How can we safely test a mains powered device?

Question

We have small non-isolated 230VAC devices that transform mains to LEDs running at 32V, 20mA. The lights are directly connected to live, neutral and earth like a standard lightbulb would be. Since there are some problems with the durability of the devices, we plan on thoroughly testing them to find the weaknesses in the product. Our assumption is that when the device is plugged in or powered when the AC voltage is at its highest level, the steep flank is not filtered out which results in the breaking of resistors. We of course want this the testing to be safe for the testing equipment and ourselves.

Up until today we have been testing with a residual-current device in between mains and the DUT in combination with a circuit breaker. The DUT is earthed.

My question is whether this is a good setup for testing the devices or would it be better to use a mains isolator to plug the devices into while not connecting the DUTs to earth. The oscilloscopes would be earthed and equipped with a differential probe.

Lastly, how should the testing method change when another device is tested that is also connected to mains but has a transformer on-board?

Answer 1

'If you think there is an issue with current spikes at turn-on near mains peak, you can test for that particular problem, deliberately creating that condition and applying it to the DUT. Of course you might want to analyze the circuit and see if it should or could be a problem.

"Testing to find the weaknesses" blindly is not likely to be thorough since there will be inevitably be some failure mode (perhaps induced by the users) that you have not thought of. Examples that come to mind are physical abuse, use in an inappropriate location such as inside a shower or outdoors, rapid repetitive on/off for some reason (children playing simulating lightning, or in a marquee chaser display), abnormal temperatures because of fixture design or location, voltage spikes or sustained overvoltage because of particular geographic location and lightning strikes or bad power regulation etc.

Answer 2

Given that the issue is resistors burning out, you have to consider the energy levels needed to do that. You not only need to exceed the power rating of the component - you need to do it for some period of time. You can experiment with this yourself - take (say) an 0.125W resistor and give it 0.25W with a bench supply. You will likely see that doing this for one second doesn't have a lot of effect (or does it? I am not going to do the experiment for you.) If you run the device at 2x or 10x rated power, how long does it take to fail?

You should do this for the actual components which are failing and then try to figure out what is a likely cause. Given that most transients are short lived (a few ms) events, is it likely that they contain enough energy to cause burning? Can you somehow simulate such an event? And if you make a change to the circuit to protect against such events, do the failures you see stop?

And very importantly - are there other possible explanations? Perhaps the circuity has some odd pathology where excessive voltage somehow ends up across the resistor for an extended period?

This kind of analysis is not easy (especially when the failure only happens to a small percentage of units and then very occasionally) but these are the kind of ways that I would try to test ideas about what the cause might be. It is very easy in these situations to get hooked on one idea about the cause - and then months or years later discover that it was something completely different.