I have a switch mode power supply, outputting 48V 20A that has a pair of NTCs (inrush limiters, I suppose) that get quite hot.
I know they’re supposed to get hot, that’s how they work. But how hot is normal?
At 25% output (1A AC input), I measure 100 degrees Celsius with my FLIR. At half output (2A AC), it’s already up to 160 degrees. Does this seem about normal? How hot should I expect it to get at full load, 4 Amps? (I know I^2 * R, how much does the R go down?)
The markings on the NTC says “NTC 5D-15”
The Wiki article about NTCs has a picture of a blown NTC due to getting too hot, I hope to avoid that!
NTCs used as inrush limiters can run pretty hot. They’re not the most efficient way to limit inrush current, and it’s not the most effective (they won’t limit if the current is toggled while they’re still hot), but they are the cheapest.
That NTC seems to be this one, which is rated for 6 A at 25 deg. C ambient, and 200 deg. C working temperature, so 160 deg. C is still within its limits.
It’s possible to work out the resistance if you know the working temperature, ambient temperature and current as the thermal dissipation coefficient figure can be used to calculate the dissipated power, but easier to just measure the voltage and current. Sadly the datasheet doesn’t give a characteristic curve or coefficients thereof so it’s not possible to say what the resistance will be at a particular current without measuring the volt drop at that current. Keep it under 6 A at 25 deg. C ambient and you won’t exceed the spec, but it’s always nice to back it off a bit and have some safety margin. It’s even nicer to know a few characteristics so you can predict the maximum safe current at 30 deg. C ambient, say, but in this case the manufacturer doesn’t supply them so it’s up to the user to do their own validation.
Yea, I hate NTC thermistors that are used for limiting inrush current.
A better but more costly[sup]1[/sup] way to limit inrush current is to use a power resistor in parallel with the contacts of a time delay power relay. When power is first applied, the contacts are open and all the current goes through the resistor. After a second or two the relay contacts close, and then all current goes through the contacts.
[sup]1[/sup]More costly for the manufacturer. You could build such a circuit pretty cheaply using surplus parts.
Wow, great find on the datasheet, thank you!!
Good to know I should be safe at 2 Amps / 160 degC with some margin to spare. I’ll have to check the temps again if my application goes up to the full 4+ amps.
Poking around, I now found out that the PCB has the two NTCs close together connected in parallel!! Surely that’s not right! Seems now I have to worry whether they’re underrated for the big filter caps in the supply, since theory says one of them is going to get the full hit of the turn-on.
I’m fairly sure the circuit uses the well known 110V/230V voltage doubler trick. It has a small 110/230 selector switch on the PCB and the two big caps are rated 200V/1000uF. So the caps are in series and I can take 250V RMS at 500uF, which gives only 31J (E = 1/2 * C * Vpeak^2).
That sounds low for an NTC (actually, 500uF also sounds low) so would it be reasonable to assume that this will be OK, even though the NTCs are wrongly connected in parallel?
Thanks for the help in any case!
Correct; NTC thermistors shouldn’t be put in parallel.
Yea, but this assumes a ballast/equalizing resistor is placed in parallel with each cap when the two caps are placed in series.
If ballast/equalizing resistors are not used when the two caps are placed in series, one of the caps will probably explode soon.
Thank you! Good point. I checked and fortunately it has 150k resistors across the caps (one of them covered in a glob of circuit glue).
Yeah, it’s a Chinese power supply (and cheap too), but so far I’m very happy with it, good efficiency, very good power factor. But always a good idea to check for gremlins.