Hi everyone,
I just read that capacitors produces heat and this heat needs to be dissipated. When installed in a PCB, this is not a problem as the copper traces can act as a PCB.
I want to do a simple unregulated power supply using point-to-point (think tube wiring). The most inexpensive capacitor I've seen within the capacitance required/desired is a snap-in.
Any ideas on how to dissipate the heat?
The capacitor will be mounted directly to the bottom metal plate of the chassis using clamps (with the capacitor "hanging" upside-down).
some ideas/thoughts that comes to mind:
PS. I have not used snap-in or screw capacitors before. my experience is limited to using soldering (<4700uf and less) caps to PCBs.
I just read that capacitors produces heat and this heat needs to be dissipated. When installed in a PCB, this is not a problem as the copper traces can act as a PCB.
I want to do a simple unregulated power supply using point-to-point (think tube wiring). The most inexpensive capacitor I've seen within the capacitance required/desired is a snap-in.
Any ideas on how to dissipate the heat?
The capacitor will be mounted directly to the bottom metal plate of the chassis using clamps (with the capacitor "hanging" upside-down).
some ideas/thoughts that comes to mind:
- Perhaps I can use a heatsink grease between the "top" of the capacitor and metal plate to dissipate the heat? or maybe the copper wiring soldered to the pins is enough to dissipate the heat?
- My assumption with screw mount caps is that with a screw capacitor, I can use a copper plate to act as "PCB tracks" which would then act as heatsink. It would be easy to attach the copper plate to the capacitor because I can use a screw.
- Maybe there's no need to be concern at all?
PS. I have not used snap-in or screw capacitors before. my experience is limited to using soldering (<4700uf and less) caps to PCBs.
Last edited:
I think better first to ascertain whether your caps will indeed get hot in your particular use case. Heating is from (ripple current)^2 * ESR. So what's the particular cap and how heavy a load is it going to see?
may I ask that's "^" in your formula? I tried searching the net and didn't find anything useful. it's ripple current squared? thanks again
Yes, it is squared. Power dissipated in a resistor is current squared times resistance.
Power dissipated in a perfect capacitor is zero. A real capacitor can be approximated as a perfect capacitor with a resistor in series.
Power dissipated in a perfect capacitor is zero. A real capacitor can be approximated as a perfect capacitor with a resistor in series.
Sorry but lots of wrong assumptions here:
Please state HOW MUCH heat are we talking about, what´s the heating mechanism and some actual capacitor value and size as example
Start thinking about that and come back.
Examples with numbers are preferred, blanket statements only add to confusion.
Everything produces heat, even fridges, but that´s not the main point.
Please state HOW MUCH heat are we talking about, what´s the heating mechanism and some actual capacitor value and size as example
Please consider that ´aluminum* capacitor body has about 1000 times larger area than puny pads and traces, is exposed high in the air and surrounding your presumed heat source.
Start thinking about that and come back.
Examples with numbers are preferred, blanket statements only add to confusion.
I don't think most heat transfer in large electrolytic capacitors is through heat sinking provided by the PCB, most heat would be transferred to the case by conduction and then convected or radiated from there. Snap ins are a bit better in this regard but not a significant mechanism to count on IMO. Provided you have chosen an appropriate capacitor you should not have an issue. Clamping to a metal chassis likely will provide a better thermal path for dissipation than a PCB..
Yes and you can calculate the ripple current from the ripple voltage I = C dV/dt where C is the capacitance, dV is the ripple voltage p-p and dt is the relevant time interval (typically ~8mS or so for 50Hz mains).
Take an example - you have 33,000uF and your ripple voltage is 2V peak to peak. This gives you a ripple current of 8.25A. If the ESR is 10mohm the power dissipated is 680mW which is negligible for a large capacitor.
I wouldn't overthink things too much.
Those people online that suffer from paranoia spread unfounded rumors a lot, so it's questionable babble.