The Classroom

Voltage Regulators

Daniel Koch

Issue 54, January 2022

A canvass of the different types of voltage regulators makers encounter, and how and when to use them.

Voltage Regulators are an item the maker comes across quite frequently. With makers coming to the field from so many backgrounds and levels of knowledge, it is possible someone can build a very complex circuit adapted from other smaller designs, and never know how the parts work. Many people use voltage regulators as they see them in online circuits and designs, not knowing how they are properly used beyond what is in the circuit they found online. Others have a sound idea of how they work and how to use them effectively, but still aren’t sure of some of the lesser-known features or traps. In fact, we even discovered some things while researching for this article!

In addition, there is an ever-increasing array of products with which you can regulate voltage within your projects. Very efficient modules can now be bought to step up voltage from batteries to the 5V needed for Arduino, or higher.

There are cheap regulator modules and expensive ones, and the difference can sometimes be mystifying. We’ll cover the difference between linear regulators and switch mode power supplies, look at some basic regulators made from discrete components, then look at some of the common linear regulators. Finally, we’ll summarise some lesser-known designs of regulators and the concept of hybrid power supplies.


While it may seem simple, the purpose of voltage regulators in our experience is not always understood. Many people think of them as a way of reducing voltage, and this is indeed how they are often used. However, their absolute purpose is to provide stable, consistent voltage. This often occurs in conjunction with a voltage reduction, either as a means to an end or a desirable outcome, hence the misconception.

This is why we don’t consider the basic resistor to be a voltage regulator, even though it could be argued that it is. A resistor depends on the relationship between its value, the current in the load, and the voltage passing through it in order to provide regulation. In something like an LED, where the forward voltage drop of the LED is somewhat constant, and if the power supply is already constant, then strictly the resistor regulates current, not voltage.

If the supply voltage fluctuates or current draw increases, then the voltage dropped across the resistor changes. This is not a property of a voltage regulator. A regulator should maintain a constant output, within its design considerations, regardless of variation in load and supply voltage. Of course, that’s not an open-ended statement. Many designs have a minimum overhead, an amount by which the input voltage must be higher than the output voltage to maintain regulation.

Variation above this point should not show on the output, but variations under that value may show through. That is for basic regulators. Some more modern switch mode devices are capable of increasing or decreasing the voltage they are fed so that the output is constant whether the input is over or under the target value. There are also many designs of regulator which are designed to boost a voltage to a constant level from a varying lower voltage, but the principle remains the same.


The two most common forms of regulator that the maker encounters are linear integrated circuits (IC) and switch mode regulators. Linear regulators are a circuit which uses some form of reference voltage to control the load voltage and either dump excess to ground via a resistor as heat, or in some other way transform the voltage. The effect is of a variable resistor, regardless of whether the IC uses an actual resistor for this or some other set of circuitry such as semiconductor junctions to provide resistance or shunt current. Because of this, the input voltage for a linear regulator must always be higher than the output voltage. Another common factor of linear regulators is that they turn any excess power into heat, which must be dissipated.

By contrast, switch mode regulators are active devices which use a high-speed switch to turn the current on and off, which is then smoothed by a capacitor, inductor, or network of both, to produce a stable average voltage. Because there are inductors involved, switch mode power supplies can be constructed to have an output voltage higher than their input voltage.

Some designs are fixed, having either a step up configuration for creating an output voltage higher than the input (known as a boost converter), or a step down configuration to give a lower output than input voltage. Other designs are more advanced and can create a stable output voltage from an input voltage above or below the output, even if it varies. It is important to note that this is still within certain parameters, and the switch mode power supply (SMPS) will not function as a rectifier on its own. Current must still be flowing in one direction.


The most basic demonstration of a voltage regulator comes from stringing several silicon diodes in series, with a current-limiting resistor in front. Silicon diodes have a voltage drop that does not vary with current nearly as significantly as many other components. The response curve is not flat, but it is much flatter than most. By stacking diodes which have a reasonably known forward voltage drop, a very basic regulator can be created. The series resistor is necessary to prevent the diodes from exploding, as they will not current-limit themselves to their one amp limit. The voltage drops of the individual diodes add together to give an overall value which, when connected to the resistor, forms one half of a voltage divider.

