Simple Power Supply Load Testing Question

[Rix2357]

I have a simple question about testing a switching power supply under load.

I have an engineer that says a power supply from a very reputable power supply vendor.

The power supply is rated for 5V 4A +/- 5%.

My engineer says he has tried hand holding a 25W resistor with 1.6 ohms resistance and also 2.5 ohms resistance to the power supply leads and also hooked up an oscilloscope to measure the voltage across the resistor.

Is this a proper way of testing a power supply? I can see that in one of the test configurations, the voltage will shoot up to over 7V, then drop to 2V, and then stabilize at 2.5V.

The manufacturer has said their power supply doesn't have a problem and that they use state of the art load testers that can test at 0% load and up to 75% load etc.

Hope I can get some help on this.

thanks,

Rix

 

[BrownOut]

Man, that's a BS answer from the manufacturer! It's a crock. Work with someone who knows what they are talking about.

 

[colin55]

Testing with a resistor is a true and genuine way to test the current capability of anything. You can feel the heat with your fingers and you know it is delivering energy - current - power.
Sometimes the output can be high frequency and (50kHz) and this will upset the reading of a multimeter, but a resistor test will be genuine.

 

[Rix2357]

Sorry, I think I may have found one of the problems. The power supply is rated for 3A, so testing with the 1.6 ohm resistor is 3.12A. That was the worst case, but I still need to check with a lot of other stuff. Looks like my engineer did one thing wrong, but I'm sure there are other issues either with the manufacturer or engineer.

Second question, should I worry about the over and under shoot voltages?

Thanks so much for the quick reply.

Rix

 

[colin55]

Wrong thread

 

[BrownOut]

Wrong thread, colin Heh!

 

[BrownOut]

Second question, should I worry about the over and under shoot voltages?

Thanks so much for the quick reply.

Rix

Yes, you should be able to get specs for that.

 

[Reloadron]

I don't get the use of the scope to look at voltage? When loading a power supply I place a scope accross the load to look at any ripple or noise content. Beyond that, yep, a resistive load (or load bank) is generally used to test a power supply. Additionally running it off a variac will be sometimes used to check its ability to regulate unler load with varying line voltage changes.

Just My Take
Ron

 

[UnloadedQ]

I don't get the use of the scope to look at voltage? When loading a power supply I place a scope accross the load to look at any ripple or noise content. Beyond that, yep, a resistive load (or load bank) is generally used to test a power supply. Additionally running it off a variac will be sometimes used to check its ability to regulate unler load with varying line voltage changes.

Just My Take
Ron

A scope is also used for dynamic load testing where a minimum load resistor is connected and a full load resistor is switched onto the supply output by a relay or transistor. The scope is synced to this switching so the voltage can be checked for droop and overshoot. The same is done using a relay to apply one of two incomming AC voltages (max and min line voltages in the US are 85VAC and 132VAC) and the affect on the output voltage observed. Doesn't have to be super rapid. Usually a cycle of 3 seconds is adequate if you don't know the dynamics of expected usage.

 

[indulis]

First, load tranisent testing is only a sign of conditional stability, but probably the only test you can do external to the supply. A simple switched "R" is probably not realistic... put some "C" in there too. That "C" will also mess with the loop... best to try it with your actual, if not higher, load C.

 

[mneary]

From no load to overload and back is not a condition that a power supply normally experiences. You have to think about how you really plan to use this component, and is this a fair test. Will you ever do something remotely similar in the application?

 

[BrownOut]

Also, don't expect a SMPS to work with no load. Find out the compliance current specification, and make sure the test current is within parameters.

 

[unclejed613]

the drop to 2.5V sounds like foldback current limiting, which is used in a lot of power supplies. when you go over the rated current of the supply, it begins reducing the output voltage, is this an open frame supply?

 

[indulis]

Also, don't expect a SMPS to work with no load.

True, in theory SMPS don't like to run with no load, but they are routinely spec’ed that way and do in fact run fine at zero load (bode plots verify stability under the no load condition). While it is topology dependent, some PWM's enter a "pulse skipping mode" (making bode's impossible), but they behave nontheless.

 

[unclejed613]

early SMPS supplies would go into shutdown if they were unloaded. the early IBM PC power supplies are an example of this. the simplest way to test them was to connect a hard drive to them.
most newer SMPS supplies aren't so picky, especially ones of the "wall wart" variety, they should have voltage at the output whether loaded or not. some "green" SMPS supplies may revert to the old way, however, to save energy when the load is removed, or the device it's attached to turned off.

a lot of power supply testing is done with active loads as well, usually a MOSFET and a variable voltage source. this is for dynamic testing. for instance a power supply rated at 10A is operated at 9A, and the mosfet is fed a repetitive pulse that periodically takes the current to 15A for 10mS to see how well the supply can handle the transient loading beyond it's continuous rating. you don't need to get so fancy with a test load, a simple resistor will suffice.

 

[MikeMl]

As a minimum, you should test the supply's delta V and transient response with a sudden application of rated load, and when the load is suddenly disconnected.
Testing ripple both loaded and unloaded is also a good thing.