Switch mode regulators
There are different types of DC-DC regulators. The most common are linear and switch mode regulators. Everyone that has tinkered with electronics has probably used a 78xx series linear regulator. They are easy to use since they don’t need any special components. With most linear regulators you just put a capacitor on the input and one on the output and they will work. Always check the datasheet to see which type and capacity is recommended. Unfortunately linear regulators are very inefficient, because of that they can produce lots of heat. For example if you feed a 7805 (5V linear regulator) 12V and you draw 1A from it, the regulator will need to dissipate 7W of heat. So that 7W goes to waste, not ideal for a battery powered or energy efficient device.
About 2 years ago I started playing with electronics and was browsing Tayda electronics voltage regulator section for a regulator that could output more then 1A. I found the LM2576 5V 3A regulator and ordered some. I didn’t know what I had bought until I read the datasheet.
The LM2576 is a switch mode regulator. It’s a pretty old design, that’s why it’s so cheap ($0,76 @ Tayda). I’m not going to explain how they work, there is wikipedia for that.
Switch mode regulators are more efficient then linear regulators. They can not only lower the voltage (buck converter) some of them can also up the voltage (boost converter) or they can do both (buck-boost converter). Most switch mode regulators need some more components then linear regulators to make them work.
For the LM2576/LM2596 you need 2 capacitors (input/output), a big inductor and a schottky diode. You also need to take into account that you can’t just place the different components like you want on a PCB. The datasheets of these regulators will have a reference design to show you how to place them to get the best results. Not following these guidelines or using components with not the right specifications and you are risking that they just don’t work or produce lots of noise. I have designed a couple of PCB’s using the LM2576 and LM2596 and got them to work, so it’s not that hard 🙂
If you don’t want to design your own PCB’s you can buy small boards with a switch mode regulator on Ebay or from Chinese webshops. So today I decided to test some of the boards I have bought from China.
LM2596
The most popular board is one with an LM2596 adjustable regulator. The LM2596 is a 3A 150kHz stepdown regulator, it’s a pretty old design. More modern switch mode regulators switch at +1MHz.
LM2596 board
You can buy these boards for 1,1€ from Chinese sellers on Ebay. I have a dozen of these boards, some have blue PCB’s, some green others red. They are not exactly the same, some have different capacitors. The ones I have with the blue PCB have capacitors rated at 35V, the ones with the green PCB have a 50V capacitor on the input. The boards have a small potentiometer to set the output voltage.
I tested the boards with an input voltage of 8V, 10V and 12V and with a 2A load. I used a Re:load V2 as a constant current dummy load.
@ 8V
output 2A @ 5V
input current 1.65A = efficiency of 75,8%
@ 10V
output 2A @ 5V
input current 1,32A = efficiency of 75,6%
@ 12V
output 2A @ 5V
input current 1,09A = efficiency of 76,5%
These results don’t say everything… With a constant load of 2A the board became to hot to touch. The small PCB can not dissipate enough heat. After about 15 minutes the regulator started to oscillate and the input current started to rise. I would not recommend using these boards with a constant load of more then 1A. I even destroyed one board when trying to draw 3A from it.
LM2596-oscillating
I have used LM2596 regulators on my self designed boards, the 5V version in a TO220-5 package. Using a good heatsink and Nichicon capacitors I get an efficiency of 80% and have no trouble with heat.
MP1584
Another 3A adjustable switch mode regulator that I have bought is the MP1584. The MP1584 is a 1.5MHz step down regulator. I bought the regulator on a small board from Electrodragon.
MP1584 3A switchmode voltage regulator
@ 8V
output 2A @ 5V
input current 1.48A = efficiency of 84,5%
@ 10V
output 2A @ 5V
input current 1,08A = efficiency of 92,6%
@ 12V
output 2A @ 5V
input current 0,9A = efficiency of 92,6%
After drawing a constant load of 2A for more then 10 minutes the board became very hot. But the output voltage stayed stable with a peak to peak voltage of only 160mV. Pretty good regulator / board.
KIM-055L
Next regulator I tested is the KIM-055L board I bought on Ebay. This is a 5V 5A switch mode regulator. The board I have has a big heatsink.
KIM-055L voltage regulator
With an input voltage of 8V the board can not output 5V. So I only tested it at 10 and 12V.
@ 10V
output 2A @ 5V
input current 1,02A = efficiency of 98%
@ 12V
output 2A @ 5V
input current 0,85A = efficiency of 98%
Even after drawing more then 2A from a longer period of time the heatsink became barely warm. The output voltage stayed stable, peak to peak voltage was 120mV.
Hi, thanks for this interesting benchmark. I will take into account the results when I buy next regulators.
I was thinking on buying some cheap LM2596 to get 5V@1A but after reading your article they seem to have a poor efficiency.
[…] prezitna un test facut asupra surselor reglabile in comutatie. Sursele sunt alimentate la tensiuni diferite si pentru un rezultat mentine aceasi tensiune si […]
Good data, thanks for sharing. I may need to affix heat sinks to the regulators on order.
Thanks for sharing this valuable data. You saved me significant testing time. The MP1584 seems to be the efficiency and size winner as a BEC regulator in RC planes and Multirotors.
Two of them could be paralleled for twice the current output with “OR-ing” diodes and the output voltage can be trimmed to compensate for the output diodes voltage drop.
If anybody wants to know how to make those LM2596 boards stable, you can do it simply by replacing the output cap with the low ESR ones recommended in the datasheet. (Different caps are recommended for different output voltages): http://www.onsemi.com/pub_link/Collateral/LM2596-D.PDF
Is there a way to control the output current on the MP1584. So that it will act as a CC-CV regulator?
Not sure if that is possible with the MP1584, the datasheet doesn’t show an example circuit for such a setup.
I see that your “LM2596” board is running at ~50khz which suggests it’s actually a LM2576 or something similar.
[…] slight spike occurring at 51kHz. I didn’t think anything about it at first, but I came across this comment, which leads me to believe that I have a LM2576 (which has an internal oscillator at ~52kHz) as […]
Hi,
I used the Kim055 and with 2A load the output voltage dropped to 4,4V, which is too low to charge eg a smartphone. Did anybody else have this problem too? And what to to about it?
That’s strange, I’ve tested mine with a constant load of 3A and the voltage stayed stable.
Great writeup, thank you! Some additional testing I’ve done on the MP1584 module to see load regulation and temperature. My goal is to regulate from old laptop power supplies 16 to 19VDC down to 5VDC. Adjusted the board potentiometer to 5V out @ 1A load, ran at each load for 15 minutes.
Load Vout Temp
0A 5.28V 23C
1A 5.05V 63C
2A 4.84V 110C
My conclusion: with that much voltage drop (19v to 5v), much above 1A will cook the components. Could try adding a heat sink, but that kind of ruins the purpose of the small form factor. Did not repeat the temp tests with a 12V supply, but a quick check showed nearly the same output voltage at 12Vin in as 19Vin with no adjustment to the board. This makes it nice to set once an use with different power bricks.
Thank you for being informative and helpful!!