By Chris Francis
Switching regulators are everywhere – buck, boost, SEPIC, Cuk, flyback etc. They are for converting power from one voltage to another in a relatively efficient manner. Whereas with a linear regulator you might lose 50% of the power or more in heat, switching regulators can reach efficiencies of close to 100% for synchronous converters. To choose which you need you need to specify whether you are increasing, decreasing or inverting the voltage. Here we will look at the buck converter.
A buck converter or buck regulator is for stepping down the input voltage. The basic configuration is shown below with an ideal switch:
By simulating the circuit you can start to see what is going on. The switch would most likely be a MOSFET in a real circuit, although quite a few regulator ICs have the switching transistor built into the control IC. There is no control loop in the circuit shown. In a real switching regulator you would need a control loop to alter the duty cycle of the switch control waveform to maintain the desired output voltage.
What you are effectively doing is storing energy in the inductor and transferring that energy to the output. The amount of energy is 0.5LI2 where L is the inductance in Henries and I is the inductor current in amps. The longer the switch is turned on the higher the inductor current and hence the higher the stored energy. So, for a light load the switch would be turned on for less time in each cycle i.e. a lower duty cycle.
The output capacitor smoothes out the energy bursts but there is still some residual ripple, as shown. The current doesn’t actually drop to zero each cycle so all the stored energy is not removed from the inductor.
While a larger inductor stores more energy for a given current, large inductors are, well, large. So, the trend is to keep the inductor small and increase the frequency. For a given inductor and current you can obtain more power out of the converter if you can switch faster. This does rely on having enough time to allow the current to build up and devices fast enough to switch quickly. Fortunately devices are a lot faster nowadays than 20 or 30 years ago so you will see switching frequencies of several Megahertz whereas in the past they would have been a few tens of kilohertz.
To work out the efficiency you need to find the losses in the system. With a simulator you can plot the power in each device. A real transistor in place of the switch would account for some losses, partly due to the finite switching speed which is one reason why you cannot just keep increasing the switching frequency without losing out somewhere. The switch resistance is important as well but that can be very low. Inductor losses are another source of energy lost to heat. However, a big loss particularly for low voltage supplies is the diode. The diode is necessary to clamp one end of the inductor to ground when the switch is turned off. Even with a Schottky diode you will lose 0.3V to 0.4V, and if your supply is 1.2V (for an FPGA power supply for example) that can result in a large loss.
MOSFETs with very low ON resistances are readily available so one solution to improve efficiency is the synchronous converter. Many of these are available with the switches built in to the control IC such as the Texas Instruments TPS63020. This has internal switches and achieves efficiencies of up to 96%. The principle of the synchronous buck converter is shown below.
The diode is simply replaced by another switch and that switch is driven in anti-phase with the series switch i.e. when one switch is ON, the other is OFF and vice versa. In practice it is a little more complex because you want to avoid both switches being ON at the same time because that will short the input supply voltage to ground. With finite device switching times to consider (and different ON and OFF switching times) you would need to carefully engineer the delays to maximise efficiency. That is why an integrated solution works well. With complete knowledge of the characteristics of the devices within an IC the manufacturers can build in the appropriate delays to ensure maximum efficiency. If you are doing it with discrete components you need to add appropriate delays yourself to suit the devices you are using to avoid a switch “clash” and wasted power. You also don’t want the switches to be both off for too long otherwise the free end of the inductor will swing to a high negative voltage until something breaks!
The effect of the synchronous switch can be seen by looking at the junction of the switch/diode/inductor and comparing it with same point on the synchronous version:
Whereas the diode drops around 0.6V (right image) the voltage drop when the diode is replaced by a switch is virtually zero – mainly because in this example the switch resistance is 10 milliohms. Even with a 0.1ohm switch the voltage drop would be less than 0.1V – considerably less than with the diode. With an average diode current of around 0.5A, that is a lot of lost power which can be saved.
Filed Under: Commentaries • insights • Technical thinking, Power supplies, Power Electronic Tips