I'm currently migrating my old and hopefully interesting projects from my blog www.tiny-labs.com to instructables. I use this power supply regularly to power my embedded projects in the lab. It is very handy and tiny. I've used an USB port as power supply for my projects before. Some ICs need 3.3V instead of 5V as Vcc. Therefore I decided to build this small power supply. It uses a range of 7-14V (I use 7V for less power loss) as input voltage and supplies 5V and 3.3V with 1A each. Both supplies can be switched individually.

I hope you can use this power supply for your projects. If you have any questions or suggestions please feel free to ask/post them in the comments or via PM.

Enjoy :)


I got two suggestion by gm280 and rafununu concerning the heat production. They suggest to mount heatsinks to the ICs. They are right, especially for high current consumptions. Consider that the energy, in this case thermal energy, increases proportional to the voltage difference between input and output voltage. A heatsink can be easily attached with a screw or a clip directly on the LM1086. Heat paste would increase the thermal conductivity between the IC and the heatsink. There a lot of differently shaped heatsinks for the TO-220 package (this is how the housing design is called). Just ask your dealer of trust or do a quick google search. I must admit that I don't have overheat problems, but I'm mostly using the 5V line (which leads to less power loss) with a small load.

Rafununu suggests to replace the two diodes with only one Schottky diode as well.

JohnC430 suggests to limit the output current to a reasonable amount, like 600mA. That would make the power supply overheat-proofed.

Another hint by jhsa140467 concerning the heat problem would be placing another LM1086-5 right before the 3.3V circuit to reduce the voltage drop and therefor the heat generation at the LM1086-3.3

Please find all the suggestions in the comments. Thx for your support.

Step 1: Circuit Diagram

The concept is very clear. I want to have a small and handy power supply, which can provide 3.3V and 5V with at least 1A current. The output voltage should be constant even under heavy load and should have less noise.

I use to linear voltage regulator from Texas Instruments, the LM1086-5 and the LM1085-3.3. You can find typical application diagrams in the data sheet. I'll add the data sheet to the attachments. Please don't be confused, this data sheet is valid for all versions of LM1086-xx.

The two regulators are powered with the same source and the both circuits can be switched on/off individually.

This is the complete part list:

  • Power plug, any kind you find and to which you have a wall-pluged power supply to
  • 2 switches, make sure that they can handle your maximal current
  • 2 1N4004 diodes to separate the two circuits
  • LM1086-5 (for the 5V circuit)
  • LM1086-3.3 (for the 3.3V circuit)
  • 2 1k Ohm resistors for the status LEDs
  • 2 small red LEDs for status on/off
  • 4 Electrolytic capacitor (e-caps) of 100uF for smoothing the voltages and handle load switches
  • 4 capacitors of 100nF for smoothing the voltages and handle load switches
  • Some wires :D
  • Some kind of plugs for the output voltages and GND. I use male pin headers, because I use cables in my lab, which have female pin headers soldered to them

You can use different or add more LM1086-xx with another output voltage to this project. But I think 5V and 3.3V is enough for the most embedded projects.

I use eagle cad to plan the circuit and later on the PCB.

Step 2: First Prototype on a Breadboard

Before I go on, I want to test my circuit. I set up everything on a breadboard and measure the voltages at the outputs. It works perfectly. Next step is to plan the PCB layout and to route the wires.

Step 3: Layout and Soldering

Now it's time to put the circuit into a layout for the PCB. You can easily switch in eagle cad to the layout view. It takes some time and experience to place everything in the right position. There are few rules I try to follow. The capacitors should be as close to the pins of the ICs as possible. The whole design should be compact but still easy to solder. I try to separate the two circuits visually and to apply some kind of symmetry. Another important rule is to minimize the crossing over of wires, which means to minimize the number of bridges you will need. Consider the heat, which is produced by your parts, especially by the ICs.

In this case you can see my result in the foto above. I need one wire bridge (the red line). You may achieve a different version, but that is okay. There is no right solution. Just try to apply as many features, I've mentioned before, as possible to the layout.

Next step is to solder everything on a PCB. PCB is not the right word here. PCB means printed circuit board. I don't print anything. I solder it by hand with wires and not traces. But you can send your design to a company, which provides such services. Or you can do it at home. There are some technics to this. Just check instructables.com, I guess you will find some examples.

My board has a width of 35 mm (1.38") and a length of 46 (1.81") mm.

Step 4: Measurements

Congratulations, you are done. The power supply is finished and can be used for your projects.

The last thing to do, but this is completely optional, is to measure the quality of the power supply. There should be no difference, if you use the same design and the same parts. I use a digital oscilloscope to measure the output voltage without and with a load. The third measurement shows to performance during the moment a load is applied to the output.

