DIY PSU load tester

[Quasar]

Re: DIY PSU load tester

I've decided I will build one too. I'll probably take the somewhat more expensive but easier apporach and use resistors. For now, though, I'm deciding on what resistors to use where and what to house it in, as it will need some fairly large heatsinks.

Old PSU cases make great housings and you already have a fan hole.

Wirewounds are inductive by nature and should be used with caution. TRW and Dale offer non-inductive types. Aluminum cased, and they are setup with mounting tabs for chassis/heatsink. I've seen up to 250w each, but you may want to set them up in switchable banks so you can dial-in the load.

Check the surplus electronic supply listings. I know I got 3 pounds these for cheap many years ago in the 25w & 50w varieties. Hey, they were neat looking and gold colored...

Water cooling was also mentioned somewhere here (bucket type). Sounds good, but I'd like to see a circulator of some type to keep an even temp with no hot spot or boiling. Test a PSU and make poached or hard boiled eggs at the same time!

[goodpsusearch]

Re: DIY PSU load tester

My psu load tester is an old abit pentium 3 motherboard that is full of bad caps (but working), 2 bad hdds, some fans and some 10W and 20W 12V halogen lamps

[Th3_uN1Qu3]

Re: DIY PSU load tester

you may want to set them up in switchable banks so you can dial-in the load.

This is precisely why i don't want to be using resistors. The resistors could be driven by a MOSFET in a feedback loop to keep the current constant, however, at anything less than 1/2 load the most dissipation will be in the MOSFET, so you will still need multiple MOSFETs and big heatsinks for them.

Water cooling was also mentioned somewhere here (bucket type).

A bunch of used-to-be minimum load resistors thrown in a bucket of water. That's what i use atm... They do about 200W, but can't be used for more than a few minutes continuously.

and some 10W and 20W 12V halogen lamps

I had two 50W lamps. I lost them.

I've given this more thought today while cannibalizing a bunch of lame looking PSUs, and noticed i have collected about a dozen similar heatsinks and 7 80mm fans of the loud and high CFM variety + 1 92mm. This makes for some serious airflow. So, considering i don't need to test 700W PSUs all that often, i might just get away with using those for cooling.

Each heatsink can hold three parts. The number of MOSFETs for each rail has been chosen so that absolute maximum power dissipation is 35W per MOSFET. Judging by the datasheet, this value of dissipation will be enough to keep them happy even at 100C junction temperature. Thus there will be four on the 3.3v rail for a maximum load of 40A or 132W, six on the 5v rail for 40A or 200W, and ten on each of the two 12v rails for 30A each or 360W each. If you do the math, we actually have a combined power of just over 1 kilowatt! Now you will just need a way to get rid of all this heat. Btw, 30 'fets at 65 cents a pop is $20. Cheap 'nuff.

I'm actually thinking that 40A capability on 3.3v is overkill, and i may move one of the MOSFETs to the 12v rail to increase its capability a bit. But more on that later.

While i have serious doubts that heatsinks from cheap PSUs can dissipate 100W each even with lots of cooling, this remains to be tested. Even if they can only do half that, this still leaves 500W to play with, and 360W of that on 12v. That'll still be able to blow any cheapo PSU.

I have also estabilished a block diagram of the power supply for the metering and feedback circuit. There will be a boost converter using my favorite circuit (a 555 timer and a comparator) which will output 14.5v. This will be powered from the 5vsb rail of all places and i'll let you guess why. Then this voltage will be brought down to 9v with a linear regulator. This should bring both rock solid and clean power to the metering circuit.

I've also given the ripple meter a few more thoughts. Since opamps with useful gain at high bandwidth are expensive, the amplifier will use common small signal NPN transistors. Which you can find, guess where... in a dead PSU of course! The aim of the ripple amplifiers is to have everything use the same 2.5v reference.

[shovenose]

Re: DIY PSU load tester

So is the schematic going to appear please please plsase i dont know how to do thisnotherwise. Lol tyvm!

[Th3_uN1Qu3]

Re: DIY PSU load tester

It's going to appear once i get it done... There will also be PCB layouts.

[370forlife]

Re: DIY PSU load tester

Just a little hint, you can go on onsemi's website and request samples of stuff. I think up to 100 items (up to 10 of each, depending on what it is) you just have to pay for cheap, speedy fedex shipping.

I got a TON of schottky and NPN transistors this way. When they came I only got 9 packages of parts. I noticed my 30A, 60V to-220 isolated schottky's were missing. Oh well, I thought. 1 week later a package arrives all the way from taiwan with those parts in it.

[Th3_uN1Qu3]

Re: DIY PSU load tester

you just have to pay for cheap, speedy fedex shipping.

Which is close to $40 for Romania. Not that cheap. However i did hear something about a local Farnell distributor, so i'll go check it out on Monday. Lemme see what i can find there.

[370forlife]

Re: DIY PSU load tester

Still, I got over $400 in free parts. Even with shipping to Romania thats cheap as hell.

[Th3_uN1Qu3]

Re: DIY PSU load tester

Update. I've earned enough $$$ to allow me to complete this project, especially coupled with the fact that i'll be getting a FREE big heatsink. So, time to finish the design specs and BUILD IT!