However, unlike a purely resistive voltage divider, the semiconductor half maintains a fairly constant voltage drop if the input voltage changes. Adding a load means that some of the current from the resistor/diode junction now goes through the load and not the diodes, altering their voltage drop slightly. The system is also not as stable as more advanced regulators. As the input voltage increases, the current through the resistor increases too, as the voltage dropped across it increases. The relationship between the voltage drops changes, while in a purely resistive voltage divider, it does not.

In our experiment, we connected four 1N4004 diodes in series with a 100Ω resistor rated at 1/2W. The forward voltage drop of the 1N4004 is typically 0.6V to 0.7V. Using a semiconductor tester, the values were measured at 0.667V, 0.676V, 0.658V, and 0.665V, all at 5mA test current. Strung together, that’s 2.666V.

At 5V, this means the resistor drops 5 - 2.666 = 2.334V. Ohm’s Law I = V/R (current I is the voltage drop V divided by the resistance R) tells us that there will be 23.3mA flowing through the resistor. However, the diodes will add a certain amount to this. At a worst case with no resistance from the diodes, 23.3mA gives a power dissipation (P = V x I) of 2.334V x 0.0233A = 0.054 Watts.

We hooked up the circuit to 5V and measured. Actual current was measured in series as 21.5mA. The diodes are providing some opposition to current flow, but not much. Measured voltage drop across the resistor was 2.233V, and the drop across the diode string was 2.917. Measured supply voltage was 5.14, so this is fairly close after decimal rounding errors in the multimeter are factored in.

However, challenges ensued when running on 12V. With the same diode voltage drop of 2.666V, we were left with 9.334V dropped across the diode, with a current value of 93.3mA. At that current and voltage, the power dissipated in the resistor is 0.87W That’s quite a bit more than the 0.5W the resistor is rated for. Without a higher-rated resistor and not wanting to use two 47Ω in series because 100Ω (measured at 99.8Ω, too!) works so well for the calculations, we overcame this by turning power on only in short bursts.

Luckily, we have our PCBite holder and magnetic probes on hand, so we could set up for the measurements and then turn on the power. Measured current was 88.9mA, voltage drop across the diodes was 3.148V, and across the resistor 9.1V. That agrees well with the measured supply voltage of 12.26V after the resistance of the spring-hook connectors is considered.

It is important to understand that as soon as a load is added, things change. Accordingly, we reconnected the 5V version and connected a green LED. This one has a forward voltage of 3V and a current draw of 30mA. That means this ‘regulator’ should run it without any smoke coming out of the LED (or any other component). We now measure 23.7mA current through the circuit, 2.401V drop across the resistor, and 2.723V drop across the diodes with the LED in parallel with the overall diode string.

You can see from the results once we measured the current through each branch of the circuit at each voltage that while it does work, something far better is required.


We have covered Zener diodes before as a Classroom topic, in issue 46. Be sure to check that one out for full details. It can be read for free online. Regardless, a summary is still in order. Zener diodes are a particular construction of PN junction, which behave like a regular power or signal diode at first. They can pass current in the forward direction, and block it in the reverse direction. However, Zener diodes are used in the reverse direction, and at a given designed voltage, they break down and conduct. A normal diode will do that too, but the breakdown voltage is much higher and the diode is destroyed when this is exceeded. A Zener diode is not destroyed, but simply passes this excess voltage to ground. The voltage measured across it will be its rated Zener voltage, regardless of what is being supplied to it.

This makes the Zener diode a basic regulator component. However, it is not magic. It has a power dissipation limit, and it isn’t terribly high. The term ‘power dissipation’ means the amount of power that an electronic component can turn to heat. In the case of the Zener diode, that’s a side-effect of dumping the excess to ground over the special PN junction. On the retail market, they are available in 1W and 5W versions. This means 1W or 5W of current through them, regardless of how much is actually turned to heat.

As with other diodes, Zeners will not current-limit themselves. A resistor must be used, and the current for both the Zener and the load flows through this. That on its own screams out ‘inefficiency’, unless you make the resistor as small as possible. Even then, with excess voltage dumped to ground and quite a loss no matter what on the resistor, these designs are not winning any energy star ratings any time soon.

Note that the load is shown as a resistor, but it may be anything else. Inductive and capacitive loads are not especially compatible with the limited current available from a Zener/resistor regulator, so they are generally used with loads that are relatively consistent and linear like a resistor is. LEDs are a good example.


The regulators above are known as ‘shunt’ regulators. This term applies to any device where current is shunted to ground, diverted away from the load. This may be the excess voltage as in the Zener circuit, or a portion of all of the current, as in the diode circuit. Circuits like this are generally low-current, and low efficiency. They can also only sink current, which means they absorb it. Regulation is provided by the effect of this on the load voltage.

