custom/home made gadgets, etc.

[Th3_uN1Qu3]

Okay, i just rigged up a quick and dirty power tester, just to see how it works. I have long wanted to build me one of those, but i don't repair enough computer PSUs to warrant building it. The idea however, has been in my head for a long long while.

So i just used two IRF740 'fets that i had lying around and put them on an ATX PSU heatsink. They have 0.55 ohm RDS(on) each. In parallel that makes a minimum resistance of 0.275 ohms which would be 91 watts at 5 volts and 524 watts at 12 volts. PSU under test was my old trusty 200W Seasonic, rated 20A @ 5v and 8A @ 12v. You do the power math. As expected, no problem at 5 volts for the PSU, however, even with a computer fan blowing onto the heatsink where the 'fets are installed, it quickly got very, very, very hot and i needed to stop testing as to not blow them. This is with just 90W, one could imagine what would happen if those two 'fets would have to dissipate 500W.

Actually their maximum dissipation IF you can keep the case temperature at 25C is 125W each, so it is physically impossible to eat 500W with just two. Also, at 100C case temp (a more practical value), they are rated for only 6.3A continuous current. So, let's do some math. 2x 6.3A = 12.6A. 12x 12.6 = 151.2W. So you could load 150W on the 12v rail with two of them. But there is another little thing that is called power derating... In the case of the IRF740, this is 1W/degree C. So with the case at 100C, you would be left with just 25W dissipation capability in each MOSFET, or 50W total!!!

This is why you will need a lot of them. Assuming your cooling sucks and they will sit at 100C case temp, for a 500W load you are going to need 20 of those. With proper cooling you could get away with 10 or 12.

And of course, there's more... Like any resistor, the MOSFET's internal RDS increases with increasing temperature. If you want a "set it and forget it" solution that doesn't have to be monitored all the time while it's testing the power supply, you'll want a feedback loop that adjusts the drive to the MOSFETs to keep power consumption constant. This sounds like a job for a microcontroller already. And honestly, i don't know a bit about micros, i'd just do something with an opamp. But for a truly automatic tester, where you input the desired loads via a keypad and get results on a nice LCD screen, you will definitely need a micro.