DC-DC switching voltage regulator: Buck converter

Power MOSFETWhen I took a power electronics course at MIT (6.334) almost two years ago I instantly fell in love with the simplicity and ubiquity of switching power supplies. They are so simple and interesting that I decided to devote a few posts here describing how they work. Voltage regulators are interesting and useful to learn about because they are used everywhere. Electronic components (for example, those within your home electronics equipment) often require different voltages, so voltage regulators included in the equipment to generate those voltages. The first question to answer in this post is “What is a voltage regulator”:

What Is a Voltage Regulator / Converter?

Buck (step-down) and boost (step-up) voltage converters take an input voltage source and either drop the voltage down (buck) or raise it up (boost). Buck and boost converters are basically the same circuit, except that for a buck converter you place the voltage source on one side of the circuit and for a boost converter you place the voltage source on the opposite side of the circuit. This post describes a buck (step-down) converter.

What Is a "Switching Regulator" (What Does "Switching" Mean)?

It is called a "switching regulator" because the way it generates new voltages is by switching the source voltage on and off very quickly (using a power MOSFET). For a buck converter, the output voltage becomes the average of the source voltage being turned on and off. For example, if the source voltage is 12 volts (V) and you leave it switched on for only half of the time, then the output voltage will be 6 V – or half of the input voltage (i.e., 12 V * 0.5 = 6 V). If you leave the source voltage switched on for only 1/4 of the time, then the output voltage will be 3 V – which is 1/4 of the source voltage (i.e., 12 V * 0.25 = 3 V).

Simple Buck Converter Explanation

The circuit below shows the most basic buck converter. There are a few points of which to be aware before getting into the details of how the buck converter works. First, note that voltage across an inductor can change instantaneously but current through it cannot (i.e., current ramps up or down). Second, assuming stable converter operation, the average voltage across the inductor will be zero (i.e., the average voltage at Vx will be the same as Vout). Third, current through a capacitor can change instantaneously but voltage across it cannot (i.e., voltage ramps up or down).

Basic buck converter: N-channel MOSFET for switch 1, flyback diode for switch 2 and no input or output damping.

Now, getting to the details of how the buck converter circuit above works… Let us assume that the buck converter is already running and we are trying to get a snapshot of what its operation looks like. In other words, for now we will not look at what happens when the circuit turns on after having been powered off for some time, but instead we will look at it after it has already reached steady-state.

  1. First note that there are two inputs to the circuit: Vin and Vswitch.
    • Vin is the energy source – we will be transferring energy from here to the load.
    • Vswitch is a control signal used to turn on and off the MOSFET to control how much energy we transfer from Vin.
  2. Assume we start observing the circuit when Vswitch is high and therefore the N-MOSFET is on, allowing Vin to pass through.
    • In this case, Vx = Vin
  3. Being a buck converter, Vin is larger than Vout, so with the MOSFET on Vx > Vout is also true.
    • With Vx > Vout, this means that the current through the inductor (left to right) is gradually increasing, which also slowly charges up the voltage on the capacitor.
  4. After some amount of time (e.g., a few microseconds) we force our control voltage Vswitch to go low, which turns off the N-MOSFET.
  5. Because of the fact that the current through an inductor cannot change instantaneously, when the MOSFET turns off then the current through the inductor will take the path of least resistance which is through the diode.
    • This means that Vx will be approximately 0 V.
    • In practice, Vx will actually be a slightly negative voltage (GND – VF,diode), but for this explanation we will assume it is 0 V.
    • Using a diode in this manner is so common in circuits that there is a special name for it: a "freewheeling" diode.
  6. With Vx = 0, or in other words Vx < Vout, this means that the current through the inductor (left to right) is gradually decreasing.
    • At some point the current through the inductor will be less than the current needed by the load Rload, and so current will start to come from the capacitor. When this occurs, the voltage across the capacitor will start to decrease gradually.
  7. After a few more microseconds we force our control voltage Vswitch to go high again, turning the N-MOSFET back on, and the cycle that we’ve just observed starts all over.

Calculating Vout Versus Vin:

Now remember one of the first notes I pointed out – that in steady-state operation the average voltage across the inductor will be 0, and this means that Vout is the same as the average value of Vx. This tells us that Vout will be somewhere between 0 V (the lowest value of Vx) and Vin (the highest value of Vx).

In fact, we can control what the output voltage will be by controlling how much time we leave the MOSFET turned on relative to how long we leave it off. This on/off relationship is called the duty cycle. The longer we leave the MOSFET on, the closer Vout will be to Vin. If we represent this duty cycle with the letter "D", then the equation for Vout becomes:

Vout = D × Vin

Simple!

Important Considerations to Take

  • Input and output damping should be used (not shown in figures above). Damping prevents the input and output voltages from oscillating to very high levels that are out of the components’ safe operating areas (SOA). Damping is achieved with a bulk capacitor and relatively small resistor in series between the input and ground and/or between the output and ground
    • Rdamp value should be chosen correctly for best performance.
  • For a simple voltage converter, make sure that the MOSFET is left on long enough so that during the time that the MOSFET is off the current through the inductor does not drop to zero (although this is allowed in more advanced voltage converter designs).
  • The Vswitch voltage level needs to be at least several volts higher than the Vin voltage level. This is because the N-MOSFET’s gate-source threshold voltage is several volts, and when the N-MOSFET is on then its source voltage will equal Vin.
  • The switching frequency of Vswitch must not be too fast or too slow. The switching frequency is usually somewhere between 200 KHz to 1 MHz depending on the converter design.