Series regulators, on the other hand, are placed between the supply voltage and the load. They use some form of varying resistance which carries the whole current that passes to the load, and turn excess voltage to heat. They are generally more efficient than shunt regulators, but also generate substantial heat.

This is often because they can cope with much more current than anything else. A Zener diode run at its maximum will get pretty hot, too! A small amount of current is shunted to ground across a reference, in order to regulate the resistance against a constant value, but the majority passes across the series resistance.


The real value in Zener diodes for most makers is as a voltage reference, and in that role, they feature in the construction of linear IC regulators. These bring together the components needed to produce a reliable, smooth voltage regulator with reasonable current capability, into a single package that can easily be used and connected.

They eliminate many of the calculations that would be required to design a regulator circuit from scratch, and end up with generally three connections: Input voltage, Ground, and Output Voltage. They are more stable than the basic regulator circuits above because the incorporate feedback from the output and compare it to the reference, adjusting the resistance to maintain a stable output.

In reality, the function block diagram is a drastic simplification, but all of the components in this image of the internal schematic fall into one of the areas in the function block diagram. This internal schematic is from the 7805 regulator, a three-terminal +5V regulator that many makers have come across at some point.

We reproduced this from Texas Instruments’ datasheet, but other manufacturers would show it as well. Of note, D1 and D2 are both Zener diodes used within the device to produce reference voltages. Q16 is the main transistor which has its base voltage varied so that the resistance between its emitter and collector changes, and reduces the output voltage in response to feedback.

Linear regulators are available in many forms, but we’ll concentrate on the ones available on the retail market, the ones most makers have access to. The majority of these are three-pin devices in a TO220 package, with some available in TO92 packages. Standard versions have an overhead of around 2V to 2.5V at a minimum, below which regulation is not guaranteed or the output voltage drops. They come in a range of voltages, but typical over-the-counter options are +5V, +6V, +9V, +12V, and +15V.

Some retailers have more voltages. The fixed voltage regulators are so easy to use that you could be forgiven for thinking you’ve missed something. Connect the supply voltage to VIN, ground the GND pin, and the VOUT is the regulated output. Ground is common between supply, device, and load.


In the connection diagram, however, we have shown two capacitors. One is recommended on the input, at a value of 0.22µF to 10µF, if the regulator is placed more than 150mm from the point where the input supply is filtered after rectification. Because in most maker cases, that would be an external location, it is better to use this capacitor. Tantalum types are favoured, but electrolytic capacitors work just fine for most roles. Some datasheets don’t specify at all. For the output capacitor, a 100nF ceramic type is specified, but the datasheet notes that it is not expressly necessary.

It does not help stabilise the output under normal conditions, but does improve the response to transients, which are small fluctuations which the capacitor helps to fill in. Transients are caused by other parts of the circuit, not the regulator. While both of these capacitors are optional, we feel that there are very few situations in which they shouldn’t be included. Both the cost and space needed are minimal, and they add extra certainty to the design.

The main consideration for practical use is heatsinking. We have a whole section on that shortly, but for now, the more current that the regulator will have to pass, the more it will heat up. On most common regulators, the TO220 package allows easy and direct connection to a heatsink. Most devices are rated at 1A, while the ‘L’ in some part numbers (78L05, for example) denotes a low-power version. These usually have a current limit around 100mA, and are in a TO92 package. These can still have a heatsink but it is not so easy to attach and the package is not designed for it. Additionally, finding a heatsink to suit is not easy at a retail level.

7805+5V 1ATO220ZV1505Z0505
7806+6V 1ATO221ZV1506Z0506
7809+9V, 1ATO222ZV1509Z0509
7812+12V, 1ATO223ZV1512Z0510
7815+15V, 1ATO224ZV1515Z0512
7824+24V, 1ATO225ZV1524Z0514
7905-5V, 1ATO226ZV1525Z0525
7909-9V, 1ATO227ZV1530-
7912-12V, 1ATO228ZV1532Z0528
7915-15V, 1ATO229ZV1535Z0530
78L05+5V, 100mATO92ZV1539Z0460
78L12+12V, 100mATO92ZV1542Z0462
79L12+12v 100mA TO92-Z0468
LM317+1.2V to 37V 1.5A adjustableTO220ZV1615Z0545
LM337-12.2V to 35V 1.5A AdjustableTO220ZV1620Z0562
LM3940+3.3V 1A LDOTO220ZV1565Z2680
LM2940CT-5.0+5V 1A LDOTO220ZV1560Z0592
LM2940CT-12+12V 1A LDOTO220ZV1562Z0595
LP2950-33+3.3V 100mA LDOTO92-Z1025
LP2950-5+5V 100mA LDOTO92-Z1027


