capacitor heatsinks

Dead chicken alert.

Not a lot of cooling is going to occur with the heat sink on the wrong side of 6 or 8 mils of plastic jacket, particularly with the fins butted right up against the mounting surface.

Cheers

Phil Hobbs

One would have to do the numbers. 8 mils of plastic conducts heat a lot better than a couple inches of air.

Just define the problem as beginning at the surface of the cap.

They look cool, anyhow.

Audio Fool Market.

Cheers

Heat kills 'lytics. It might make sense.

--

John Larkin Highland Technology Inc

Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators

They need to be gold plated. And a very high price. With a special blue heat sink grease.

Most failures tend to be at point defects. I think you would have to know the failure mechanism to determine if these devices are useful.

I'm not ruling out if they are useful or not, but I'd like to see the science behind them. Failure analysis is a very specialized and narrow field. A thankless job, but you get the best lab toys on earth. If I had access to a SEM, my G-jobs would be never ending.

Bad idea. While self-heating from ripple current might justify a heat sink, the bulk of the heat damage comes from the adjacent CPU, switching devices, and power supply heat sink. At best, the increased mass of adding a heat sink will slow down the rate at which the capacitor gets hot. The little clip on heat sink doesn't seem to have very much additional heat sink area to make much difference. It will catch more air, if there's air flow, but often the cooling air is hotter than the capacitor self heating, making that a potential loser. The little clip on heat sink will also need to be mounted high on the capacitor (after removing the vinyl coating) to catch the air flow. Looks like the fins are in the wrong direction for horizontal air flow, so vertical air flow will be needed, which is always obstructed by the PCB. Oh yeah, there's never enough room on the PCB for the electrolytics, much less for electrolytics with heat sinks. The necessary added PCB space would better be served by adding additional electrolytics, so as to distribute the ripple current between more caps. Or, just leave some air space between caps in order to get more air flow and less turbulence.

However, looking at the tiny drawings, it appears that the heat sink is wrapped around a "computer grade" capacitor with big screw terminals. In order to get 2-3 C/W thermal resistance, the heat sink would need to be fairly large. However, those large caps usually don't need additional cooling. It's the small caps that might benefit from additional cooling.

Incidentally, if the aluminum can electrolytic manufacturers though that additional cooling was necessary, they could have easily corrugated the aluminum can and thus increase the surface area. That would also make the can stronger, so they could use thinner aluminum.

Drivel: Phil Allison gave me the clue as to how to measure self heating. The capacitor ESR decreases with increasing temperature. I fired up an ATX power supply on the bench, with an exposed PCB, and running a handy motherboard. About 90 watts total draw on a Kill-A-Watt meter. I then pulled the plug, shorted the power supply capacitors, and measured the ESR (as distributed among several caps in parallel). From the ESR value, I could estimate the temperature inside the aluminum can. About 40C was typical, hardly anything worth getting excited over.

The heat sinks are for 3.03 INCH diameter electrolytics. Not milimeter.

Cheers

Perhaps, the only time I fried a cap that size, the polarity was reversed. With large ripple currents dumping 30W of heat does have advantages. There must be a market with the power inverter guys that do the big stuff. One rep pointed out he sell thousands of TO-260 case devices to a customer because of the RdsON vs the Larger IGBT modules. Thats the industrial market, its pretty big and can drive innovation like this heat sink.

Cheers

Thanks. I missed the dimension in the title, but caught on in my next paragraph (below). Most of my comments are still applicable for both large and small caps.

I find this comment odd: "To function reliably in today's densely packed systems, the capacitor must be kept cool. Our Cap Coolers? improve the performance and extend the life of 3.0" electrolytic capacitors."

If the current switcher technology is a "densely packed system", where are they going to find the room to put the heat sink? It can't be installed near the base of the cap (against the PCB) because it's unlikely to be any room and there's zero air circulation. It might fit near the top of the capacitor can, above the jungle of parts below, but only if the cap isn't surrounded by other tall caps. I just don't see it doing any good. In any case, the fins are in the wrong direction.

What does that mean? Adding the heat sink will make the cap hotter? That would make the cap a cold sink.

A small cap has a better surface to volume ratio.

It means that I caught on that the heat sink was for large size caps as I went along, and didn't bother removing any references to smaller caps at the beginning. Sorry(tm).

