Choosing a voltage regulator

Chances are that if you look inside the electronic items in your home or office you will find plenty of linear regulators and you have probably used a fair number of them in your own projects. Being so inexpensive and simple to use, they are routinely used to regulate circuit supply voltages without much thought or consideration. A simple and effective method made popular by the introduction of three-terminal fixed voltage IC regulators. Despite many newer components now available, regulator designs that date back to the 1970’s are still widely used. Is there a better option?

Linear regulators

The basic idea behind a linear series regulator is that a pass transistor is placed in series with the load to create a variable resistance, creating a voltage divider that is automatically adjusted in response to changes in load or source voltage to maintain a near constant voltage.

Linear regulators are very popular and have a wide range of uses, but their main disadvantage is lost power due to dissipation through the pass transistor. A linear regulator, is so called because the pass transistor is operated in its active (or linear) region and is therefore creating a resistance, like any resistor it has to dissipate the excess energy in the form of heat. As the load current increases the pass transistor has to dissipate more power as heat.

Linear regulators are most problematic in low voltage – high current situations, requiring large heatsinks or even fans to remove the heat generated. It should be self evident that the more heat generated the more energy is wasted and therefore efficiency falls.

Here is how to calculate power losses:

Power wasted = (input voltage – output voltage) x load current

For example, we have a 9V battery and you want to power a digital clock circuit based on a microcontroller that draws 10mA, and a display that draws approximately 50mA. Both the microcontroller and the display require 5V so we choose the ubiquitous 7805 regulator to step the voltage down to 5V from 9V.

Power wasted = (9V – 5V) x 0.06A = 0.24W

0.24W is not too bad for power losses. The 7805 can easily handle this without a heatsink, the power consumption is so low that the battery life will be very long anyway so there seems little point in considering anything else. If you were designing this circuit aiming for the lowest manufacturing costs and you do not care about the cost of batteries, you would be perfectly happy with this arrangement. If however we dig a little deeper and compare the total power consumed against power wasted it doesn’t look so good.

Total used = 9V x 0.06A = 0.54W

45% of the power is wasted, we could significantly improve battery life if only we could increase the efficiency of the regulator.

Reducing the battery voltage could potentially help reduce losses, for example using a 7.5V battery, but there is a problem with the doing this with the 7805. The voltage in a battery is not constant, it has a discharge curve where the voltage falls as the battery discharges, attempting to compensate for the falling voltage during discharge is one of the reasons we want a regulator in the first place. A linear voltage regulator requires a minimum voltage headroom where the input voltage is always higher than the output to function correctly. The minimum voltage differential is called the dropout voltage, if the input voltage drops below this the regulator can no longer maintain a constant output voltage and will begin to drop below our target voltage.

Most standard linear regulators have voltage drops of 2V or more, including the 7805. The input voltage must always be kept at least 2V higher than the output voltage to guarantee a constant voltage. At first glance it may seem using a 7.5V battery in this case would mean that we have a 2.5V overhead so we have the perfect solution. When we take the typical discharge curve into account for 1.5V alkaline cells, after 4 hours use, the battery pack voltage will have dropped to 7V and we can no longer guarantee that the circuit will work. The batteries have plenty of power left in them but can no longer be used to power our circuit and will need replacing prematurely.

Powering our circuit from a 7.5V mains power supply that will not drop below 7V, would mean we can confidently run the regulator from a voltage that is very close to the dropout threshold to achieve maximum efficiency possible with this type of regulator. While not perfect it is likely to be good enough to be accepted for such a low power circuit. What however, if we have no choice but to run from a battery or just need further efficiency gains?

Low dropout regulators

Before we consider a switching regulator we have a much simpler option, the low dropout regulator (LDO). The LDO regulator is a high efficiency version of the linear regulator. As the name implies the amount of headroom required, or dropout voltage, is lower than a standard regulator.

As we have a choice of what battery we use to power our example circuit, we can improve our efficiency simply by choosing a 7.5V battery made from 5 x 1.5V cells and an LF50 LDO regulator that has a 0.45V rated dropout voltage. Our cells will be able to run down to 1.1V each and our battery will be producing 5.5V. A typical alkaline cell dips to about 1.1V when it is exhausted so we should be able make full use of the available battery capacity.

If we now re-visit our calculations we find the amount of wasted power has reduced.

Power wasted = (7.5V – 5V) * (0.06A) = 0.15W

With 28% of our power wasted, we have achieved an efficiency of 72% just by choosing an LDO regulator and a well matched battery voltage. If we have the luxury of powering this device from a 6V mains power supply that will not drop below 5.45V our wasted power drops to just 0.06W, 90% efficient. We clearly have very little to gain by using a switching regulator in these circumstances.

Standard regulators can usually be easily replaced by LDO regulators, so why are they not used more widely? The main reason is simply convention, often a part continues to be used simply because it has always been used. If the part is inexpensive and good enough the attitude tends to be why risk potential problems by changing to something else. In hobby electronics the problem is exacerbated by the way circuits are shared. The designs are usually deliberately based on easily accessible parts that tend to change very little over time. Certain components end up being included by convention, creating an informal standard.

There are a few considerations when using an LDO regulator. Care must be taken to observe the maximum input voltage specification, which can be lower than that for standard regulators. The output impedance is typically higher than standard regulators making them less well suited for driving very low impedance or capacitive loads. Some LDO regulators (particularly older designs) may need specially chosen external capacitors to maintain stability.

The low overhead is usually achieved at the expense of power handling, a similarly priced LDO regulator will typically have a lower current rating than the standard type. To increase the power rating a larger transistor is required and inevitably the price increases. As modern electronic components use a fraction of the power it is likely that the high power capability of a standard regulator is far more than required and it can be replaced with a similarly priced LDO with a lower power rating.

