Voltage regulators come in two main flavours, linear and switchmode (DC-DC). The output of a linear regulator is always lower than the input, this may only be by a couple of hundred millivolts depending on the regulator but will always hold true.
A switchmode regulator can have an output that is higher than the input (boost mode), lower than the input (buck mode), greater or less than the input voltage but opposite polarity (buck-boost), greater or less than the input voltage and the same polarity (SEPIC). For this blog, we will concentrate on buck regulators.
The biggest issue with Linear regulators is their potential for heat dissipation when driving heavy loads. The power dissipated by the regulator itself is a function of the voltage it is dropping and the current required by the load. This can be expressed as (Vin – Vout)*Iload. Suppose we have a small circuit that draws 200mA at 5V and we wish to power it from a 12Vdc supply. In this situation, Vin = 12V, Vout = 5V and Iload = 0.2A. This means that the regulator will dissipate (12-5)*0.2=1.4W. Therefore the regulator is drawing around 115mA from the supply and turning it into heat.
If the circuit required 3V instead of 5V as many modern micro-controller circuits do, then the situation is even worse. Our regulator is now wasting 1.8W of power. That’s 3 times the amount of power being consumed by the circuit itself!
With this in mind, if a linear regulator is required by the circuit for other reasons, then it is essential to keep both the voltage drop and the load current as small as possible, otherwise, thermal management may be required.
Buck regulators mitigate the heat dissipation issue by chopping the input voltage and storing the average in an inductor and capacitor. By doing this quickly (10,000+ times a second) an acceptable approximation of the required output voltage can be produced. To ensure the output is suitable for the load, buck regulators use a diode, inductor and capacitor to help smooth the large spikes into a DC voltage.
So far, we have identified that linear regulators can generate a lot of wasted heat and that buck regulators are generally a lot more efficient, but just how much more efficient are they?
In general, buck regulators come in at around 80-90% efficient, that said some designs are now pushing greater than 95%. With this in mind, let’s take a quick look at the power dissipation of a buck regulator in place of the linear regulator in our examples above.
For the moment, we are going to assume that the buck regulator we have replaced the linear with is 80% efficient, everything else however stays the same.
Our first circuit draws 200mA at 5V which equates to 1W. Given that our regulator is 80% efficient, the circuit will draw a total of 1.25W from the power supply. Therefore the voltage regulator is drawing 0.25W. A far cry from the 1.4W being consumed by the linear regulator. In the second example, our load required 200mA at 3V, therefore drawing 0.6W, in this case, the buck regulator will consume around 0.15W giving a total power draw of 0.75W. Notice how with a buck regulator, the total power consumption has reduced, whereas, with the linear regulator, we still consumed the same amount of power from the supply, it was the ratio of useful power to wasted power that changed.
If our friendly switchmode regulator is such a great device then why do people still buy linear regulators?
Whilst the buck regulator sounds like the answer to all of our problems, it does have drawbacks of its own. Firstly cost, generally, a minimum of 5 components are required:
- Input capacitor
- Swiching regulator
- Flyback diode
- Output inductor
- Output capacitor
Compare that to a linear regulator with its input and output capacitors and we have just added an extra diode and inductor into the mix. These extra components also add to the board space required by the regulator. For some bigger designs, however, both of these may be offset by the thermal management requirements of the linear regulator (i.e. a heatsink and possibly a fan).
Being a switching device, (some buck regulators can work in the MegaHertz region) there is the potential for significant RF (radio frequency) interference to be generated. This often requires careful component placement and trace routeing on a PCB to ensure that emissions both at the fundamental frequency and harmonics thereof are kept to acceptable levels. Some switch mode regulator designs can be built on stripboard however they are unlikely to run as efficiently as they are capable of, while others may not run at all.
So how do we decide?
Ultimately it falls to the designer to decide firstly, the most important requirements of the circuit being designed and secondly, which solution best meets those requirements. Often, as with many things, this will be a compromise.
When making the decision, however, it is recommended that the designer keep the power equation in mind, as even small load currents can result in large dissipations. For example, a 100mA LED reading light could result in a 2.1W regulator dissipation if its intended environment is a lorry with a 24V electrical system.
Does this mean the linear regulator is dead?
Certainly not, the linear regulator still has a place in modern electronics design. For example, a predominantly 5V design may include 1 or 2 3V only components, given the voltage drop is low and (depending on the components) the current is very small, a linear regulator lends itself perfectly to the application.
Another situation where a linear regulator may prove useful is in a noise-sensitive application, for example, audio devices or sensitive sensor circuits. In these situations, it may be difficult and expensive to filter the output of the switching regulator sufficiently. Therefore a linear regulator may be used, either on its own or after a switch-mode regulator to give the required signal quality.
Whilst certainly not exhaustive, the following links may be of interest to anyone considering using a switch-mode regulator