Input voltage must be maintained above the value stated in the datasheets, otherwise the output will vary. The phrase ‘overhead’ is sometimes used, but most of the datasheets use ‘dropout voltage’. It is the amount above the output voltage that the input voltage must stay. Many linear regulator designs will reduce their output when their overhead voltage is too low. However, we have encountered some, usually cheap generic ones where manufacturers have cut corners, where regulation fails altogether. This can be very problematic.

Consider the situation where 9V worth of AA batteries are being used to power a 5V sensor. The LM7805 from Texas instruments (TI) is a 5V regulator with a dropout voltage of 2V, meaning an input of 7V is the minimum. As soon as the battery voltage falls below this, TI’s device will reduce its output. However, we have found cheap ones where suddenly, there is no regulation and the full input voltage is passed to the output. For a sensitive 5V circuit, 7V or so may well cause failure. Buying brand-name devices will help protect you against this.


Low dropout regulators are another linear device, in many ways the same as the standard type above, with a reduced variety of voltages. The big advantage to these is that regulation continues much closer to the output voltage. For example, the LM2940CT-5.0 is a 5V device with a dropout voltage of 500mV at 1A. That means a theoretical minimum input voltage of just 5.5V, 1.5V lower than the LM7805. The trade-off is higher cost, over four and a half times the cost on most listings. However, if you are working with batteries, this can make a big difference to the usable lifespan of your design.

Like their standard cousins, low dropout regulators are available with different maximum currents. Because their main value is in low-powered applications, many on the retail market have current ratings in the milliamp range. There are some exceptions, and one item we looked at has a current rating of 3A for a 12V device with a 1.3V dropout, while the standard 3A version, the MC78T12 has a dropout of 2.2V to 2.5V. That may seem trivial, but if you are running 12V regulated devices or circuits in a car, for example, where battery voltage without the engine running sits around 13.8V, that extra headroom is very important.


Convenience and size are not the only reasons to use a regulator IC rather than building a circuit. They come with a range of built-in protection features to help avoid damage to themselves. Texas Instruments’ datasheet for the LM7805 even declares them to be “essentially indestructible”. If the output is short-circuited, a current limiting feature stops excessive current from passing through the regulator. This will save the regulator from being destroyed, but does not initiate shutdown on its own, so the load may still be damaged further than the fault which caused the short circuit in the first place.

IC regulators also have thermal shutdown. When the junction of the series transistor inside (the part that acts like a variable resistor) gets too hot, another circuit shuts down the whole regulator. At this point, no current passes, so the load is unpowered. In addition to these features, many regulators incorporate something called ‘safe area protection’ or ‘safe area compensation’.

This means that the internal circuitry adjusts output to maintain safe operating conditions for the device. The maximum current available is automatically reduced as the device heats, to stop thermal runaway. Some devices also have a reduced current capability at very high overhead voltages.

If in doubt, always consult the datasheets. These can be daunting for those unused to them, but the really important stuff for a maker usually has headings, and the rest if contained in graphs and tables. This is where it is easier for people to miss things, so look carefully at the titles of each graph.

Some to look for include ‘Maximum Average Power Dissipation’, ‘Output Voltage Vs Temperature’, and ‘Dropout Voltage Vs Output Current’. Also, scan the tables. Look closely at the ‘Absolute Maximum Ratings’ section, because these are the critical numbers. The ‘Electrical Characteristics’ table contains useful information, too. Again, these look daunting at first, but look line by line, and you can usually work things out.


Despite this protection, there are still things you can do to destroy a linear regulator IC. One of these situations occurs when large capacitors are used across the output. This is not needed for stability or smoothing but is used in circuits where the load current will fluctuate or momentarily exceed the regulator capability. During normal operation, all is well. However, if there is a situation where the input is short-circuited, the input is at a much lower potential than the output, and current can flow through an internal diode and some transistors from the large store in the capacitors. This can cause permanent damage, although only when the input is shorted.