Benefits & Drawbacks

Some of the benefits and drawbacks to using a buck converter are:

  • (+) Very simple design
  • (+) Very good efficiency
  • (-) Must compensate for voltage drop across diode and N-MOSFET
  • (-) Switching nature of design means inherent voltage ripple exists on the output
  • (+/-) Must be extremely careful not to violate any component’s safe operating area (SOA) (e.g., make sure to use capacitors rated for high ripple current and design safely for the high power dissipation that occurs during Vswitch transitions)

Tweaks & Tradeoffs

  • You can replace the freewheeling diode with an N-MOSFET to prevent the voltage drop across the diode, but then the Vswitch control logic must be much more sophisticated to prevent both FETs from being turned on at the same time.
  • Use zero-current switching (ZCS) and/or zero-voltage switching (ZVS) to increase efficiency, but then need much more circuit design effort and more sophisticated Vswitch control. (Here are some really good slides on ZCS and ZVS from the University of Colorado. You might want to start with page 29.)
  • Use snubbers to to help keep components within their safe operating areas (SOA), but then you will either lose efficiency or will need a relatively complex converter design.

8 thoughts on “DC-DC switching voltage regulator: Buck converter”

  1. Hey Mike,
    I stumbled across your website and found your projects very interesting. I have a (brief) background in physics and electronics, but work in the IT industry today.

    I was hoping to get your input on a project my son and I are working on. We’re trying to build a tiny 5V DC, 5 -10 Watt power supply from 120 VAC. We’ve outlined that there are 3 parts to the typical wall brick – Voltage Reduction (transformer), rectifier (bridge), and regulator (zener and cap). First we tried eliminating the transformer, and used an X2 cap to reduce current to about 40 mA. This works fine for a fixed load, but we’re shooting for a USB style charger which could see variable loads. Then we just went bridge rectifier and cap, giving us 170 VDC, then resistors to get 5 VDC. This seems like the simplest, smallest solution, but have to dissipate 5-10 W via resistors. We experimented with rectifying it and creating a DC-DC buck converter, but it seems the inductor will be rather large. My goal is to make it fit inside of a typical plug head, thereby “hiding” the power supply.

    Would you be willing to provide any feedback on this project?

    Many thanks,
    Patrick Reynolds

  2. Patrick,

    It is very difficult to go all the way from 120VAC to 5VDC with a single rectifier and buck converter and avoiding the transformer. A few things to consider:

    1. The transformer provides isolation. Without isolation, if the wall outlet was wired incorrectly, then when someone plugs their phone into your transformer-free charger they will be picking up a device that could be sitting at 120V potential above ground and get a big shock! (See: http://www.diyaudio.com/forums/power-supplies/116436-120vac-straight-wall-rectified.html)
    2. Going from 120VAC (or around 70-80VDC when half-rectified), down to 5VDC requires at best a 7% duty cycle on the switching regulator, and the frequency must be very low for the high voltage swing. The low frequency causes bad output ripple and/or means you will have a huge inductor (like you mentioned previously).
    3. Check out Linear Technology’s website, they are arguably the most advanced manufacturer of switched power supplies. The fact that they don’t produce something like this already is saying something, considering how ubiquitous it would be. These are the closest products I could find on their website: http://www.linear.com/product/LTC3810 and http://www.linear.com/product/LTM8027
    4. Try using Linear Tech’s free LTspice program. I use it all the time and it is very useful for simulating any of these circuits that you are curious about before you actually build something. (http://www.linear.com/designtools/software/#LTspice)
  3. Patrick,

    Take a look at Power Integrations’ LNK302 or LNK304. It does exactly what you are looking for. Regulation is not very tight, so if you want 5 volts precisely, use the LNK304 to make 7.5v or so, then use a linear to get you to 5vdc. As noted, these solutions are NOT ISOLATED, so be ye wary.

  4. Wonderfully Explained …

    You said t

    ARDUINO UNO R3
    The base frequency for pins 3, 9, 10, and 11 is 31250 Hz.
    * The base frequency for pins 5 and 6 is 62500 Hz.

    **** The switching frequency is usually somewhere between 200 KHz to 1 MHz depending on the converter design.

    How would i be able to switch with such high rates. I have max 62.5Khz on pins 5 and 6. Awaiting your guidance.

    23year old, solar power enthusiast, Working on Solar Max Peak Power Tracking.

    Regards.

    1. A few hundred kilohertz is actually pretty slow. You can probably get the pins to toggle faster than that. If it is indeed the “base freqency” as you’ve mentioned, then that usually implies that it is the “base” for other frequencies that you can generate – so you can probably change multipliers and other settings to get different frequency outputs. Perhaps look at the PWM features of the microcontroller. Take a look in the datasheet to see what other options might be available – the datasheets and user guides usually have everything you need.

      Good luck!

  5. Hey i’ve done some of boost converter project but it seems to be when the voltage is step up the current output seems to reduce. Is there any circuit that can boost voltage as well as boost current at the same time?

    1. No, that would generally be impossible. The amount of power going into the converter (voltage * current) must be the same as the power going out of the converter. What you are describing is generating more power than what you are putting in (same description as the so-far-impossible perpetual motion machine or cold fusion). In other words Vin * Iin = Vout * Iout (minus some efficiency losses in the form of heat).

Leave a Reply

Your email address will not be published. Required fields are marked *