Inrush into offline flyback via 10R resistor

Hello,

We are doing a 30W PFC’d Flyback SMPS with an input stage as in the attached. We have a 10R inrush resistor which is ROX1S (datasheet below).

Just to see what would happen, I changed the X2 capacitor, C1 from 220nF to 2u2.
The 10R inrush resistor was damaged the next time I switched the power supply on, obviously due to inrush into the 2u2. (there was also a 1A fuse [not shown in schem] which did not blow). The resistor did not appear physically damaged, but was open circuit when checked.

I would not expect a surge rated 1W resistor to blow under these circumstances.

Nothing in the datasheet says it would blow up due to such a short term power spike…..i mean, as you can see in the LTspice simulation, the power peak is 10kW and the waveform shape is like a 1.5us/25us ‘spike’.
I mean, surely this is far too short a time to heat up such a resistor to the point of damage?

(the power supply worked fine when I replaced the 10R and changed back to a 220n X2 capacitor)

When the 220n X2 capacitor is fitted for C1, then the power peak is 8kW and the spike is of a 500ns/2.5us shape. How would we know that this wouldn’t cause wear out over time and blow up?

[we are now trying to come away from NTC’s for inrush limitation as they don’t work well when hot, and also, inrush relays etc are too expensive for us…as such, we now want to have an inrush resistor, sized to dissipate no more than 400mW for the sake of efficiency…after this , I confess I will probably put an NTC in aswell].

The resistor cannot have heated up much at all in the 26.5us of the power surge……so this actually makes me wonder, what is the long term effect of inrush into the X2 capacitors is? …..what is the time wear out process for them, years, decades?

ROX1S resistor datasheet:
https://www.te.com/commerce/Documen...pdfEnglishENG_DS_1773271_E.pdf3-1625892-1

It is true but rarely talked about: peak power does matter.

Think of it like a laser pulse: total power is small over one second but the peak is very large. There is no time for the energy to dissipate over the whole body. The conventional ratings apply for steady state.

But it is same for other devices too: a transistor rated for 1W power cannot take 1kW over one mS. There is too little time for the local heat to spread over the whole body. Of course it all depends on the manufacturing technology.

I guess (but not really sure) good old composition resistors can take peak power better. We need to be very conservative when it comes to peak power. The resistor now acts more like a fuse.

The resistor cannot have heated up much at all in the 26.5us of the power surge...

Click to expand...

It is the same way the common filament lamps burn out with a very modest voltage spike...

The energy dumped in a narrow area due to a spike CANNOT dissipate fast enough.

Thanks, the problem as you know with carbon resistors is that they absorb moisture over the years.
It is a shame the ROX1S doesnt give any kind of (i^2).t rating, or peak power pulse rating.

Ideally you should try to find the joule rating of the resistor. It's tricky to do but many manufacturers do have some guidelines on the joule rating and a smaller percentage of resistors (much) have explicit joule ratings:
https://www.ohmite.com/surge-high-energy/

This line has good surge curves at least:
https://www.ohmite.com/assets/docs/res_as_smd.pdf

As I mentioned elsewhere recently we switched to using joule rated in-rush PTC's with a bypass relay and a DC bus sense circuit to verify full charge before closing the bypass. It's a solution I like quite a bit. Though a PTC can't work without the bypass which is overkill for this I assume.
https://en.tdk.eu/download/1537850/ab335b2067bf7f0f55819b22912d7faa/ptc-icl-pb.pdf

resistors, many of, are thin film deposit, onto an Al2O3 substrate thus the short term rating of the thin film is really very low, we have tested a lot of 100 ohm, 2W, 3W, 4W, and 5W resistors - wirewound - and found all of them to be very disappointing - except the best made 5W white ceramic (rectangular) ... a spiral of thin film on a ceramic tube just does not cut it ...