Step 5: 3.3V Output Without Load

There is no load applied to the 3.3V output port. The voltage has a small noise of +/-10mV. There is no voltage drop. This is more than enough for my purposes.

Step 6: 3.3V Output With a Load of 330mA

In this setup I want to measure the performance of the power supply under a load. The simplest way to apply a load is to use a resistor. I use a big 10 Ohm resistor. Make sure you have one, which can handle some power. I use a resistor which can handle 4 Watts. It's pretty big.

V = R*I --> I = V/R --> I = 3.30V / 10Ohm --> I = 0.33A

I apply a load of 330mA. The noise level is the same +/-10mV. There is a small voltage drop as well of 20mV. This is still enough for my purposes.

Step 7: 3.3V Output With a Load Switch to 330mA

The noise level and the voltage drops are important for a power supply. Another important criteria is the behavior during the switching process. I use a MOSFET to switch between two loads (from small to big). The MOSFET is triggered by a function generator with a square-wave signal. This signal is used to trigger the oscilloscope as well.

The noise level is the same +/-10mV. The voltage drops immediately a bit. There is no resonance or big amplitude.

Step 8: Conclusion

The power supply suites my requirements. The voltage is smooth with less noise. The small voltage drop under heavy load is negligible. The process of switching the loads does not interfere with the performance.

I hope you enjoyed the small instructables :)

