I have a switched-mode PSU for a gaming laptop. It has a bad habit of burning out the switches on mains extension panels when I turn it on (two in the last four days). The turn-on surge is always audible and spectacularly visible (sparks) if I plug in with the power on. I don't think the PSU is faulty (I have two to compare), probably just not that well-designed. I need to turn it on and off fairly regularly - it's not possible to leave permanently powered on.

I qualified in electronics (before SM-PSUs!) and write software these days, but I understand roughly what is going on.

Is there anything I can put in series on the input side to reduce the spike? A ready made device would be great, but I'm happy to build something. I'm guessing some sort of inductor will do the job, but I'm not sure of all the factors/trade-offs involved and I'd rather design well than just over-specify, plus I don't have any serious test equipment left.

Or is there another approach?

The PSU is specified as 100-240V / 1.7A AC in, and 19.5V / 6.15A / 120W DC out. I am in the UK where we use the full 240V.

  • $\begingroup$ This sounds so wrong. No 120W laptop should affect mains. Have you tried it on another outlet? In NA, we have 15A and 20A outlets. UK should have the same. Is your laptop far from the mains panel? Can you move it closer to test? Take it to a friends house and test. Something is significantly wrong. $\endgroup$ Commented Oct 1, 2020 at 18:33
  • 1
    $\begingroup$ @StainlessSteelRat Agreed. A straight 120W draw shouldn't affect much. The issue is that some switched mode PSUs transiently draw high current at startup to charge a capacitance. This transient current spike represents a lot more than 120W for a brief period of time. In the UK, the maximum current for domestic cabling and devices is 13A, so 13A * 240V = 3120W. $\endgroup$
    – riemannzz
    Commented Oct 3, 2020 at 9:06
  • $\begingroup$ The transient spike probably exceeds this, albeit for a fraction of a second. My issue could be either a poor PSU design or a failed component. The poor design would probably be that not enough attention was paid to input current limiting. The failed component would probably be something like the thermistor as detailed below. $\endgroup$
    – riemannzz
    Commented Oct 3, 2020 at 9:07

1 Answer 1


OK, as it's been so quiet, I've looked at this in more detail.

The accepted solution seems to be an Inrush Current Limiter based on a negative temperature coefficient (NTC) Thermistor. This sits on an input line and can be before rectification (ac) or after (dc), but obviously must be before the capacitance that presents almost zero impedance at startup. At room temperature, the thermistor has an impedance that significantly restricts the inrush current. As current flows through, it heats up and the NTC implies that its impedance drops with increasing temperature. At some temperature, it reaches a balance where heat is being radiated at the same rate as it is being generated and a steady state is reached.

Turning to more quantitative issues:

  1. Typical impedances vary from less than 1Ω to 100+Ω at room temperature, typically given as 25°C in manuals.
  2. The impedances can drop to fractions of an Ω at working temperatures which can easily be in excess of 200°C. This high temperature must be allowed for in terms of mounting and ventilation.
  3. The dissipation constant (power lost per unit of temperature above surroundings) or "thermal inertia" of the device means that it can take quite a few seconds, even a few minutes, after power-off for the temperature to drop to a level where the impedance has significantly elevated again. If the PSU is turned on again within this time frame the original inrush current problem will at least partially manifest as the high-temperature/low-impedance combination will not restrict it enough.
  • $\begingroup$ Good job. You can accept your own answer to mark the question solved. $\endgroup$
    – Transistor
    Commented Oct 30, 2020 at 17:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.