Sizing Resistor Wattage for One Short Pulse

Monte Carlo simulation in some SPICE variant is probably the way to go unless you enjoy pain

What SPICE simulator will model thermal effects? How would Monte Carlo analysis help? Isn't Monte Carlo for determining the random effects of part tolerances in a system? That would not be helpful here because the circuit is too simple.

I do not need spice to know the pulse duration and shape. That is a trivial Laplace Transform problem.

What I need to know is roughly what the resistor's peak temperature will be.

Some resistor companies (Panasonic?) give curves of pulse power vs. time for just this sort of situation. You'll have to decide what's the equivalent rectangular current pulse, but once you make that decision, you can pretty much read things off the chart.

Note that for the MOSFET current rating to mean much the gate needs to be turned on hard -- the transient where the MOSFET is still dropping considerable voltage and the current is high will dissipate far more power in the MOSFET than the fully on condition -- so you need to pay attention to your gate drive, to make sure the FET comes on fast enough that it doesn't fry.

...

The maximum transient heat pulse is if a full 16V is applied to a discharged capacitor; half the energy in an RC circuit goes into any resistance during such charging, so you have a heat pulse in each capacitor of

To my mind, if that won't boil the electrolyte and break a seal, you don't need to worry about that surge in the self-heating of the capacitor. Unless there is a risk of a bored operator flipping an on/off switch a LOT, or unless you have elements (switches, wires), that might fail in such a surge, by quick processes (look up "exploding wires"), maybe you don't to take any precautions.

Resistor-and-relay shunt is the most elaborate Resistor-and-MOSFET NTC resistor (heats up after turnon and becomes more conductive) inductor Fuse Resistor

are all series additions that would limit inrush current, at varying costs, and ought to get some consideration.

On Sun, 12 Jul 2015 21:50:30 -0700, Artist Gave us:

And it will eventually fuse open, sometimes closed.

You want a "bulk device", as in bulk resistance medium, as in carbon comp or a wire wound will also work for this application.

Surface mount is a thin metal film, and the pulses will do (create) tiny damage sites on it. Think "liver cirrhosis". Eventually, the value will drop. If the pulse is large enough, it will happen with each stroke.

Perhaps I'm not understanding your problem correctly, but given the component tolerances you're talking about, isn't this the sort of situation that Monte Carlo is made for? You say that the capacitors have tolerance of 20% - would your mathematical analysis for the appropriate rating of the resistor to limit the inrush current hold in the unlikely, but possible situation where you get a batch of capacitors both 20% larger than nominal, or when the MOSFET threshold voltage is higher than you expect? Or some combination?

If the circuit is as simple as you say, why not just set the whole situation up in SPICE and have the computer run a couple thousand simulations, varying the component values around stochastically. You can probably find the thermal time constant of the particular resistor package you're interested in (probably good to use a 'bulk device' as someone else mentioned) - you can then look at the worst case power dissipation on the histogram averaged over several time constants and derate a bit further from that.

So long as the resistor's power dissipation rating is not exceeded by the pulse when averaged over an appropriate window, I guess I'm not sure why you need to know exactly what its peak temperature will be?

I am not interested in numerically calculated probability density curves. I am interested only that in the worst case scenario the ripple current will nor be exceeded. That is not a Monte Carlo problem.

The MOSFET has a gate driver that will drive the gate with 5V. I know that 3.3V from the processor's gpio will not be enough.

Is this also true for thick film surface mount?

Right. The OP should use a "pulse-withstanding" resistor:

or

(You just need the right search-fu.) ;)

Cheers

Phil Hobbs

Yeah, I know nothing of capacitor in rush current. But I'd guess that it is different than ripple current.. that's happening 60 times a second.

George H.

On Mon, 13 Jul 2015 06:47:26 -0700, Artist Gave us:

There are SMD resistor form factors sold that are supposed to be able to handle lifelong pulse operations without significant change. I just do not know who the makers are as I made HV supplies, and dealt with bigger pulses, and it was back in the thru-hole days.

There likely are devices that would suit your needs though.

You can form the leads of a short WW or carbon comp to fit 0805 or so layout pads too.. More labor intensive and not typically pick and place, but they do work.

The resistor I now consider is 4-1879233-1, this is 5W, 4.7ohm, wire wound:

On Mon, 13 Jul 2015 07:15:12 -0700, Artist Gave us:

Looks pretty nice. The load life stats show that it can

circuit board around it likely will be too, since you say the pulses will be momentary. You could extend one of the pads (masked of course) under it to help conduct any heat into the PCB substrate..

Why is Ileak involved? It's typically very low for polymer caps.

Ripple is bad for electrolytic caps because it cooks them. Single high-current events produce little heat. You can charge any electrolytic cap to rated voltage and short it with a screwdriver and make an impressive BANG, maybe weld the screwdriver to the terminals, but it won't hurt the cap.

Do you have any reason to expect that polymer aluminum caps need current limiting? You could try it.

• Err.. You better take a course in simple electronics. Foistly, a current limiting resistor would be in SERIES. Secondly, "inrush" is not "ripple". Thirdly, the MOSFET's on resistance may give sufficient current limiting; net current may be far less than what you want/need.

• No, you cannot assume that; go back to basic electronics.

• Leakage current in the presumed scenario has nothing to do with the simple circuit described so far; go back to basic electronics.

• Nope; go back to basic electronics.

Den mandag den 13. juli 2015 kl. 06.50.06 UTC+2 skrev Artist:

wild idea,

stick one of these:

in the ground lead of the caps, permanently turned it with a resistor to input.

-Lasse

The resistor is in parallel with the MOSFET, not the capacitor. Both the MOSFET and the resistor are in series with the capacitor. The resistor and MOSFET are between the negative terminal of the capacitor and the ground. When the resistor is no longer needed to limit current, because the capacitor is as charged as can be, the MOSFET shunts the resistor.

The leakage is a factor because when the capacitor is fully charged it drops voltage across the resistor. When the MOSFET turns on that voltage drop suddenly disappears, and that means a sudden voltage change on the capacitor. Because of this very high dV/dt when the MOSFET turns on what limits capacitor current is its ESR.