Every electronic or electric project needs power, but how to select what kind of power supply you project needs.

There are so many types out there, what are the things to keep in mind when browsing your local store or digging in a box of recycled parts?

In this instructable I hope to give you a clear idea of what the differences and caveats are.

Note: I only cover mains powered projects here.

## Step 1: Specifications

First you need to find out what specifications you are looking for, and they are all interlinked.

• Output voltage(s)/power
• Cost
• Input voltage
• Efficiency
• Ripple/noise
• Size
• Durability

You can spend days looking trough all sorts of folders, catalogues, websites and so on for your ideal power supply.
Most likely: it does not exist and you will have to settle for less.

## Step 2: Output Voltage/power

For most people this is the only consideration when selecting a power supply.
Of course this is the most important, your project needs the voltages and power, you can't change that.

You can however change the way you supply it.

Example:
If you need 5V for an Arduino and 12V for your motors, a buck converter would be a good idea instead of a PC power supply or 2 wall warts.

## Step 3: Cost

Cost seems simple, your junkbin is free and the store has simple price labels.

But it is not so clear cut, efficiency plays a big role.

Example:
We need 5 Volt 0.5 Ampere for an Arduino project with a screen that is going to be on permanently.
In your junkbin you find an old 12V wall wart, your Arduino accepts 7-12V on the input so you are done.

These old wall warts are usually only about 45% efficient and have a high consumption.
The Arduino uses an linear power supply to bring the 12V down to 5V.

You use 5V 0.5A -> 5*0.5 = 2.5 Watt.
The power supply in the Arduino converts 12V into 5V by turning the 7V difference into heat.
This means that at 12V it is still 0.5A, so 2.5W becomes 6W.
The wall wart is 45% efficient, so 6W output means about 13W input (45% of 13W is 6W).

So of the 13W that is used from the mains only 2.5W is actually used.
10.5W wasted in heat, over the course of a year is about 92kWh wasted in heat.
The rough average kWh price in Europe is about 20cent, in the USA about 12cent. (dollars or euro's don't really matter in these estimates)
That means you are wasting 11-18 dollar/euro per year on the wasted power in your free power supply.

A simple USB charger is about 70% efficient and outputs 5v directly which reduces the wasted power from 10.5W to 1.1W.
This is only 8 kWh /year, or a saving of more that 10 euro/dollar per year, plenty to buy a USB power supply.

## Step 4: Input Voltage

In mains powered projects the input is normally mains power, but mains is something different around the world.

The mains power supply around the world varies from 110V to 250V and everything in between and lots of 'mains' powered project are actually powered by a output of some other equipment such as the USB port of a server.

In general, power supplies with greater possible input ranges are less efficient.
Power supplies with a switch (such as the picture) don't have that problem.

Higher input voltages also lower efficiency (too bad I'm in a 240V country).

Newer power supplies get better every generation. This is mostly because the internal frequency in power supplies gets higher and higher which makes power transfer more efficient.

## Step 5: Efficiency

Efficiency is the amount of power that goes into your power supply compared to the amount of power that comes out.

Here are some efficiency figures for power supplies you are likely to find.

• Old heavy wall wart 40-50%
• Modern switchmode wall wart (e.g. usb chargers): 70-80%
• Computer power supply: 60-90% mostly dependant on age: older=worse.
• Normal transformer: 60%
• Toroid transformer: 70%
• Open-frame supply: 80-90%

Of course these are rough estimates, in general these rules apply:

• The newer the design, the better the efficiency.
• More powerful power supplies are more efficient (For example the transformers used by energy companies are up to 98% efficient even though they are normal transformers)

The efficiency also varies wildly over the amount of power consumed.
You will normally find the highest efficiency with the power supply at about 75% load.

Don't be temped to use a high power supply at lower loads, efficiency drops like a brick at low loads.

## Step 6: Ripple/noise

In digital designs it usually doesn't really matter.

Ripple: How flat is the output voltage under load.
Ripple always has the frequency of the internal frequency of the power supply.
Transformers will always be at the mains frequency (50/60Hz).

Switch mode (newer adapters, computer supplies, open frame etc) will usually have a frequency in the 5-150kHz range, which often changes if the power supply is under load.
This can be a show-stopper for a audio system.