If the input is simply disconnected, there is still a higher potential on the capacitor side of the regulator, but with open-circuit inputs, it has no path. If there is a situation where input shorting can occur, such as in some switching setups, then an external diode should be added to provide a current path.

Another big “DON’T!’ is having an output voltage higher than the input voltage. This cannot occur natively from the regulator alone, and is the result of another part of the circuit when it happens. This could be voltage spikes from inductive loads or contamination from additional step-up circuits further down the line, or any other situation where a higher voltage could exist. In this case, the possibility and mechanism of damage is the same as for a short-circuit input: The input is at a lower potential than the output, so current flows the wrong way through components that cannot handle it.

While in most maker circuits, deliberately generated higher voltages are less common (such as a voltage doubler to more efficiently run LEDs in series), inductive spikes are more likely. Remember flyback diodes and filter capacitors when using motors and relays. Importantly, the situation can exist if battery power is used and the input falls below the output. Of course, this will only apply if the load has enough capacitance to store enough charge to cause a damaging current. In circuits without large capacitors, when the battery voltage falls below the dropout voltage, the output falls, and so generally avoids the problem. The battery voltage tends to fall slowly, also helping to avoid the problem by allowing capacitors to discharge slowly without a big potential difference.

A poor ground connection can cause trouble for the regulator, too. However, if the path between the regulator and ground is poor or becomes disconnected, it is the connected circuit that is most likely to incur damage. When ungrounded, the regulator does not regulate, and instead passes nearly the full input voltage to the output. When designing a circuit layout, ensure a clean ground path and minimise the number of connections between the regulator ground pin and the ground connection itself.

Transient voltage is a term used to describe fluctuations in voltage above and below the rated supply voltage. For example, on a 12V DC supply, a sudden dip to 6V is not a transient, but a dip below ground (0V) to -0.5V is a transient. Likewise, a spike of 16V is a transient. Positive transients can cause damage if they exceed the input rating. If you have a 7809 regulator on a 12V DC supply (The LM used earlier for the 7805 from TI is a manufacturer code. The family is the 78XX series), it has an input voltage maximum of 35V. A transient of 15V will not cause a problem. A transient of 50V will.

However, negative transients are likely to be the bigger problem for the maker. A negative transient of just 0.8V can damage the regulator. Negative transients are caused by any situation which reverses the current flow, and can be a side-effect of poorly-protected inductive loads as well as positive spikes.


Linear regulators work by having a series pass transistor (just a transistor in series between input and output with the current passing through it) which is rarely in saturation. It is kept only partly conducting, thereby enabling the regulator to work. This generates heat, and quite a lot of it. Low-current regulators like the 78L05 100mA 5V unit are in a TO92 package, and their rated current can be drawn without heatsinking, because the package is not designed for it (however, you can heatsink a TO92 and it helps). On packages such as the TO220 which are designed for a heatsink, the maximum rated current is declared with this in mind. The maximum current figure cannot be reached in free air with the metal tab of the package as the only heatsink.

Bob Harper wrote a great article on heatsinks way back in Issue 24. He details all of the maths required to work out the heatsink required for a given situation. The information you will need to calculate a suitable heatsink is contained in the tables near the beginning of the datasheet. There is a whole table on ‘Thermal Characteristics’, but some information can be found in the other tables too, which will be needed. Thankfully, however, you don’t need to do the calculations yourself. We still encourage you to read ‘Cool Components’ from Issue 24, which you can do for free online.

However, there are two other ways to figure out a heatsink. One is in the graphs in the datasheets. There are several tables dealing with temperature, and often, information can be found showing power dissipation for a given heatsink rating in °C/W. Power dissipation is the amount of voltage dropped, multiplied by the throughput current. For example, a 5V regulator with a supply of 8V passing 1A is dropping 3V, therefore dissipating 3V x 1A = 3W.

Helpfully, several online calculators can be found that allow you to add data gleaned from the tables and graphs in the datasheet, and calculate the heatsink you need for your specific situation. Even if you don’t understand the parameters required, that won’t stop you. Thankfully, they are consistently named and often use symbols which are also consistent, appearing in both the calculators and the datasheets. Just look for matches even if you don’t know what the item is. We have provided a link to one in the ‘Reading and Resources’ section that we use.