If your supply voltage is fixed, for example our clock circuit has to be run from a 12V (actually up-to 13.8V) car battery? Our input voltage is fixed and an LDO regulator is not going to help, it will perform exactly the same as a standard regulator. The only way to improve efficiency in this scenario would be to use a switching regulator.

Switching regulators

Although linear voltage regulators continue to enjoy widespread use, switching regulators are increasingly being used due mainly to their greater efficiency. Perhaps you feel you should be looking at switching regulators but are uncertain if they are the right fit for your application.

Let’s consider a new example, suppose we have a circuit that draws 1A that also requires 5V from a 9V supply. How much power is wasted in a linear regulator now?

Power wasted = (9V – 5V) x 1.0A = 4W

4 Watts is a lot of waste heat! The 7805 would get so hot it would certainly be too hot to touch and without a suitably large heatsink likely exceed the rated maximum operating temperature. Even with a suitable heatsink, 4W is a big waste of battery capacity for no benefit, you are going to go through batteries like there is no tomorrow. If the source voltage needed to be increased, to 12V, we would have 7W of heat to disperse with a linear regulator, this would require an enormous heavy heatsink.

In situations where there is a high current load and/or large voltage differential between the input and output a switching regulator offers significant efficiency gains. To calculate power losses with a switching regulator, we simply have to measure the power drawn from the supply and compare it to the power consumed by our circuit. Here is how to calculate power losses:

Power wasted = (input voltage x current drawn) – (output voltage x load current)

Input power: 9V x 0.6A = 5.4W
Output power: 5V x 1A = 5W
Wasted power: 5.4W – 5W = 0.4W

Our power losses have been reduced to around 0.4W or just 10% of that wasted with the linear regulator.

As the name suggests with a switching regulator, the pass transistor is used as a switch, the transistor is either saturated (on) or non-conducting (off). A transistor operated in its active region will be required to dissipate a great deal more energy than it would when saturated resulting in little heat generation while saturated. By rapidly switching, small chunks of energy are allowed to pass, by varying the duty cycle, altering the ratio between the on and off time the amount of energy can be controlled. The output is a series of pulses between ground and the unregulated input voltage, clearly this can not be used as a power source as it is. To convert the series of rapid pulses into a usable DC power supply the output is fed into an LC (Inductor Capacitor) filter, we now have a DC voltage, magic!

Switching regulators are enormously flexible, unlike linear regulators, boost switching regulators can be used to create higher voltages than the source. A class of switching regulators known as boost/buck regulators can seamlessly provide a regulated output from a wide range of source voltages that can be higher or lower than the output voltage. With such enormous efficiency gains and flexibility you may ask why anyone would still be using linear regulators, where is the catch?

The most significant reason switching regulators are not used as often as they otherwise might is the up-front cost. Significantly more components are required, and therefore higher costs are involved when including a switching regulator in a circuit. In a competitive market place traditionally a buyer is more interested in the purchase price than running costs. Of course the small amount of extra cost at the time of purchase will soon be recouped in the cost of batteries or electricity in a device that is used a lot, purchasers are more aware of this than they used to be.

Switching regulators are a little more complex to use, they require a little more planing than a linear regulator. A suitable inductor value needs to be selected dependent on the input voltage and output load current. A low ESR capacitor is usually used on the output and a catch diode is required. The additional complexity may be a little off putting at first, but it really is not as difficult as it might at first seem. Data sheets usually provide reference designs and inductor selection guides, if you want things even easier however, there are many low cost ready-made modules now available.

Noise can be a problem, all of this rapid switching creates electrical noise, ripple on the output and even sometimes audible sound from a resonating inductor. Obviously this has a number of implications, if for example you are making an audio amplifier this noise can easily appear on the output as annoying interference. Perhaps more importantly however is that if a circuit is being designed into a product then it must pass EMC emission tests. A switching regulator adds significant additional emissions to the design that may make it more difficult to meet the required standards.

The best of both worlds

As switching regulators are noisy it is common to see a switching regulator doing the heavy lifting while an LDO linear regulator provides power to sensitive parts of a circuit. A typical example would be a circuit that needs a variety of voltages such as 12, 5 & 3.3. A switching regulator can be used to bring the voltage down to 6V and a linear regulator used to provide the 5V followed by 3.3V. Instead of the 5V regulator having to drop 7V it only has to drop 1V, the use of the linear regulators also helps reduce some of the switching noise from the supply to the 5V & 3.3V rails.

Conclusion

Clearly a switching regulator is not always desirable except where your main requirement is for efficiency or where there are large loads to be regulated that could not easily be achieved with a linear regulator. If you are building a device that will see occasional use, do you really care that it might cost £10 more to run over its lifetime? If however, that device is to be operated on a camping trip half way up a mountain then perhaps getting the maximum life out of a set of batteries would be your main concern. Equally if the device is intended to be operated 24 hours a day it could be costing you twice a much each year to run than necessary, again a good reason to take a closer look at efficiency.

If you have a low power application where the input voltage is selectable, look at an LDO regulator for greater efficiency. As a general rule of thumb, if your linear voltage regulation solution is wasting less than half-a-watt of power, a switching regulator would be overkill for your project. If your linear regulator is wasting several watts of power, you most certainly would want to consider replacing it with a switching regulator. Where the input voltage is much larger than the output, a switching regulator will offer significant efficiency gains and may in some cases be the only viable option. If your circuit is likely to be sensitive to the ripple caused by a switching regulator then a hybrid solution consisting of a switching regulator and an LDO will help reduce the unwanted noise. Try to avoid using standard regulators where possible unless nothing else will work or you really don’t care about efficiency.