power resistor dummy load for PSU testing

[FrankBuss]

I found a nice and cheap power resistor for load testings. 500W/8ohm for EUR 17:

http://cgi.ebay.de/ws/eBayISAPI.dll?ViewItem&item=370855728850

Currently I use it for testing my new Siglent SPD3303D and so far the PSU survived 32V/4A (=128W) for an hour. A detailed review will follow.

But the 500W rating of the resistor is very optimistic: It got very hot after some minutes when I tried it, (up to 100°C). Maybe it is the usual PMPO audiphool rating? Then I mounted it on a 4cm thick aluminium block (with lots of thermal grease) :



and now it gets 70°C after an hour and the block gets 35°C at the edges (@25°C room temperature).

What do you use for PSU testing and for power supply circuit testing? Electronic loads are nice, but you could buy 20 resistors for the price of one of these devices, and a resistor works for AC, too, which is even more expensive for electronic loads.

[SeanB]

You mean like this guy?


Dummy load resistor by SeanB_ZA, on Flickr

[FrankBuss]

Nice, even adjustable. What are the specifications and dimensions?

[SeanB]

IIRC it is a 10R 100W unit.

It has a companion of a 1R5 50W unit.


Resistor in case by SeanB_ZA, on Flickr

[ElectroIrradiator]

Aluminium clad Welwyn wirewound power resistors, dunked in a glass container filled with water (distilled or tap water to taste).

The resistors can survive being boiled forever, as long as you remember to pull them up hot. This prevents water from being sucked through any cracks in the glass seals as the resistor cools.

No need for a base plate. Just use resistors with appropriate power ratings.

[alm]

But the 500W rating of the resistor is very optimistic: It got very hot after some minutes when I tried it, (up to 100°C). Maybe it is the usual PMPO audiphool rating?

Depending on the type of resistor, 100°C may be a perfectly fine operating temperature. But that 500 W rating is generally only true when used with a sufficiently large heat sink. There's no way such a relatively small resistor is going to dissipate 500 W through convection.

What do you use for PSU testing and for power supply circuit testing? Electronic loads are nice, but you could buy 20 resistors for the price of one of these devices, and a resistor works for AC, too, which is even more expensive for electronic loads.

Electronic dummyloads are more convenient than a box full of power resistors that you have to mix and match: just set the current/resistance/power and you're done. Good electronic dummyloads can do modulation (eg. alternatingly pull 0% or 100% current 1000 times per second) to test stability. Small dummyloads (eg. the re:load) are quite affordable, but 500 W gets spendy.

[nctnico]

I agree electronic dummy loads are easier. Either way you need a heatsink and a fan. A good resistor based dummy load which can dissipate a couple of hundred Watts continuously is also expensive. I build this one to aid in testing & certifying poultry stunning equipment. It's designed for 600W and it consists of 8 100W resistors mounted on a tunnel-shaped heatsink with a 120mm fan. The parts alone cost about €150,-. Oh and I won't even attempt to bring it on an airplane as hand luggage 

[FrankBuss]

I agree electronic dummy loads are easier. Either way you need a heatsink and a fan. A good resistor based dummy load which can dissipate a couple of hundred Watts continuously is also expensive. I build this one to aid in testing & certifying poultry stunning equipment. It's designed for 600W and it consists of 8 100W resistors mounted on a tunnel-shaped heatsink with a 120mm fan. The parts alone cost about €150,-. Oh and I won't even attempt to bring it on an airplane as hand luggage 

Neat construction! For the airport security, do it like Jeri Ellsworth, when she explained her jerry-rigged mobile phone with alligator clips: "I'm an electrical engineer, trust me!" 

[Harvs]

For just simple load testing anything can be fine (you can even grab a toaster off the side of the road and re-configure the nichrome wire to give you the resistance you want.)

Where I find an electronic load essential is in loop stability testing for a SMPS.  With a variable supply, electronic load, function gen and scope you can quite quickly un-cover resonances in the control loop (that lead to sudden loss of phase margin) that clients have tested using power resistors and declared perfect.  At the moment I'm still doing it all by hand, but I'm part way through writing the software to automate this process.

[FrankBuss]

Where I find an electronic load essential is in loop stability testing for a SMPS.  With a variable supply, electronic load, function gen and scope you can quite quickly un-cover resonances in the control loop (that lead to sudden loss of phase margin) that clients have tested using power resistors and declared perfect.  At the moment I'm still doing it all by hand, but I'm part way through writing the software to automate this process.

This sounds interesting. How do you do it? Using the function generator to control the electronic load and then doing some slow frequency sweeps while testing the regulation with the scope? Is square wave sufficient for the load or do you need any other waveform? Why a variable supply?