So far, we’ve dealt with positive voltages. We said earlier that the input voltage must be higher than the output voltage, but that can still apply in negative voltages. Dual rail supplies are less common in maker circuits but have applications in amplifiers and some sensor situations. Any situation involving the current crossing zero (as opposed to reaching it but not crossing over into negative) is Alternating Current and requires a dual rail supply, with a positive rail, a negative rail, and 0V (GND) in the centre. However, this still requires a filtered, reasonably stable input, so it’s no good just putting AC with rectifying diodes on the inputs to the regulators and hoping for the best.

The 78XX family has some upside-down cousins, the 79XX family. There aren't as many of them, at retail at least, but the common voltages are present. These are the dedicated negative counterparts to each of the positive regulators, and are otherwise the same. They’re just as easy to use, but making a dual rail supply is a little harder than a single-rail supply. The most common design requires a centre-tapped transformer, which are not easy to find as pre-wired or plugpack arrangements. The transformers themselves are easy enough to buy, but mains wiring is illegal if you are unlicensed, regardless of whether the wiring plugs in or is fixed.

There are ways to produce a dual-rail supply from a single winding, but they require a lot more explanation that we can justify here: We want to explain the concept, so we’ll stick with the most common design. The centre tap of the transformer becomes the ground, and each end of the winding goes to the bridge rectifier. On the other side of that are the filter capacitors, which smooth the pulses from the rectifier.

Note the polarities of the polarised capacitors: In the negative half of the circuit, the positive terminals of the capacitors go to ground, and the negative terminals connect to the rails.


If the output current of a chosen (or available) regulator is not high enough for your needs, there are ways to boost it. In essence, this involves an external series pass transistor, using the regulator IC to control its base and utilise all the internal circuitry that comes with the IC. It does not make a lot of sense to use the 78L05 100mA regulator and a transistor to give a 1A output when the 1A 7805 is so readily available, but it does make more sense to use a 7805 and the right transistor to give a 4A output.

The circuit uses a high-current power transistor, and because it is on the high side of the load, it is a PNP type. The negative regulator equivalent uses an NPN type. Because the transistor is not operating at saturation, the constant current figure for the transistor is not going to be the same as the output of the whole regulator circuit. In fact, we found it very difficult indeed to find a way to calculate the current output that was both reliable and understandable. What is important is the current gain of the transistor in use. The current gain, shown as either Hfe or ß in datasheets, is the critical figure. Use the lowest value if there is a range. The gain must be greater than or equal to the desired maximum output current divided by the regulator current maximum.

It is important to note that this circuit is not protected against short circuits! If the output is shorted, there is no protection, and the transistor will overheat. Also, the transistor must be mounted on a suitable heatsink, because a significant amount of heat is generated by this circuit.


The fixed regulators in the 78XX family can be configured with an adjustment ability. There are also adjustable voltage regulators on the market. These are very useful if you want a continuous voltage adjustment, or just a value not easily obtained through the fixed versions. Adjusting the fixed range is useful if you just want a slightly different voltage, such as 5.5V from a 5V regulator. This circuit will not produce a lower voltage than the rated voltage of the fixed regulators. It works by attaching the ground connection of the regulator to a voltage divider and fooling the internal reference. R1 is fixed, but R2 can be fixed or a potentiometer wired as a variable resistor.

The input voltage must still be 2V to 2.5V higher than the output voltage. IQ is the quiescent current for the device, the current flowing out of the ground terminal as part of the normal operation of the regulator circuit. This is around 6mA for the TI LM7805 but check the datasheets for your device. Vn is the nominal output voltage for the device, such as 5V for the 7805. Don’t forget order of operations!

Some phone calculators do forget, and simply perform the calculation left to right. So, if you are getting odd results, just stack it with brackets. On a phone calculator, the equation might look more like: Vn + ( ( ( Vn ÷ R1 ) + IQ ) x R2) = VOUT.

In most cases, however, it is simply better to use a dedicated adjustable regulator. The most common of these is the LM317, or its negative counterpart, the LM337. The LM317 is a 1.2V to 37V output continuously adjustable regulator. Besides being adjustable, all of the information regarding the fixed regulators is valid, including the fact that they generate heat and require heatsinking. In fact, that could be more of a concern with the variable devices than the fixed ones, because it is more likely that there will be a big difference between input voltage and output voltage. All of the protections apply, and the output current can be boosted in the same way as the fixed regulators. However, there is a 1.2V to 33V 3A version, the LM350, but it is in a TO3 package which many people find painful to use.

