I TAKE ABSOLUTELY NO RESPONSIBILITY FOR ANY INJURY, DAMAGES OR UNLAWFUL CONDUCT CAUSED BY FOLLOWING THE BELOW INSTRUCTIONS. THOSE INSTRUCTIONS ARE PROVIDED *AS IS* IN THE HOPES THAT THEY WILL PROVE USEFUL, BUT WITH ABSOLUTELY NO GUARANTY OF FUNCTIONALITY OR FITNESS FOR A PARTICULAR PURPOSE. IT WORKED FOR ME, I HOPE IT WORKS FOR YOU. BUT YOU DON'T GET A WARRANTY, AND YOU CAN'T SUE ME IF SOMETHING GOES WRONG. IF YOU DO NOT AGREE TO EVERY PART OF THIS, STOP READING NOW.
One of the most important things to test when diagnosing any power supply problem is load regulation. I can't count the number of computer supply's that I have come in contact with that test 100% fine when out of the computer, then fail miserably when actually being used. Now... it's simple enough to throw a suspect power supply in a device then probe it's voltages with the device running, but it's not very safe or professional. Also, what if the supply only fails when very near it's rated max power? How can you be sure that your device is drawing enough load needed to trigger the failure?
All of these problems can be solved with an adjustable dummy load. Here is the schematic for the most basic version...
How it works
The above circuit wastes Watts in the N-FETs resistance in the form of heat. It does this by creating just enough resistance in the N-FET's for a very specific amount of current to flow. The amount of current is measured by the 100m Ohm current sense resistor and is sent to the Op-Amp for feedback. The Op-Amp is set up in a "negative feedback" arrangement with the N-FET. This makes it so any inconstant performance in a single FET will be corrected for, ensuring that all of the FET's will share the current evenly. This allows us to connect all the FET's in parallel and control them with just a single control voltage. The control works through the magics of the OP-Amp arrangement, which will attempt to create the same voltage on the (-) input that is sees on it's (+) by adjusting the FET dynamically via the Op-Amps output.
The Op-Amps themselves are active circuits containing many transistors. Because of this, they need to be powered. It is not shown in the schematic, but the Op-Amps will also need decoupling capacitors of about 1uF or more connected to the power pins, and as close to the chips as possible. Adding these prevents the dummy load circuit from fluctuating out of control. Note that if you try to power this circuit from the supply you are testing, the Op-Amps will not get enough power during start up, causing the system to go into a dead lock up. For this reason, you must power this circuit with a separate power supply. This also has the added benefit that if the voltage of the supply we are testing happens to fluctuate, it won't change our current drain setting along with it.
With the exact circuit above and a 12V power supply for testing, the current drain per each FET with the knob turned to it's max position will be about 5 Amps. The Watts will be about 50 per FET also. Note that this can be changed higher by using a smaller value for RA, a larger value pot for RB, or using a higher voltage for the circuit supply. Make sure you don't exceed the SOA of the FET or it will get destroyed, regardless of how well you are cooling it. I recommend you stick with the schematic values unless you know exactly what you are doing. I also recommend you use a regulated supply to power the circuit. This prevents many problems that I'm not going to get into here. Google "Three terminal voltage regulator". Again, do NOT power it off of the supply you are trying to test, it will not work.
The above circuit wastes power in the N-FET's. This makes them produce large amounts of heat. If you do not properly remove this heat somehow, you *WILL*destroy your N-FETs very quickly. For a small (around 200 watt) dummy load, you can simply heat sink the N-FETs like you would any transistor with a properly sized heat sink and fan. However, above a certain power level it becomes impractical to do this alone. If you are planing such a large dummy load, then it is advisable to solder the backs of the N-FET's to a copper pipe, then run water through it. Here is an example of just such a construction...
And here it is when being used.
As junky and cobbled together as the above prototype is, such configuration still works pretty good for just recirculating a single bucket of water. It remains portable and can easily scale up to 1000 Watts by it's self. If you need more Watts, you will have to attach the pipe to a water faucet and pump cold water straight from the well into one end, and have the other end run into a drain. And you may need a heat spreader (metal scour pads) inside the pipe to give the water a better thermal interface with the pipe. This final cooling configuration has practically no wattage limit. It could easily scale up to 10k Watts with domestic utility water.
Anything more is absolutely absurd for DIY, you shouldn't be considering it.
Conclusion and credits
This circuit as is shown above is pretty powerful for loading up most power supply's. And it can be taken from it's base form and made in to some complex load devices if one desires. The original thread that spawned this circuit can be viewed *HERE* there is more information and details there than there is here. You will have to do the leg work on sorting it all out.
I would like to personally thank...
jocanon, For bugging me in the first place, which enticed me into doing stuff.
dougy83, for suggesting that the dummy load be linear, which helped me "out of the box".
ronv, for doing a massive amount of personal work with this, as well as sending me a load of free parts.
Extensible dummy load. 50 plus Watts per transistor.
I TAKE ABSOLUTELY NO RESPONSIBILITY FOR ANY INJURY, DAMAGES OR UNLAWFUL CONDUCT CAUSED BY FOLLOWING