I've also found a source of really cheap analog panel meters ($4 each), so i'm good to go. This has me thinking that i should also have analog ripple meters, should be more accurate than just a couple LEDs. Since they're so cheap, i might even drop some voltmeters in while i'm at it, but you do have four meters, don't you?

If you haven't figured out how to make the 0.001 ohm sense resistor, 46.5mm of 18AWG will do. That too, you can get from a power supply. It'll only be dissipating 1W at the maximum current of 30A, so there should be no problem.

Here's the summary so far:

• One TLC272 (single supply dual precision opamp) will be in charge of keeping the load constant. One for each rail.
• Simple class A amplifiers with 2SC945 transistors (that you find in EVERY cheapo PSU) will be amplifying the ripple and driving all the meters.
• A small switcher using the 555 timer IC will boost up the 5vsb supply to 15v, which will then be dropped down to 9v with a linear regulator for best filtering. This 9v supply will drive all the metering and control circuits. The switcher will use single transistor forward configuration, and again, a transformer core from a generic power supply will be used. More specifically the one that does the 5vsb.
• And finally, a single TL431 will serve as voltage reference for everything.

There will also be a thermally controlled PWM fan driver using as sensing element a NTC also recovered from a power supply. I might use the 555 again here, who knows. For now all i can say is that it will be fed from the 15v supply. I plan to run 10 80mm fans. At 0.24A each that's 2.4W, well in spec of the 5vsb rail. The control and metering circuits are expected to use up a few hundred milliwatts at most. It will also have a thermal shutdown function.

[shovenose]

Re: DIY PSU load tester

sounds interesting but complicated.
when i build mine im going to use my coolmax psu tester thing as a voltage readout...that way i dont need t hook up 4 multimeters

[PCBONEZ]

Re: DIY PSU load tester

Those things don't put enough load on a PSU to tell if it's good.
-
They will tell you if it's bad, but not if it's good.
There is more than one way for a PSU to be bad.
.

[shovenose]

Re: DIY PSU load tester

yes i know. i am just going to use it for voltage readouts. and to turn on the pu. but i will have a selectable auxiliary load piece

[shovenose]

Re: DIY PSU load tester

So i looked on the onsemi site (to get some free samples), they dont carry irf530. What would do thr same thing?
Btw i may as well use more expensive components if i can get them free right?

[Agent24]

Re: DIY PSU load tester

This looks very interesting... Will keep an eye on this thread!

[Th3_uN1Qu3]

Re: DIY PSU load tester

Heads up, here comes the first batch of parts. And btw, i found $3 meters. They will obviously need new scales but that isn't a problem. They happen to also be the right sensitivity for my ESR meter and are large enough to read comfortably (though not as large as i'd like, but oh well).

They only had one in stock, which will go into my ESR meter, but i placed an order for eight more. Looks like we will have analog ripple meters after all.

Also i ditched the idea of powering the electronics of the tester from the 5vsb rail, because this would make me unable to meter ripple on 5vsb, which is quite important. I remembered i had a small 12-0-12 mains transformer, it will do nicely if a little high, so more heat to burn for the 7809 regulator, but at the amount of cooling this thing will have, it won't be a problem. Oh yeah and i bought two 7809s. 3A oughta be more than enough for running 8 fans and a few opamps.

[shovenose]

Re: DIY PSU load tester

what sort of fans are you usign? because if you use powerful fans you will go way past 8amps

[shovenose]

Re: DIY PSU load tester

exactly what mosfets are those? i want to order some of those...and how enormous must the heatsinks be.

[Th3_uN1Qu3]

Re: DIY PSU load tester

what sort of fans are you usign? because if you use powerful fans you will go way past 8amps

Standard 80mm fans recovered from PSUs.

exactly what mosfets are those? i want to order some of those...and how enormous must the heatsinks be.

I'm still waiting for that heatsink, my friend is too lazy to look for it in his basement. I'll run a few tests on heatsinks recovered from power supplies, but as they'll have to burn 100W each i doubt they'll be enough.

The MOSFETs are IRF530s like i said.

[Th3_uN1Qu3]

Re: DIY PSU load tester

I've installed three of the MOSFETs on a generic ATX PSU heatsink to test it. As i didn't build any of the electronics yet (gotta go pick up the rest of the meters), i hooked the gates up to my adjustable power supply.

So, can a generic PSU heatsink take 100W of heat without burning the devices attached to it? Yes... if it's sandwiched between two fans. Even so it runs at 100C which is too high to be comfortable. The IRF530s are rated for 175C maximum temperature and 1.7C/W junction to case, which means that with the heatsink at 100C the devices are operating very close to their ratings. I'm just starting to realize the amount of cooling that this thing will need...

And there you go, that's why generic PSUs blow at 250W or so, at 70% efficiency the primary switchers simply get too hot.

[Agent24]

Re: DIY PSU load tester

Yeah, I was thinking about JonnyGuru's PSU tester and from what I've seen it's pretty big, and he always comments about the fan noise.

I agree, you are probably going to need some pretty big heatsinks and a lot of fans...