PSU test

1. Jan 5, 2016

totota

Hi.

I have to test a couple PSUs (ripple, regulation, etc.) that are rated 12 V 62.5 A. In order to do so, I need a test load; I was thinking about using nine 2 ohm 100 W 5% resistors (about 3 usd) in parallel (giving 0.22 ohm 900 w). Taking into account the tolerance of the resistors, I get a total current in between 51.43 A and 56.84 A and, for a single resistor, a maximum dissipated power of 76 W.

For the resistors to be able to dissipate their rated power, they need to be mounted on a heatsink (12" x 12" x 0.125" aluminum sheet it's recommended in the attached datasheet). I would like to use a different type of heatsink (specifically, the ones used on computer's CPUs), but I don't know how to tell if that replacement it's valid.

To summarize, my questions will be:
1.- Is this the best way to test the PSUs?
2.- Is my reasoning right?
3.- How do I determine a different heatsink (the datasheet does not give a thermal resistance)?

File size:
19.1 KB
Views:
84
File size:
190.2 KB
Views:
50
2. Jan 5, 2016

meBigGuy

You can go roughly by sq inches of surface area. (294 sq. in. surface area per resistor)

There is a graph for the effect of reduced heat sink area. (100W resistor = 75W at 60% area)
You also need to consider the ambient air temperature.

You need to determine the effective surface area of the cpu heatsinks or their thermal resistance to ambient air.
I'm surprised you can't find thermal resistance for them.

The resistors are tested at 250C so 40W at 25C ambient would mean 225/40 = 5.6C/watt for the resistors in free air
100W @ 25C on the plates would mean 225/200 = 1.12C/watt

You will have to research the characteristics of a flat 12x12x0.125 plate to calculate thermal resistance to the plate.

Last edited: Jan 5, 2016
3. Jan 6, 2016

Tom.G

You could put the resistors in a big bucket of water. The water will boil before the resistors overheat (you may want to use running water to avoid boiling.) Since the resistors are encapsulated there shouldn't be any corrosion problem in the relatively short time needed for the tests. Alternatively, put them in a bucket of oil, light motor oil will do, then use a water jacket around the bucket. A larger bucket with circulating water would suffice. Of course use metal buckets for their thermal conductivity, just be careful to insulate the electrical connections.

4. Jan 6, 2016

davenn

not unless they were totally insulated, else the water will only short circuit everything
and with 12V @ 62 Amps that would be a spectacular and bad result !!

D

5. Jan 6, 2016

Merlin3189

I hesitate to tangle with davenn, but does it do any service to promulgate these myths?
Water is not a good conductor and certainly would not provide a "short circuit", especially when talking about extra-low voltage circuits carrying 10's of amps.
To put the boot on the other foot, would you feel safe working with a mains circuit, whose protective earth was made via a "short" connection through a tank of water?

A lot depends on the physical dimensions and placement of the conductors and on the composition of the water, but a rough calculation for seawater between two metal plates 100cm2 1cm apart came to a resistance of about 500 Ohm. That would conduct 24mA at 12V. Even allowing an order of magnitude for safety, 12 V 250mA is hardly going to make much difference here. And at 3W I'd hardly be buying tickets to see that "spectacular result".
If OP is going to be handling 50 A currents, albeit at extra-low voltage, maybe he would be more careful and use tap water, gaining another order of magnitude reduction in current.

6. Jan 6, 2016

jim hardy

Automobile headlamps make a handy load for such tests at 12 volts, readily available and not too expensive
just be aware they have inrush current when first turned on that's several times their steady state current
so don't switch them on all at once.

7. Jan 6, 2016

Averagesupernova

I have done this as well. I used sealed beam bulbs with a 2 filaments. They can be wired any combination of ways. Put the 2 filaments in series, parallel or just one at a time and of course separate bulbs can be wired up in different configurations. Once the filament has been warmed by starting it with both of them in series they can be switched to parallel.

8. Jan 6, 2016

Hesch

Use about 12.5m , 1mm2 varnish insulated copperwire, wind it to at coil and drop it into a bucketful of water ( with ice cubes ).

Then you will have about 0.22Ω, which you are free to adjust by varying the length.

