Controlling Inrush Current in a Power Supply

[Matthew Beasley]

If the inrush surge current of some item is so * horrendous * that it
must never happen, then a missing cycle detector is essential, followed by
instant shut down and eventual re-start of the equipment.

How much do you expect the cap voltage to drop in one cycle? If it's only
10's of volts, it's not going to be the full inrush.

 

[D from BC]

How much do you expect the cap voltage to drop in one cycle? If it's only
10's of volts, it's not going to be the full inrush.

If it's what I think it is.. its zero crossing turn on inrush control.

The inrush limiter can kick in on the next zero or the zero after the
power is turned back on.

But yeah...A brief power disruption with that type of inrush control
won't let the bulk caps discharge much and there will be a lesser
surge upon power up.
Assuming the load goes hi-Z during bulk capacitor undervoltage.
D from BC

 

[colin]

At current, the NTC in my app is the hottest component. A little
annoying for a 90% eff convertor.

I find the thermal insulation option interesting to keep the heat off
other components (if worthy) or the reduce the temp. inside the box.

I think I get the idea...
I'm imagining that a thermal/power equilibrium is reached at a lower
resistance when the NTC has insulation.
In other words...
NTC open air power disp > NTC insulated power disp.
(I^2*R1) > (I^2*R2)
where
R1=NTC in open air
R2=NTC insulated

But...slower cool down as you mentioned before.
D from BC

yep basicaly thats it, the thing to bear in mind is that a certain amount of
energy has to be dumped into the NTC
if its to limit the current until the capacitors charge up,
you can work it out but its probably a bit tricky,
but is probably similar amount of energy to the energy stored in the
capacitors.

the thermal mass of the NTC should be such that it absorbs this energy
before it gets hot so that it loses too much of its resistance.

the actual thermal reistance is only relevant to cool down time,
although you need to do the calculations to make sure the final running
temperature doesnt impact on its lifetime,
but it realy depends if you need to consider short power interuptions or
not.

personaly I find it anoying that the new psu I had to put my pc becuse the
capacitors exploded in the previous one it makes the lights flicker yet it
claims to have pfc and inrush limiting.

why they put 2 200v electrolytics in series I have no idea, the center tap
isnt used for anything.
it had the bad make of caps, it lasted 3 yrs wich seems par for the course.
cleaning the electrolyte off and replacing the caps didnt stop it from
blowing a fuse.

but yes the disipation and heating of nearby components is akward
I think we used to put it on very long leads,
with it glued inside a tube made of glass fibre to support it and insulate
it a little bit.
mainly to stop the heat going into the pcb.

this also stoped bits of it flying off as shrapnel when we did the
conformance testing where it had to survive any component being shorted out
lol.

I dont know if a relay would be a simpler alternative for you,
but it wouldnt need to be a big one as it switches always at zero voltage,
and only sees the maximum voltage accros the open contacts for a short time.

Colin =^.^=

 

[Phil Allison]

"Matthew Beasley"
"Phil Allison"

How much do you expect the cap voltage to drop in one cycle? If it's only
10's of volts, it's not going to be the full inrush.

** The above applies to transformer isolated PSUs where the major source of
inrush current is due to magnetic saturation of the iron core.

With a 1 kVA transformer, the initial urge peak can be many hundreds of
amps, subsequent peaks are of all of the same polarity and extend for about
20 cycles before the event is over.

Typical "soft start" schemes seen involve high power resistors, NTCs,
relays, triacs etc but normally fail to work if the interruption is only a
few cycles. A blown AC fuse is then the result.



....... Phil

 

[legg]

I got a case of: Thermistor Vs Something Else

+----(high side lim)--+-----+
| | |
\/\/\/170VDCpk 180uF Rload~85ohm
| | |
+-----(low side lim)--+-----+
|
Com

I'm not too familiar with inrush thermistors.
I know they get hot.. which can suck.
Don't hotter electronics die sooner...??

I looked at a thermistor app note...
http://www.epcos.com/web/generator/...ta__en.pdf;/PDF_InrushCurrentLimiting_AN2.pdf
groan...

The design process looks like a pita at first glance.
Bad enough I'm turned off by the heat emitted.

It's basically cold resistance, maximum current rating and maximum
energy rating.

You're looking for a >2.5A, >5R part with 'minimal' energy limit
requirements.(often stated as capacitance at operating voltage).
If ripple current is high, get an accurate rms measurement and act
accordingly.

Energy handling capacity determines thermistor diameter, so anything
over 11mm rated for inrush limiting will likely serve in this case.
The smaller parts have the advantage of diissipating less energy to
reach their quiescent operating temperature (ie improved efficiency).

Active inrush limiting is another subject.

RL

 

[MooseFET]

Nah...
The inrush problem is in the front end of an offline smps.
I did alot of work to reduce the power inductors for the smps...
Not gonna ruin the design with a relatively massive inductor.
It's like a tumor.

I going to use a thermistor.

Extra heat! Ick poo ick poo ick poo

But I'm still interested in alternative ideas...

D from BC

A resistor and an SCR in parallel, I think, is one of the better ways
to go. You delay the start on the DC-DC until the capacitor is
certain to be charged. An extra winding on the transformer triggers
the SCR when the DC-DC starts. It is a low parts count way to go.