if a wire breaks and contacts ground, the PLC input is true. This can potentially result in undesirable machine behaviour (for example, start push button input turning on). When a wire in a PNP circuit shorts to ground, the PLC input is false.
Nevertheless, while PNP sensors are standard in the United States and Europe, the NPN variety still enjoy some preference in Asia. This means that in the worst case, an organisation could find itself having to carry extra stock of sensors or input modules to ensure compatible pairs could always be set up. This aspect of the installation would also need to be tracked and managed.
Mostly, though, the situation is eased with a degree of flexibility. For example, many input modules (especially the IP20 versions that are most commonly installed in cabinets) can be wired either NPN or PNP. However, note that all of the inputs on that module need to be either NPN or PNP. You can’t mix and match them. Additionally, new sensors are coming into the market that can be wired or configured as either NPN or PNP. Alternatively, many PLC cards can be wired to accommodate either NPN or PNP and do not require hardware changes.
These points are more extensively discussed in a ‘Control Design’ magazine Q and A document titled ‘How to decide between PNP and NPN’.
Other transistor technologies
So far, we have discussed the basic forms of BJT and FET transistors and their variants. However, other types also exist; two key examples are the Insulated Gate Bipolar Transistor or IGBT, and the Gallium Nitride (GaN) high-electron-mobility transistors (HEMTs).
The Insulated Gate Bipolar Transistor or IGBT is a cross between a BJT and an FET transistor. It combines the high input impedance and high switching speeds of a MOSFET with the low saturation voltage of a bipolar transistor to produce another type of transistor switching device that is capable of handling large collector-emitter currents with virtually zero gate current drive.
IGBTs offer the output switching and conduction characteristics of a bipolar transistor but are voltage-controlled like a MOSFET. IGBTs are mainly used in power electronics applications, such as inverters, converters and power supplies, where the demands of the solid state switching device are not fully met by power BJTs and power MOSFETs. High-current and high-voltage BJTs are available, but their switching speeds are slow, while power MOSFETs may have higher switching speeds, but high-voltage and high-current devices are expensive and hard to achieve.
Figure 6 shows that the insulated gate bipolar transistor is a three terminal, transconductance device that combines an insulated gate N-channel MOSFET input with a PNP bipolar transistor output connected in a type of Darlington configuration.
As a result, the terminals are labelled as Collector, Emitter and Gate. Two of its terminals (C-E) are associated with the conductance path which passes current, while its third terminal (G) controls the device.
Gallium Nitride (GaN) high-electron-mobility transistors (HEMTs) comprise another transistor technology. They are opening new possibilities by offering several advantages over silicon MOSFETs. GaN transistors can achieve a much higher dV/dt slew rate, and thus can switch much faster than silicon MOSFETs, significantly reducing switching losses. Another advantage is the lack of a reverse-recovery charge, which causes switch-node ringing with traditional silicon MOSFET designs.
The transistors are also regarded as combining a high breakdown voltage with high efficiency to perform well as power devices. After use in mobile base station power amplifiers and radar sensor applications, further expansion into power conversion, in equipment such as server power systems.
With low on-state resistance and high-speed switching performance GaN HEMT offers potential for device miniaturization, reduced power consumption and lower costs through further developments in the technology, together with circuits that maximise these advantages.
Transistor application example – FET common source amplifier circuit
The common source FET amplifier circuit is one of the most widely used of all FET circuit configurations for many applications, with a high level of all-round performance. It provides current and voltage gain along with satisfactory input and output impedance.
Figure 7 shows a typical common source amplifier circuit, as developed by electronicsnotes. The input signal enters via C1; this capacitor ensures that the gate is not affected by any DC voltage coming from the previous stages. The resistor R1 holds the gate at ground potential. Its value could typically be around 1 MΩ. The resistor R2 develops a voltage across it holding the source above the ground potential. C2 acts as a bypass capacitor to provide additional gain at AC.
The resistor R3 develops the output voltage across it, and C3 couples the AC to the next stage whilst blocking the DC.
Sourcing the right transistor
Having understood and defined a set of parameters for your target transistor, the next step is to find a real, available device that possesses this set. One way of doing this is to use Farnell’s parametric search engines to filter out suitable candidates. The ‘Bipolar Transistors’ section, for example, can be searched on Polarity, Collector Emitter Voltage, Transition Frequency, Power Dissipation and DC Collector Current as well as Compliance, Packaging and Manufacturer.
Similarly, the ‘RF FET Transistor’ area can be filtered by Drain Source Voltage, Continuous Drain Current, Power Dissipation, Operating Frequency Min and Max as well as Case, No. of Pins, Max Operating Temperature, Compliance, Packaging and Manufacturer.
(iv) Named after John Milton Miller, who first described its effects in 1920.
Transistor types and circuits - Date published: 4th September 2018 by Farnell element14