The circuit presented is not the most basic one in the LM317’s datasheet. In fact, the most basic looks exactly like the circuit for adjusting the fixed voltage regulators. This one is further into the datasheet and is designed for increased stability and ripple rejection. Take note also of the different capacitor values compared to those previously shown. The IC functions by adjusting its output so that there is a 1.25V drop across R1, between the output and the ADJ pin. This causes a constant current to flow through R2, along with the 100µA current from the ADJ pin.

There are various combinations of components, and other ways of displaying the formula that you may encounter. Many designs ignore IADJ, use a nominal 120Ω resistor, and flip the equation:

where R1 is already set at 120Ω, and VREF is 1.25V. This formula works well and we have succeeded with it many times. It leads to the values shown in the circuit above, good for a 1.2V to 25V power supply. Just remember an adequate overhead (2.5V minimum) at the maximum voltage, and that heat will be an issue.


The 78XX series can also be used as current limiters. In this mode, they do not reduce voltage, so input voltage should be close to the desired output voltage (there is a minimal loss internally). A single resistor is used from the IC output to what would in a normal circuit be the ground rail, but in this case, it is the output rail. The voltage drop across it is used in the calculation, and relies on a few factors.

The formula shows that output current (IOUT) is found by dividing the voltage drop across the resistor by the resistor value, and adding the quiescent regulator current (IQ). However, in most uses, you will already know the desired output current, and will have to adjust R1 until it gives a voltage drop at that current of the right value to give the output current you need.


There are always reasons for and against a particular design choice, and using a linear regulator is no different. Whether or not you should use a switch mode power supply or a linear regulator depends on the end use, and your own skillset. Sometimes it’s convenient to buy a pre-made module, and most of those are switch mode. They run cooler than linear versions, too. However, linear regulators are inherently rugged, and even basic designs incorporate a lot of protection just from the IC used. They are second to few when it comes to noise-free outputs, which is critical in audio situations and many sensor situations. They are also cheap, and readily available in a range of fixed voltages with good tolerances.

Contrast that to having to design your own switch mode power supply if one is not around that suits your needs. These have a far higher component count and careful design is critical to avoid excessive noise and give a clean output. Additionally, cheap capacitors are notorious for exploding on the high-speed switcher of a switch mode power supply, and makers cannot always choose what they get. We’ll cover the details od=f SMPS in another Classroom, because they will take a whole article to do justice to.


As an aside, some power supplies are produced with both switch mode and linear elements. These exploit a property of electricity in that the higher the frequency, the lower the resistance provided by conductors.

Switch mode power supplies (SMPS) work by first rectifying the incoming AC into DC, then chopping it up into a high-frequency wave. This is often a square wave, but sine wave designs exist, particularly at the upper end of the quality bracket. The frequency also varies markedly. At a high frequency, the voltage is fed into a transformer, before being rectified again on the other side at the transformed voltage.

The effect of the high frequency and lower resistance is that very small wire can be used for the transformer. Not only does this make them dramatically lighter, it also makes them cheaper. Despite the design required and the component count, it is still cheaper than the large amount of copper required to wind a transformer of equivalent capability at 50Hz.

The side effect here is often noise. Even with good filtering, it is hard to eliminate the switching noise. This is where having a linear regulator comes in handy. The SMPS gets the voltage down (or up) to near where it needs to be, then the linear regulator is used to smooth off the top. This helps reduce noise and gives a very clean output.

Even more advanced designs such as some lab or benchtop power supplies even have circuitry inside to vary the output of the SMPS, so that there is always a minimum dissipation in the linear regulators when a variable output is used.


As summarised in the tables earlier, there are plenty of linear regulators available over the counter at Jaycar and Altronics. Online retailers vary, particularly those aimed at the Arduino and Raspberry Pi type making rather than making with discrete components. There is nothing wrong with that, by the way. It just means that sometimes you’ll only find pre-built modules, usually SMPS, for the voltages commonly used in those fields.

If you need something more specialised, you will have to look at somewhere like Element14. Element14 is a commercial supplier aimed at trade customers, but who deal with the public. While they are happy to sell to individuals, bear in mind that you are expected to be self-sufficient.

The website can be daunting, but you can find things that you can’t get at retail. For example, we recently purchased a bunch of 78L33, which is actually a 3.3V rather than a 33V regulator in a TO92 package with a 100mA output. These will make a future project much, much easier and more compact, but Element14 was our only source of them.

We were going to cover switch mode power supplies, but to do them justice, they need a full article on their own. So, watch this space!