[SeanB]

Dummy load for testing ground support alternators for aircraft was very simple, a 44 gallon drum filled with salt water with 3 rods on an insulating frame on top. Adjust length in water to adjust current and top up the water as it boils off to keep the current constant. 115VAC 3 phase at 50A cooks it pretty fast. 28V tester was an old steel bed frame with 5mm thick wire springs stretched across it, and a gap in the short side with a bakelite spacer. Add springs and they glow red hot to white hot, as you needed to get 28V at 8kA as a full power test. That one stood on a pile of bricks, as otherwise the white hot steel spalled the concrete apron.

[Legit-Design]

For just simple load testing anything can be fine (you can even grab a toaster off the side of the road and re-configure the nichrome wire to give you the resistance you want.)

Etching/milling from normal FR4 pcb also works as reasonable resistor/heater.

http://reprap.org/wiki/PCB_Heatbed
http://www.instructables.com/id/PCB-Heater-Diy-Joule-heating/?ALLSTEPS
Even if the resistance varies with temperature it is an easy way to dump some load for testing purposes. PCBs are also easy to stack up for higher power handling. Fan cooling might be a good idea for longer durations and higher power handling. For trace sizing I found kicad pcb calculator useful for quick reference.

[fpliuzzi]

Where I find an electronic load essential is in loop stability testing for a SMPS.  With a variable supply, electronic load, function gen and scope you can quite quickly un-cover resonances in the control loop (that lead to sudden loss of phase margin) that clients have tested using power resistors and declared perfect.  At the moment I'm still doing it all by hand, but I'm part way through writing the software to automate this process.

This sounds interesting. How do you do it? Using the function generator to control the electronic load and then doing some slow frequency sweeps while testing the regulation with the scope? Is square wave sufficient for the load or do you need any other waveform? Why a variable supply?

Recently I came across this simple to build electronic load that's controlled by an arb/func generator. If I not mistaken, this short article might be related to the test setup described by Harvs. I need to put one of these together myself because it looks like it could be quite useful for power supply testing.

http://gpete-neil.blogspot.com/2012/03/building-electronic-load-with-general.html

Regards,
Frank

[nctnico]

Where I find an electronic load essential is in loop stability testing for a SMPS.  With a variable supply, electronic load, function gen and scope you can quite quickly un-cover resonances in the control loop (that lead to sudden loss of phase margin) that clients have tested using power resistors and declared perfect.  At the moment I'm still doing it all by hand, but I'm part way through writing the software to automate this process.

One of the best ways to test any power supply is to use a dummy load made with bipolar NPN transistors. The collector of an NPN transistor behaves like a current sink so it has a very high internal resistance. Any loop instability will be uncovered this way. You don't even need a control loop in the dummy load, just thermal runaway compensation.

[Harvs]

Where I find an electronic load essential is in loop stability testing for a SMPS.  With a variable supply, electronic load, function gen and scope you can quite quickly un-cover resonances in the control loop (that lead to sudden loss of phase margin) that clients have tested using power resistors and declared perfect.  At the moment I'm still doing it all by hand, but I'm part way through writing the software to automate this process.

This sounds interesting. How do you do it? Using the function generator to control the electronic load and then doing some slow frequency sweeps while testing the regulation with the scope? Is square wave sufficient for the load or do you need any other waveform? Why a variable supply?

Sorry I didn't explain myself well.

The function gen is used with an isolation transformer and used to inject a small signal into the feedback loop. The scope is then used to look at either side of the injected signal, which effectively gives you the open loop response which you can produce a bode plot, or just look for the gain cross-over point and measure phase margin.  There's some very good detailed application notes out there on this (TI maybe?), I've got a paper copy of one here somewhere that I'll try to find.  Effectively a poor mans network analyser, but it is surprisingly effective.

However, to get a true picture you need to perform this analysis at all speced input voltages and load combinations (and well into overload).  Often it'll look just fine, but then just sweeping the load and/or input voltage can uncover a point where the phase margin disappears.  Obviously it's impractical to test all these points by hand, but a reasonable procedure is to start at no-load, find the gain cross-over point then start sweeping the load and voltage to see what happens.  By repeating this at a few points it becomes apparent quickly if there's any dodgy points.

[KJDS]

I'll either strap a few 100W RF 50ohm dummy loads together, or a selection of HP6632Bs configured as current sinks, or sometimes a combination of the two.

[alm]

There's some very good detailed application notes out there on this (TI maybe?), I've got a paper copy of one here somewhere that I'll try to find.  Effectively a poor mans network analyser, but it is surprisingly effective.

Is this the one you mean:
NSC/TI AN-1889: How to measure the loop transfer function of power supplies

[Harvs]

There's some very good detailed application notes out there on this (TI maybe?), I've got a paper copy of one here somewhere that I'll try to find.  Effectively a poor mans network analyser, but it is surprisingly effective.

Is this the one you mean:
NSC/TI AN-1889: How to measure the loop transfer function of power supplies

Yeah that's the one I was thinking of.