<p>For me it looks like another boring DC-DC, which can be found anywhere for $1. What's benefit of _your_ converter? It even doesn't work from wall socket! (read: &quot;requires ONE MORE device AC-DC&quot;)</p><p>Note: LM1086 can take up to 30V input, you confuse people with lower figures.</p><p>Note 2: Even for RaspberryPI-3 it's recommended to have 3A DC. Don't know why, but it's official.</p>
<p>You are right, there are cheap power supplies out there. I've mentioned this before, but this is not the idea about DIY. You can buy everything online. It's fun to build something on your own from scrap parts. But I get your point.</p><p>I've decreased the input voltage to limit the power loss and the heat generation.</p><p>About the the Raspberry Pi, <a href="/member/SebastianC107">SebastianC107</a> asked for 2.5A and I've checked the FAQ from RasberryPi, which recommends a 2.5A power supply for the most hungry model. Please see: <a href="https://www.raspberrypi.org/help/faqs/#powerReqs"> https://www.raspberrypi.org/help/faqs/#powerReqs</a> . I agree with you, that I would use a wall-plugged power supply here.</p>
In my mind if you DIY, then your goal is not only &quot;repeat the wheel&quot;, but make it better/unique. Or cheaper. :) Hope you'll also find something to improve! Good luck!
<p>Hi, </p><p>Thanks for the great work.</p><p>Can I replace the LM1086-x with the LM1084-x to increase the Amps to 5+ amps?</p><p>Or Can you please provide a circuit to do the same? </p>
<p>Thank you. Please see the data sheet of the LM1084: <a href="http://www.ti.com/lit/ds/symlink/lm1084.pdf"> http://www.ti.com/lit/ds/symlink/lm1084.pdf </a> . The LM1084 and the LM1086 are identically constructed, so there shouldn't be any problem. You have to consider to use thicker wiring and check the passive parts for their maximal current ratings (diodes, switches etc.). I would definitely add a heatsink to the LM1084. </p><p>You would need the LM1084-5 and the LM1084-3.3, the adjustable version would need a different circuit. There are typical application circuits in the data sheet. I've found a blog post about the LM1084, which might be worth reading for you: <a href="https://electronics.stackexchange.com/questions/116215/linear-regulator-lm1084-5-0v-capacitors-choice"> https://electronics.stackexchange.com/questions/1...</a></p><p>This circuit is a bit simpler and they are talking about the right capacitors. I hope this is helpful :)</p>
<p>I love it!! I do have one that I built a long time ago that steps the 5 volt USB down to 3.3 volts for those projects, and several power banks that I built from salvaged laptop batteries with an adjustable power supply one from 1.25 volts to 11.6 volts, and another from 12 volts to 24 volts as well as my super pack that runs at a straight 32 volts for 50 and 100 watt LEDs. This, however looks like a great little project to have on my desk for instant relief. Perhaps I will build this and hook up several 18650's for power. I think, if I dig around, I do have one of these that I ordered off Ebay that does exactly the same thing, but it isn't as much fun as building your own.</p>
<p>Thx, I'm glad you like it. I've used old laptop accumulators for several projects as well. They work pretty good with a cheap charging regulator. Are you using your 100W LEDs as a flashlight? That must be very bright :D</p><p>You're right, there are a lot of cheap power supplies available, especially on eBay or aliexpress. I had a lot of fun building my own version :)</p>
<p>Could I Adapt it to provide 2.5 A? I'm trying to build a Raspberry Pi Power supply</p>
<p>Referring to the data sheet, the LM1086 has a typical maximal output current of 2.7A. That means it would be sufficient for your demands. But keep in mind that you might increase the diameters of your wires and use parts, which can handle the current (like the switches, connectors etc.). You would definitely need a heatsink on your ICs.</p>
Thanks for your reply, i will use just 5V circuit and a heatsink.<br><br>Great article
<p>Thx and you are welcome :)</p>
<p>Idea, connect another 5V regulator before the 3.3V regulator. That will decrease the difference of potential on the 3.3V reg, keeping things cooler :-)</p><p>Another alternative if you don't want to increase the parts number, would be to connect the input of the 3.3V reg to the output of the existing 5V regulator.. I prefer the first option though..</p>
<p>This is a good idea. It would distribute the heat generation to more parts. Thx.</p>
It is still a good idea to install some heatsink on the regulators :-)
<p>You are absolutely right :D</p>
<p>I've added the suggestions from the comments to the project description for a better overview. Thx for your remarks.</p>
<p>With 7 VDC max input, the 3.3 V regulator will have about 3 watts at 1 Amp. Generally the TO-220 package is only good for 2 W to 2.5 W. if you limit the output current to about 0.6 Amps then you don't need any heat sinks. I suppose lost loads are lower than that.</p>
<p>Thx, thats a good idea. My projects need barely more than 200mA. A limitation would make this power supply safe.</p>
<p>Can you make for 10V too?</p>
<p>I guess you mean 10V output voltage, right? Of course it's possible. As far as I know there is no fixed 10V version of the LM1086, but you could use the adjustable one, the LM1086-ADJ. You can set the output voltage with a fixed voltage divider or a poti. There is an example circuit and a formula explaining this method within the data sheet. </p><p>There are also a lot of different kind of regulators out there which have a fixed 10V output voltage. For example the UA78xx family, the UA7810 would be the 10V version. Please see the data sheet: <a href="http://www.ti.com/lit/ds/slvs056p/slvs056p.pdf"> http://www.ti.com/lit/ds/slvs056p/slvs056p.pdf </a> . There is an example circuit as well. As far as I can see you have simply to adjust the capacitors.</p><p>I hope this helps. Cheers.</p>
<p>Tnx a lot</p>
<p>You are welcome :)</p>
<p>Looks fantastic, eventhough similar modules are fairly cheap nowadays it is always fun to make something yrself.<br>I do not see the &quot;heatsink issue&quot; as brought forward by some as a big issue. Space enough to screw a piece of aluminum to the back of the regulators.<br>You could indeed do with 1 diode if you put it before the switches. No need for a Schottky diode: the LM linear regulators are not the most efficient anyway so I wldnt bother with a Schottky, unless your input voltage was just above the LM's treshold</p>
<p>Yeah there a lot cheap and already boxed power supply out there. But as you have mentioned, it's always fun to do this on your own. I had the parts laying around and I sampled the ICs from TI directly for free.</p><p>The heatsink issue might be a problem if you have a big voltage difference between input and output, although I've never encountered an overheating issue. I added the advices to my description. A heatsink could be easily mounted and there is enough room for it.</p><p>Linear regulators are not very efficient indeed, but there are easy to setup and cheap :). A switching power supply would be more efficient. I've made once a switching power supply with -5V and 5V to operate operational amplifiers.</p><p>Thanks for your kind words. Cheers.</p>
<p>oh, i fully agree. If you have the parts laying around then that is cheaper than anything else.<br>I use linear regulators a lot, indeed, very easy</p>
<p>Not a PSU but a regulator. Heatsinks are necessary, specially for the 3.3V which must absorb at least 3.5W ! The 2 x 4004 (to avoid inversion I presume) should be replaced with only one shottky in the general supply feed line, shottky threshold is minimal. </p>
<p>Thanks for your suggestions. I've added your remarks to this projects. Cheers.</p>
<p>Looks nice and small. But without any heat sinks, I don't see it holding up for long periods of heavy current needs. Otherwise good project. You should design a PC board and have a few made with heat sinks incorporated. Just a suggestion. </p>
<p>Thanks for your suggestion and compliment. I've added your advice to my description. A PCB would be great. Unfortunately or fortunately (depends how you see it ;)) there are a lot of cheap power supplies out there. I did some PCBs before for other projects and let them produced in a low volume. That was so much expensive. But you are right, it's a great feeling to got your own PCB done :D</p>

About This Instructable




Bio: Born in Berlin in 1985, engineer, contrarian, 'The Big Bang Theory' fan, my blog: www.geek-end.de
More by Stish:DIY Power Supply for 5V and 3.3V Rotating Bike Lights Fly-In Diner - Wooden Bird Feeder 
Add instructable to: