I want to build my own switching power supply. I already know how to make a 10 Ampere Linear supply, and I'm wondering if I should bother. What do I have to learn to do a switching supply? What makes a switching supply better if they both end up giving me DC?

What I don't get is the "inefficiency" argument. Maybe linear supplies get hot, yes, but so does every laptop switching power supply I have met. Looking at a schematic of a switching power supply shows that it has at least 3 times more components; that means 3 times more work and cost to build a power supply. Why would I feed a circuit using an expensive switching power supply that gets hot and that ends up being more expensive than a linear one?

Don't both just end up giving me plain regulated and filtered DC power? I should be able to use either for every application shouldn't I?

Also, if i wanted to make 10A one, how or which component can manage 10 Amps in a switching supply? (Darlington arrays are used in linears)

  • \$\begingroup\$ If your device runs on batteries, any improvement in power supply efficiency is an improvement in how long the device can run between charges. \$\endgroup\$
    – The Photon
    Oct 20 '14 at 4:47
  • 2
    \$\begingroup\$ At some point, "It just gets hot" becomes your main problem. \$\endgroup\$
    – W5VO
    Oct 20 '14 at 5:00
  • 3
    \$\begingroup\$ Let me put it this way. Assume you want to drop 12VDC to 5VDC @ 10 amps. The switcher, assuming 90% efficiency, will dissipate 7 watts ((12-7)*10*0.1). The equivalent linear regulator will dissipate 70 watts! ((12-5)*10). The cost of the extra components for the switcher will be less than that of a cooler for the linear, not to mention the cost of the wasted energy. \$\endgroup\$
    – DoxyLover
    Oct 20 '14 at 5:17
  • \$\begingroup\$ The benefits of a switcher come in the bigger the difference in voltages, and the higher the current needed. At 10 Amps, unless we are talking 1V difference, a Linear supply becomes unfeasible. \$\endgroup\$
    – Passerby
    Oct 20 '14 at 5:36
  • \$\begingroup\$ at 10A output current u can leave LDO/linear regulators......very few available......1. media.digikey.com/pdf/Data%20Sheets/Microsemi-IPG%20pdf/… 2. media.digikey.com/pdf/Data%20Sheets/Sharp%20PDFs/PQ7DV10.pdf \$\endgroup\$
    – user19579
    Oct 20 '14 at 5:47

The answer to which one you use depends on the application, and the efficiency needs.

For example, you're asked to make a phone charging dock. The dock is powered via a 12 V wallwart, and powers the phone with 5V of power at 500mA. Using a linear regulator, 3.5W is dissipated.That's quite a bit of waste, but you're connected to the mains, and a charging dock is a big enough device, where a properly heat sunk regulator wouldn't cause a lot of heating issues.

On the flip side, suppose you're building a wearable device that operates on a small Li-Po battery, even if you designed a LDO circuit that only wastes about 1W of power, a switching circuit would be more desirable as if designed properly, you could reduce your wastage to <10% that of the linear regulator

Note: Pay attention to the efficiency curves of switching regulators. They normally only have high efficiency for small ranges of current usage, and it helps to understand what current usage your application operates on in different condition to design the most efficient power circuit. Also - laying out swtiching regulators on a PCB can be hit/miss - I've seen a lot of incidents where tiny layout issues can mess with the desired voltage out.


As DoxyLover pointed out, it's not just a matter of "getting hot". The efficiency of a linear regulator is Vout / Vin, which is really bad when there's a large difference between input and output. Consider a modern desktop CPU running at 0.9V for an extreme example.

Another advantage of switching regulators is that they can boost or invert the input voltage. If you need a positive and negative voltage from a single battery, or 12V from a 1.2V solar panel, a linear regulator won't work at all.


Linear supplies can be about as efficient as switching supplies--an on rare occasions even more efficient--in cases where the input voltage will always be slightly above the required output voltage. Unfortunately, if the input voltage is only slightly above the required voltage, then a small dip in the input voltage will leave a supply unable to maintain the required output voltage, and a small increase in the input voltage will cause a huge relative increase in the amount of power a linear supply will have to dissipate.

The advantage of a switching power supply is that it will be able to offer good performance over a wide range of input voltages. Even though, as darudude notes, many supplies have a somewhat narrow range of conditions under which they will achieve optimal efficiency, in most cases such limits stem from the fact that many supplies have a certain minimum amount of power that they will draw whether or not the load requires it, as well as a certain amount of power that they will consume beyond what the load takes.

If a 12V to 5V converter is 90% efficient when supplying one amp, but draws a minimum of 1uA from the source, then it efficiency when driving a 10nA load would be pretty pathetic (less than 1%) but the amount of power it would waste in that situation would be only 12uW--far less than the losses when supplying a full amps (where it would waste about 0.56W). If a battery-powered device will need to supply a load that consumes a full amp for one 1 seconds each week and otherwise draws 10nA, the average current draw would be about 1.7uA, of which 1.0uA would be a result of the baseline current draw, making the overall efficiency about 40%. If, however, the load consumed a full amp for ten seconds each week, the average current draw would be about 8uA and efficiency would improve to be about 80%. With the one second-per-week load, reducing the current drawn during the idle times might potentially double battery life. With the ten-second-per-week load, however, battery life would be limited by the need to supply real current to the load even if the idle power consumption could be reduced to nothing.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.