Noise: How clean is the output signal.
This is mostly based on the quality of the components used in the power supply and the operation principle of the power supply.

Linear power supplies have very low noise because there are no switching components in the power supply, this makes the design automatically absorb the noise that is in the system.
Switching power supplies effectively work by transforming small amounts of energy to different voltage levels, although this is very efficient, every switch action gives a lot of noise.

Ripple and noise can be filtered, using low-pass filters.

If efficiency is important but ripple/noise are also important (for example in an audio amp) you could look at both linear and switching at the same time.

For example if you need 12V, use a 15V switching power supply and connect a linear supply to go from 15V to 12V.
That should greatly improve the noise levels.

## Step 7: Size

The size of a power supply depends mostly on output power and efficiency.

If the power is greater you need a bigger power supply.

If the efficiency is greater the power supply will be smaller for the same output power.

Power supplies are always limited by the size of the transformer, but the higher the frequency the smaller and lighter the transformer can be.
The smaller the parts are, the more power fits into the same box.
A 1000W transformer is about 20kg, and the size of a small table.
A 1000W modern PC power supply is less than 5kg and easily fits in a shoebox.

## Step 8: Durability

How durable/reliable should your project be?

The simpler the power supply the more reliable it becomes.
The more industrial it is the more reliable it becomes.
The hotter a power supply becomes the faster it breaks.

Sorted by reliability:

1. A bare transformer, almost indestructible.
2. Old wall wart, simple in design, not much to break.
3. Open-frame power supply, usually build to last decades continuously running.
4. Modern wall wart, low power, low heat. Just don't pick a cheap one.
5. Laptop adaptor, completely solid, but get hot.
6. Computer power supplies, often don't like strange loads and cheap fans.

The single component that is usually the first to break is the electrolytic capacitors in the filtering.
These work on a water-based liquid electrolyte inside of the cans, which leaks out over time.
The hotter they get, the quicker they fail.

## Step 9: Summary

So many things to look out for and I've only scratched the surface.

A short summary.

• When powering something continuously, pay close attention to the efficiency, or it could get expensive.
• A big old power supply is usually not a good choice unless you are looking for high reliability.
• When doing analog, audio or otherwise critical filtering is important, but selecting a good power supply helps more that you would think.

Just use your favorite search engine and look for terms such as:

• Linear power supply
<p>If u are talkin bout pc , psu....I use 20+ years of pc building, repair, tech support (no I refused to go to India when the company sold out hehe)...anyhoo just go to Corsair website and buy from them. AFAI'mConcerned they make one of the best in all areas of build quality(they look purty on the inside too) for a reasonable price(think PcPower and Cooling brand). I'm using one I got in 2006 when pfc,etc was first being implemented to us &quot;enthusiasts&quot; it's specs have not deteriorated at all (yeah I test stuff cause I'm old and bored) according to limited analyzing tools. I'm an avid to a fault when it comes to oc'ing my pc so the psu has taken some blows but not an iota of change in the single-line 12v power, still reads 41.374 amps (not a fan of multiple main bus lines)....StormRosson</p>
Just a little correction - mains voltage worldwide starts at 100V, in Japan. I know this from having to help my friend find hair straighteners to take on holiday :)
<p>Request: Using on-hand power supplies in series and parallel to achieve higher voltage and amperage :-)</p>
<p>Placing regulated power supplies in parallel is problematic. The regulators often fight with each other to regulate the voltage.</p>
nice 'ible. thanks
nice article thanks for putting the effort in to help describe to us
<p>You seem to be ignoring Ohm's Law in your article. Power equals Volts times Amps. This means that when you increase Volts Amps drops for the same Watt output. So I do not know how you have arrived at your statement, &quot;This means that at 12V it is still 0.5A&quot;? Not when you are coming from 12V to 5V it isn't. A half an amp draw at 5V is two tenths of an Amp at 12V for the same power output. If this wasn't true then our electrical grid would be in big trouble.</p><p>Voltage is pressure, Amps are volume, and Watts is work. So an increase in pressure allows for a corresponding drop in volume to do the same work. Each little drop of electricity is just pushing harder. If that helps you to see it.</p><p>Watt calculations for alternating current are not like DC calculations either. There is something called Power Factor that you have to use in your equation. It has to do with the slopes, and troughs of the sine wave that AC is made of. The simple truth is that Alternating Current is not Direct Current. So Ohm's Law does not work the same there. At least not when it comes to Watts.</p><p>Noise definitely matters in digital electronics too.</p>
Sorry I missed your reply earlier.<br>The point is that most simple electronics use linear regulators.<br>If you have a device that uses 5V 0.5A and use a linear regulator to get that from 12V it will still draw 0.5A (+ a bit more for the regulator) at 12V.<br>The difference (7V 0.5A or 3.5W) is turned into heat in the regulator.<br><br>If the regulator would be a switching one than you would be right, 2.5W at 12V would be roughly 0.21A + the wasted power in the regulator.<br><br>I know very well how AC calculations work, my previous job was designing smart electricity meters, which is why I did not include them, there is a lot more to it than would be appropriate in this instructable.<br>I don't think I'll do an instructable on that, it is more math and theory than practice, wikipedia is more suitable for that.
<p>you are very very right.<br>interesting ibble. Just one thing, many 5 Volt USB chargers are sadly unsafe as they work with a capacitor and a resistor and have bad construction. One cannot see it from the outside, so.. know what you are buying</p>
<p>I'm not sure how you would make a switching power supply without capacitors and resistors.</p><p>There are a lot of unsafe ones out there, but that is usually because of isolation of the transformer/board that is too low.</p><p>If you are talking about wattless droppers, never seen those as USB charger and will probably never be build for a lot of reasons.</p>
<p>true but i refer to the real simple ones that only have 1 or 2 diodes a resistor and a capacitor and often without a transformer</p>
<p>Do you have a link to an example?<br>In my option it is impossible to make a USB charger from mains without magnetics (i.e coil), let alone a safe one.<br>If there is no transformer the USB is connected directly to the mains (i.e. you would be electrocuted by touching your phone).<br>It think even the worst manufacturers would not make something that bad.</p>
<p>Well, as you say: let alone a safe one. and that was exactly my point :-)<br>circuits like these: <a href="http://jeelabs.org/wp-content/uploads/2011/10/JCs-Doodles-page-201.png" rel="nofollow"> http://jeelabs.org/wp-content/uploads/2011/10/JCs...</a><br><br>read an article once from a guy who opened up a few of these. scary</p>
<p>I reverse looked up that image, and it is part of a series he did about making capacitive power supplies: http://jeelabs.org/2011/11/18/zeners/</p><p>I have designed products with these in them and they are really nice, but not suitable for USB chargers for 2 main reasons:</p><p>- Getting more than a few mA out of them you need a big&amp;expensive X2 capacitor, more expensive than a crap quality cheap transformer.<br>- The electocution thing.</p><p>I've seen some wildly unsafe designs, but all of them have at least 'some' mains isolation.</p>
<p>Let's get back to my original remark as we seem to go back and forth wether there are unsafe chargers or not. A simple checkof the news learns that people do get killed and that e.g. in UK unsafe chargers regardless of their design get impounded <a href="https://mrbloggyguarddog.wordpress.com/2013/08/14/unsafe-chargers-in-local-markets/" rel="nofollow"> https://mrbloggyguarddog.wordpress.com/2013/08/14...</a></p><p>Know what you buy is all i wanted to say.<br>Yet that all doesnt change you have written a great ibble</p>
