# Why can a switched-mode-power-supply be dammaged without load?

I've read that some SMPSs can be damaged if they are being operated without any load (or under a specified current). What is the actual cause of this? What is happening if it is damaged due to this?

Does OVP/UVP/OCP/CSP (etc) protect from damage due to no/low load? I've seen some ATX PSU have these protections (however none of them seem to have a specified minimal load).

How do (modern) mobile chargers and other (similar) consumer products cope with this problem? After all one may expect to be able to just plug in it to the power outlet without destroying it.

• Please be specific - what SMPSs have you cause to believe suffer from this problem? Generally boost and flyback converters suffer from no load issues but you could write pages on this and miss the whole point of what might be happening with the specific (but unspecified) device you talk about. Feb 17, 2017 at 11:03
• @Andyaka I'm in particular considering using a ATX PSU as a 12V source and I've read that one should load them to avoid damaging them. I also read other claims, ranging from it works just fine, to them just shutting down w/o load, to that it only applies to cheap PSU, to that it actually probably not applies to ATX PSU's (but other SMPS's not intended for consumer use). I'd love to hear that I can use my ATX PSU without concern of course. I don't really know if this is a general issue or only an issue for specific types (and in that case which one), so sorry if I can't be more specific. Feb 17, 2017 at 11:58

## 4 Answers

A minimum load requirement is usually due to a rather simple analog control scheme. These things sometimes can't go to 0 PWM duty cycle. I've also seen a case where continuous pulsing was required because the internal pulses were used to run the charge pump for making the gate voltage supply of a high side N-channel MOSFET.

All the reasons for needing a minimum load can be worked around. The main reason for still doing it today is to strip off every last fraction of a penny in cost. This can matter for high volume products used in somewhat controlled applications. PC power supplies are a good example of this.

Note that violating the minimum load requirement doesn't usually damage the supply. It will make the output voltage go out of spec, possibly damaging whatever it is connected to.

Seems that you're talking about offline isolated converters.

Actually, this is valid for especially Forward, Push-Pull and Bridge converters. The common thing for these is, as you might guess, having an LC output filter (Surprisingly, despite not having an LC output filter, an LLC resonant converter does also need a dummy load!).

If one of these converters with LC output filter works with a proper load which is sufficient to keep the instantaneous inductor current (remembler $di = 1/L \cdot V_Ldt$) above zero then this filter can average rectangular waveform coming to its input from the secondary winding. If there's a light load causing instantaneous inductor current to fall zero then the output filter will not perform its job and lets its output voltage to rise as high as its input voltage.

Also, this problem becomes significant on these converters if the control circuitry (which contains oscillator, comparator etc) is supplied from an auxiliary supply derived from a separate winding on power transformer (i.e. supplying converter's control circuitry from directly the converter itself. The control circuitry tries to regulate the main secondary but it's supplied from a secondary which also should be regulated). If the control circuitry is supplied from a separate, free, independent supply then there's a chance to have a stable output even under real no load (RL = $\infty$) situation. On-Semi has an application note about real no-load regulation. Note that ATX PSUs are supplied from an independent supply (5VSB).

For Flyback converters, there's something different: Flyback converters generally does not need an LC output filter. However, they do need a dummy load, because they work on something like "store-and-deliver" principle. So, if there's no (or required minimum) load to deliver the stored energy then no stable output shouldn't be expected.

(however none of ATX PSUs seem to have a specified minimal load).

Because all the outputs have a dummy load resistor. Go to this page, open a schematic and take a look at output section. You'll see resistors as dummy loads.

Why can a switched-mode-power-supply be dammaged without load?

in the 1990s one of the "highly acclaimed" class d amp designer asked me why his famous amps were blowing up under light load, :)

it is quite simple: with very low duty cycle, and insufficient drive current, the switch goes out of its linear region and is no longer fully on or fully off, dissipating the most amount of power.

that gives out the magic smoke.

the solution is to limit the duty cycle (can be hard to do) or to increase the drive -> very counter intuitive to some.

Yes, ATX & consumer power supplies are often designed to deal with no load conditions, you would have massive failure issues with them otherwise. But, all bets are off with other types of SMPS's.

I'm not sure if this is an issue here, but, some power supplies with multiple outputs needs a minimum load on one output, for the other outputs to operate in the right voltage range.