A good example of pre-heated caps is the typical ATX power supply and box. The fan is mounted on the rear panel and blows air out of the box. That means the power supply intake air supply is heated by the CPU, video card, HD, etc. That air, blowing through the power supply, will pre-heat the power supply capacitors (and everything else in the PS). By adding a heat sink to these capacitors, it catches more of this pre-heated air than it would if the cap was buried among a forest of other similar caps, therefore making the heat sink and capacitor can hotter. This is a well known problem and probably the inspiration for adding a case fan on the rear panel near the CPU to exhaust most of the hot air before it gets into the power supply. (Extra credit to the ATX designers for putting the power supply at the top of the case, where it collects the hottest air in the box).

The small caps surrounding the CPU on a computah motherboard are another example. Most quality motherboards have given up on using electrolytics and now use low profile polymer caps instead. They use low profile caps, not much higher than the CPU socket, so that the caps are not in the downward air flow from the CPU fan. One would think that the designers would want maximum air flow, but because that air is pre-heated by the CPU, it will actually heat the caps more than if the caps were outside the air flow.

See any conventional electrolytics? These are all polymer caps.

Well, let's see if that's true. Surface area (minus one end with the screw terminals which has no aluminum) is: Pi*r^2 + 2*Pi*r*h Volume is: Pi*r^2*h Surface area to volume ratio is: (Pi*r^2 + 2*Pi*r*h) / (Pi*r^2*h) or (Pi*r^2) / (Pi*r^2*h) + (2*Pi*r*h) / (Pi*r^2*h) Canceling surplus terms: 1/h * 2/r Looks linear to me for both radius and height. However, it wouldn't be linear if I had included the end of the can with the leads. (OK, where did I mangle the math this time?)

However, if it really were a problem with big caps, then the cap manufacturers would have corrugated the can to increase the surface area. With little effort, it could probably be corrugated to resemble a steel trash barrel and nearly double the surface area.

Yeah, little baby ones.

Fine, except that big caps tend to be bigger than small caps in both directions. Mice and elephants.

I remember seeing tubes with heat sinks sort of like that.

I wish people would figure out how to build big polymer aluminum caps. They tend to be low uF and low voltage. I'm struggling now with the ESR of electrolytics, which start mediocre and go to hell below 0 degrees C.

Back in late 1960's, I ran a 2-way radio shop that had UHF repeaters on various mountain tops in the Orange County, CA area. Most of the repeaters used tubes: In an effort to improve reliability and tube longevity, we added some of those slide on black heat sinks over the smaller 9 pin tubes. Tube failures rose spectacularly, so we removed them. I also tried them on various other radios, with similar results. The GE Progress Line repeaters in the pictures had silver colored tube shields, but none were the heat sink variety. The shields also reduced tube life, but not as much as the black heat sink variety.

Some glass broadcast power tubes have a water jacket and use water cooling. Maybe wrap some copper tubing around the electrolytic capacitor and pump some water through it with an aquarium pump. See the various compuer "overclockers" web sites for details.

Polymer caps are almost constant ESR over temperature. What value uF and voltage are you looking for? Maybe a PCB full of smaller caps in parallel can be found:

Capacitance range, voltage rating and polarization Polymer capacitors are made with capacitances between 10µF and 1mF. The typical maximum voltage rating is up to 35 V, but there are polymer capacitors with maximum operating voltages reaching 100 V.

Now, all we have to do is find them.

JL is correct. Your result shows that, if either h or r or both are decreased, the area/volume ratio increases. That is, smaller dimensions gives it a better surface to volume ratio as he says.

As an aside, I always thought of the function 1/x as a hyperbola. I don't consider it linear since, when plotted, it is not a straight line. YMMV.

It's always annoyed me that people confuse functions in common parlance. "Increases exponentially" when it's only quadratic, for example (referring to P = V^2 / R perhaps). Or "intensity increases exponentially as you get closer" (actually, inverse squared). Or trying to explain an hyperbola (like 1/x increasing towards infinity as x --> 0+) as exponential, when in fact it's increasing infinitely faster than an exponent.

I've been training myself to use more terms in conversation, and I'd like to think I've been more accurate; having to think about the relations in the system builds better mental models even as one speaks.

Tim

I would guess, about the same time one has to replace the capacitor. :^)

Tim

=2/rh

ie. surface area per unit volume *decreases* with radius and height (or length).

A simpler, more hand-waving, way of looking at it: Surface area of a cylinder increases as the radius, volume increases as the square of the radius.

That's why heat exchangers, radiators, tube boilers, etc. use small diameter tubes.

That's one of the reasons that George Stephenson's early locomotives were more efficient than the competition. Also how later Stanley Steamers could raise 500 (yes, 500) PSI. in a matter of minutes. Multiple small tubes.