9. Jan 6, 2016

totota

10. Jan 6, 2016

Ummm.. Regulation in my view is measured in mS & I do not think you need a final circuit that can handle the power continuously...headlamps will start with low resistance and increase as they heat up - impossible to get a quick accurate reading since when you turn them on the PSU will probably go into Overcurrent Shutdown.

If you need to test the total temp rise for continuous operation - then the thermals (heatsinks) need to be looked at as a continuous operation issue. ( Light bulbs here would be OK)

So --For Impulse use Resistors - and switch them in and out quickly (looking for stability) - for a power resitor I doubt you will kill it in 1 Second ( their DS should have a Time Based Rth curve) - if you do need a long term load - then the headlamps, lightbulb may be a good (cheap) way to go as long as you manage the inrush current.

A clear definition of the test requirements ( which follow from the performance specification) is in order. 780W PCUs?

11. Jan 6, 2016

totota

The PSUs are HP common slot, the ones used in servers.
The tests I was planning on doing where regulation with load (which should be within 5% for atx) and ripple (120 mVpp for atx).

Regarding inrush current (100 ms max?), I couldn't find the time needed for the overcurrent protection to shutdown the PSU. But, if indeed the PSU gets shutdown because inrush current, I could build a current limiter (which seemed easy, although I barely glanced at it).

PS: With "regulation" (I don't know if thats the right name), I mean that the output stays in between 12 V +- 5% under load, in this case.

PS2: Another test could be to check if the over temperature protection (if it exists) doesn't shuts down the PSUs after a certain time.

Last edited: Jan 6, 2016
12. Jan 6, 2016

meBigGuy

13. Jan 6, 2016

Tom.G

Hmmm... I didn't know headlamps plus a 60Amp current limiter had gotten cheaper than the resistors (3usd each in the OP). Guess I'll have to check that out for my next project.

14. Jan 7, 2016

jim hardy

I sure missed the $3 quote for those resistors. I'd have expected several times that must be something wrong with Mouser's site, they show around$180 for that RH100 series http://www.mouser.com/ProductDetail/Vishay/RH1002R000FJ01/?qs=%2fha2pyFaduhZltrlxzeKEwOeLQ5p1%252b2wSoGPpgSIzN0%3d [Broken]
ebay $43 http://www.ebay.com/itm/DALE-VISHAY...OR-Fast-Shipping-Trusted-Seller-/321903201980 Digikey shows more reasonable but doesn't have the 100 watt series. 50 watt run ~ 7 bucks apiece,$3 apiece would be attractive. Those are nice resistors .

Last edited by a moderator: May 7, 2017
15. Jan 7, 2016

Merlin3189

Absolutely. I was way out. Thank you for checking.
I'm not sure where I slipped up in my calculation. I looked up a table of values and did a sum in my head just to get a figure to use. Because it came out somewhere I believed it should (or maybe wanted it to), I used my erroneous result without question.

My apologies to davenn for falsely criticising him (and more so because I did not do it as politely as meBigGuy has corrected me.) I hope I'll be even more careful in checking, before I publicly disagree with someone in future.

Just for the record I now think my 100cm2 plates 1cm apart should pass about 60A at 12V in seawater and about half an amp in the most conductive drinking water.

I knocked up a test with two 10cm square pieces of pcb board held 1cm apart and I find that
my drinking water passes about 120mA, equivalent to a conductivity of 0.01 S/m
35 g/l brine pushes my 30 A psu into current limiting for a second or so before gassing increases the resistance and drops the current back to around 10 - 15 A. (Although there was plenty of gas with the brine, which could be quite dangerous, no sparks.)

16. Jan 7, 2016

jim hardy

i know when an extension cord falls off the dock into seawater
it fizzles and smokes while you're running to unplug it

good reason to have GFCI's there .

17. Jan 7, 2016

meBigGuy

Yeah --- I checked that at mouser also and was very surprised.
Maybe totoa can share where he is getting that price. If he uses the 250W resistors, they can dissipate 100W each in 25C free air.

I think resistive loads are the correct way to do PSU checking (short of a real dynamic load bank). Being able to switch/connect them in groups will also allow one to verify the step response under load.
(But, that is also an additional cost, and you need to be careful about switch/connector capacity).

18. Jan 8, 2016

Tom.G

Try searching the suppliers for "100w" in category "resistor". There are other mfg's. of resistors with a wide range of prices/availability.