Computer PSU Tester design

I am looking to design a PSU tester for computer power supplies from 100-850 Watt.  The tester will have user selectable power selection switch on each leg of the power supply and voltage and amp meters on each leg.  Can anyone suggest a good design schematic??   mike41351(_at_)aol.com

Do you mean flyback converters as used in desktop PCs. Often the 5V supply has a minimum load current to keep the supply regulated.
If you are only going to test each output individually (bearing in mind the earlier comment) the best way is with an active current sinking load,
They can be set up with a pot to the desired load current, and don't care what the voltage is, providing you meet voltage compliance and power dissipation requirements,
cheers,
Richard

The PSU Tester would employ a multi position load switch for each leg of the PSU and include a volt and amp meters. Loads wound be resistors. Currently nothing is available at technician level cost.

Problem with that is that you will have to fix the current supplied for every voltage or have an extremely large number of resistors. So you wont be able to test each leg at it's maximum current, but at a much lower one that will suit the minimum load current you could expect from each of the voltages, from memory plus and minus 12, 5, and 3.3. Otherwise it would be easy to exceed the max load current on a connector.
With modern graphics cards the 12 volt supply is often expected to supply quite a high current as well.
If I were doing it, I would use a programmable current sink, probably not much more expensive than a lot of fixed resistors, the advantage being that you could dial up the test current as indicated on the PSU name plate, and you wouldn't have to worry about the applied voltage, the meter would read that off for you.
In theory you would need to test every output simultaneously to verify that the supply holds to the name plate specks, but do you just want a go/no go test on each transformer winding and connector,
cheers,
Richard

I agree. I am not looking for a go/nogo tester, but one that would test output currents simultaneously. Can you supply a schematic??

How are you with analogue design, there is no cook book stuff around generally, you need some op amp theory and ohms law (as it applies to electrical and thermal design) and good wiring practice for high power circuits.
Basic principle in enclosed note, you can substitute an NMOS fet, taking into account power supply rail and Vgs. Reverse transistor or fet type (pnp or pmos)for negative supply testing, have to have say +-15V supply rail.
http://www.allaboutcircuits.com/technical-art...r-an-op-amp-output-for-higher-current-part-1/
For the higher voltage supplies the sense resistor could be larger to keep the power dissipation on the pass element lower, you have to check for voltage compliance, the voltage across the sense resistor and pass element has to be less than the voltage being tested, the pass element voltage needs to be checked against the data sheet, must not get near Vsat for transisor, or Vds has to be OK for drain current, keeping the pass device in its active state. And so it goes on. Spice may help a bit, but you need to design the whole thing on paper first to understand what you are doing.
If this all seems excessive, then, a lot of variable power resistors may be the answer, if they all had dials, you could set up a spread sheet according the supplies name plate voltages and currents, the results from the spreadsheet could then be dialed up on your test board.
Again, you are talking a hundred watts on some supplies, so these are heavy duty pots, they are available, but very very expensive. Additionally, the power dissipation for a pot is for the full resistance, so the power rating drops as the resistance drops, something a spread sheet could take into account.
Seems to me the active solution offers a lot of advantages in this respect.
If you are in the business of testing a lot of supplies, then the cost may not be an issue, you will have to look at the payback time.
Without knowing more details, it's hard to know what to suggest,
cheers,
Richard

Hello,
Your knowledge of op amps and transistor circuity is beyond my capability. I am generally familiar with common radio circuitry(tube era), But after encountering several computer power supplies that passed go/nogo testers, but failed in circuit, I surmised the need for a basic load tester for PSUs. I would like to see a switched load bank that would load each leg of the supply at once and have a range of wattages up to a total of 800 watts. A gang switch with contacts rated at 30 Amp and using analog meters with amp and volt readouts for each leg.After reviewing a PSUs label,the loads could be set to 90% of value and then tested and note failures I would like to build something fairly simple and reliable withe minimum part count. There is currently only go/nogo led testers that do not load the power supply. Very much an inconclusive test. Maybe build for sale, since this would be an open market.
Mike Jobe/ 618-993-6340

OK, as I said, the major problem is that you would need so many fixed resistors to cover all the options I think it would be unmanageable.
Consider, a 5V supply could have ratings from 10A, to 50A, so you might need five to ten different resistors. The same would apply to all of the other voltages. Eventually you might end up with >50 resistors to try and cover all the options, and they all have to bolted onto a big heat sink, particularly if you want to soak test the supply. Switches to handle that sort of current don't come cheap either.
The advantage of the electronic load is that you only need four or five, there is no current switching, and they will cope with any voltage, they just take whatever current you dial up from the supply they are connected to.
Realistically the electronic load is the way to go I would think.
The other advantage of electronic loads is they can wind the power supply currents up slowly, and do dynamic testing as well, which would give you more options as a selling point.
You may be able to find someone locally who could do the work for you under contract, I am in New Zealand, and am not doing much contract work these days by choice.
It should be easy to put together, and once a design is done, it's just a question of replicating it. Spice may help you to get a handle on it, if you have had an electronics background. However, spice has very many limitations, and a circuit has to be bread-boarded in the real world in the end, stability is something that spice often handles badly since exact models for the circuit are never available,
cheers,
Richard