<p>pfred2, you are wrong on this one: the current drawn by the circuit is 0.5A (at 5V). If you supply it 12V, the voltage regulator on the circuit will drop the voltage by 7V (it is a linear voltageregulator, so the extra power is dissipated as heat). When looking at the efficiency calculation, the power supply is providing 12V at 0.5A (6W), the &quot;useful&quot; power going to the circuit is 5V at 0.5A(2.5W) and the wasted power is 7V at 0.5A(3.5W). Hence the recommendation to get a &quot;real&quot; 5V output.</p>
<p>What is a &quot;real&quot; 5V output? Are you somehow suggesting that linear regulators are not real?</p>
Indeed, also thanks for replying, I missed pfred2's comment.<br><br>It is meant to give a real world example that is likely to happen but has 'bad efficiency' written all over it.
<p>somehow that link doesnt seem to work. maybe this one does: </p><p>http://jeelabs.org/wp-content/uploads/2011/10/JCs-Doodles-page-201.png</p>
Nice intro! I would point out two small things : First, old wall warts and cheap transfo rarely give the voltage they are rated for. I have some in my junk box that are rated 12v and give 14v even under load. Enough to dramatically increase wasted energy or to fry a linear transformer. Second, a 2500w transformer can be 10x10x10 cm (and one is included in each microwave) as long as it is not used for a long period. So duty cycle matters in the final choice.<br><br>Obviously, the usual warnings : do not modify any power supply unless you really know what (or watt hahaha) you are doing, unless you are good at dealing with electrical fires, pyrotechnics and auto-CPR.<br><br>Cheers
<p>Old (unregulated) power supplies giving a higher voltage than marked is indeed often the case, espessially in the Netherlands (and probabaly other countries) since the mains voltage was increased from 220 to 230V in 1985.</p><p>Modifying power supplies is indeed a bad idea if you don't fully understand what you are doing, especially with the heaps of underdesigned and unprotected supplies currently dumped for almost free on the world market.</p>
<p>I have had 12 Volt Wallwarts that were indeed 14 volts... but dropped drastically to 9 Volts with the smallest load :-)<br>yes, in the Netherlands :-)<br>Nowadays when I need 12 Volts in an otherwise 5 Volt circuit, I use a small converter that costs next to nothing at Aliexpress</p>
Hi<br>I have a hcl winbee thinclient, it shutdown itself after a few minute.due to over heat. What should i do??
Totally not related to power, but step 1 would be removing dust and checking if the fans work.<br>Step 2 is checking if the heatsinks are mounted properly.<br>Step 3 would be install additional fans.
Cool!<br>Nice guide. :D
<p>I have a pulled apart radio with ac adaptor will wiring a light globe work</p>
Without additional information I would say it will not work.<br><br>There are literally millions of transformer types out there and I can't tell which one you have, same goes for the light globe.<br><br>The chances of 2 random components matching up is small.
<p>Useful and interesting reading. I especially liked the part where you show how efficiency translates into cash!</p>
<p>Great basic primer for Power Supplies, now if you can help me with a PMS I would be greatful.</p>
Motrin?
<p>LOL</p>
<p>I'm not sure what the abbeiviation 'PMS' stands for, can you explain?</p>
I have an old time AT - not ATX - power supply that is also short circuit and overload protected. To top it all it has a 2 pole power on / off switch. <br> <br>On the other hand here in India the mains voltage can be up to 20 volts above 235V it is supposed to be. To compensate for this I had a special 245 -&gt; 20 volt 20 A isolation transformer made. Connected it in auto transformer mode. Now it simply drops app 20 volts from the mains. Believe me my energy bill has dropped drastically due to zero wasted power.
<p>Great introduction, indeed. I feel like I learned a lot all over again. Some very straight-to-the-point pointers.</p>
<p>those older wall warts powered devices that were not so finicky with power needs or had some internal power regulation inside. . they are often just 2 diodes and 2 small capacitors + transformer.</p><p>I do not know why I never thought of doing this level of analysis before thanks for the pointers!</p><p>uncle frogy </p>
I think computer power supplies operate at a higher frequency than 20khz, the quality of the switching power supply is also a good determinant of the efficiency.
Sorry, that was a typo, I meant 150kHz there.<br><br>The quality is not always a good way to determine effciency.<br>Power supplies without PFC and input filtering are more efficient, but worse quality.<br>An underdesigned/smaller transformer will also boost efficiency but will fail under full load.<br><br>A lot of poor design decisions can boost efficiency but make the quality worse.<br>Usually a poor design will also be poor ont the selection of other parts such as rectifiers, transformer core material, etc making the efficiency worse than a good quality power supply, but there are exceptions.<br><br>A good way to asses quality is weight, but cheap manufacturers have picked that up as well, adding scrap as weight into power supplies.<br>Personally I can't think of a good way to determine quality that does not include testing.<br>Counterfeits can be found everywhere nowdays and there are plenty of quality no-name brands out there.