What are the critical considerations when selecting between buck regulators or voltage regulators?

Apologies in advance if this is a basic question, I'm a total beginner.

I'm planning a project that uses USB-C Power Delivery to provide 12v power. I have a component that requires 12v. However I also have components that require 5v and also 3.3v for the microprocessor.

I am considering using either buck regulators (e.g., something similar to this SY8286ARAC) or voltage regulators (e.g., something similar to this TLV1117).

After much reading around, I've learned that buck regulators are better suited to some projects while voltage regulators are better suited to others. What I'm not clear on however is the main considerations to decide between them?

Is temperature the main consideration, or are there other important issues I should also be thinking about?

The main consideration is usually the efficiency of the voltage conversion process. Voltage regulators tend to be wasteful when reducing voltage as they produce heat as a by-product whereas buck converters do not

To add to what @UKHeliBob said,
A voltage regulator dumps excess voltage as heat, so to give you a typical example if you have a 12V supply and want 5V then 7V (12 - 5) effectively gets wasted as heat. To put this in perspective, if you are using a 12V battery then measure along the length of the battery 7/12th of the way from one end, that much of the battery is completely wasted.

A buck converter effectively swaps voltage for current, so by converting 12V to 5V you get a lower voltage out in return for drawing a lower current from the battery, and so not wasting the battery.

No voltage conversion is 100% efficient, so some power is wasted whatever kind of converter you use.

Buck regulators are also "voltage regulators". :wink:

A buck regulator (switching/switchmode regulator) or switching/switchmode power supply is almost always "better" than a linear regulator or linear power supply. But it requires an inductor and it's harder to build if you're building the circuit yourself.

Most modern "larger" (high current) power supplies are switching supplies. I don't think there has ever been a PC with a linear supply.

A small low-power linear regulator is cheaper and easier to build than a switching regulator and and since the total power is low, efficiency and wasted power/heat is usually not a problem.

The 5V and 3.3V regulators on the Arduino are linear. (It's simpler and it saves cost). But it means you can't power a lot of "other stuff" through those regulators because they can overheat and shut down.

Linear regulators can also have less electrical noise which can be an advantage in analog applications such as audio amplifiers. But that doesn't mean you can't use a switching supply/regulator as long as it's a good design, and the switching noise is often above the audio range. A lot of modern audio power amplifiers do use switching supplies.

If you are a manufacturer the extra cost of the inductor and more complicated circuit is (at least partially) offset because the components don't have to dissipate as much wasted heat so they can be lower powered and it may not need a heatsink.

In the case of a full power supply, the main transformer runs at a high frequency in a switching supply so it's small, lighter, and cheaper than a 50/60Hz transformer. But those are nearly impossible to build yourself.

1 Like

Switching supplies (and 5V USB powerbanks) can cause significant de-sensitizing of RF modules and GPSs.

If your Arduino needs a very low sleep current this is easy to arrange with linear regulators, but a low sleep current may not be possible with most swiching regulators.

So given what has been explained by others, it is useless to discuss this further unless you choose to reveal exactly what the "components" in question are.

I would also be interested to know how you propose to "use USB-C Power Delivery". :astonished:

@Paul_B thanks for your valuable contribution to this thread.

The components are a 12v display, some 5v sensors and a 3.3v microcontroller. But to be honest it is not particularly useful to get into specifics. Searching online there are many such examples explaining "with project design A, you should use a buck regulator" or "with project design B, you should use a voltage regulator". What was difficult to find (at least for me) was the general principles behind these decisions. For the purpose of this question, general principles that can be applied to any design are of most interest, and this has indeed been provided.

For USB power delivery, I will use a controller IC to negotiate a 12v supply from the source rather than the default 5v. Using something similar to this STUSB4500.

Thanks @UKHeliBob, @PerryBebbington, @DVDdoug, @srnet

This is really useful information, I've learned a lot.

One of the reasons for this question was that, while I had noticed that buck converters appeared "better" / more efficient, I'd also noticed that most arduino and esp development kits were using linear regulators. I could not figure out why. But your comments on cost and noise have helped answer this, and given me some useful background to help with planning my projects.

Thanks !

One consideration is also the availability of an enable/shutdown pin in the case that the buck/boost converter is to be controlled by an Arduino GPIO pin. This could be important for battery powered applications where the Arduino periodically say switches a boost converter to power a peripheral.

Noise, efficiency, over-load behaviour, price. For sensitive analog electronics noise
is often the overriding concern - for most other cases efficiency or price win.
For some circuits demanding heavy surge currents over-load behaviour is the
show-stopper.

A common pattern you'll see for analog electronics is combining a buck converter
and a linear regulator - the first provides good efficiency and thermal
management, the latter (an ultra-low dropout type) cleans up the buck
converter output.

No, so it will clearly be useless to discuss this further.

The Arduinos using linear regulators are really old designs. Recent ones use switching regulators as they became readily available.

As to the ESPs, they are 3.3 V systems and dropping 5 V to 3.3 V is generally not very efficient in a switching regulator (you need to use synchronous rectifiers instead of diodes) so the advantage over a simple linear regulator wasting the 1.7 V as heat is much less.

BTW cost is often roughly equivalent to the PCB area taken up - often this costs more than the components. A small linear regulator can fit onto a fraction of a square cm, whereas a buck regulator and its inductor(s), caps and schottky diode tends to be > 1cm^2.

This is why some companies sell buck converter chips mounted in a package containing the inductor all in one: https://www.digikey.co.uk/en/product-highlight/t/torex-semi/xcl206-micro-buck-converter?utm_adgroup=General&utm_source=google&utm_medium=cpc&utm_campaign=Dynamic%20Search_EN_Product&utm_term=&productid=&gclid=CjwKCAjwtpGGBhBJEiwAyRZX2uPu5LbiSBMjAaNy25QWQQoI4XjHu70n4eW6XkgTa36TE6J94I0WaRoCdg8QAvD_BwE

But requires a far bigger heatsink.

Entirely dependent on the load - sometimes yes, sometimes not an issue.

One more thing I forgot, buck converters can generate audible whining via the
magnetostrictive effect in the inductor - some (which are commonally ultrasonic)
back off to audible frequencies on low loads - this is sometimes an issue.

Tell me about it! I have a friend who complains about this all the time, most of the time I can't hear what he's complaining about, but then he is 20 years younger than me!