capacitor heatsinks

Well, let's see what can be done with what is currently available.

I'll aim for about 10,000 uF at 50 v. Picking something at almost random, I find 100 uf at 50 v at: That will require 100 capacitors.

The case size is 10.0 x 12.7 x 11.0 mm h. Since there has to be some clearance for the leads, I'll stagger the rows, which will increase the width from 10.0 to 11.5. Therefore the PCB area will be about: PCB area = 100 x 11.5 x 13.0 = 150.5 square cm. Volume = 200 sq-cm * 1.1 cm = 220 cm^3 (not including PCB) Cost = $1.80 * 100 = $180 (ouch) plus the PCB, soldering, connector, etc.

By comparison, a commodity 10,000 uF 50 v electrolytic at: Cost = approx $5.00 Volume = 25 x 25 x 50 mm = 31 cm^3

So, the polymer capacitor would have 5 times the volume and 36 times the cost (not counting the PCB). Sigh. I don't think it's polymer is going to work unless absolutely must have low ESR at low temperatures.

Well, there's always tantalum capacitors. Lets's see where that lands:

"miso = moron "

.... Phil

"John Larkin"

FYI Some cap makers app notes have this way underestimated, do your own test.

BTW:

Excessive ESR causes an electro to heat itself, solving the low temp ESR problem.

.... Phil

"John Larkin"

There is no heat source at the base of "tube".

The filament heats the cathode to a red colour by conduction and radiation, then the cathode heats the plate by radiation and the glass absorbs most of that heat.

You obviously think there is - so please explain.

... Phil

We thought of that, but the warmup time constant is minutes. This particular application has to be ready to go on the first shot.

It's a discussion group. There's no more requirement to stay exactly on topic than there would be amongst a bunch of guys downing beers at a bar.

Nice word, "amongst." You British speakers do have some nice words.

That makes no sense.

Wow, those caps would be a wonderful way to get planned obsolescence. And lawsuits.

Cheers

Phil Hobbs

Phil Hobbs schrieb:

Hello,

I have seen special power condensators in an inductive heating electronics. These condensators were mounted on a water cooled plate, the IGBTs also used water cooling. The condensators had the size of a large coffee mug.

Bye

The envelope is pretty thin, maybe 1 mm, so that's a possibility at the short wavelength end. Quartz has OH absorption peaks around 2.7 um, and drops off like a rock around 3.5 um like most other oxides. (Sapphire is an exception.)

The power spectral density peak for an object at 600K is about 8.5 microns. I ran the Planck integral for a 600 K blackbody, and with a 3.5 um cutoff, the glass transmits only about 8% of the total IR. So no, the glass is pretty well completely opaque until the plate gets up to dull red heat--the transmittance reaches 50% with a 1200K black body.

Probably the main difference was inhibiting convection near the tube envelope, as you say.

Cheers

Phil Hobbs

John Larkin schrieb:

Hello,

I have seen special power condensators used in inductive heating equipment. The do also larger capacities and voltages with polymer. But electrolytic cpacitors allow for larger capacity values.

Bye

Never mind, they're the wet type. Probably not such a huge improvement over aluminums, then.

Cheers

Phil Hobbs

At work, we regularly use the big metal tub kind (good for a few MVA). They put water cooling pipes inside the things. Smaller units (coffee mug and smaller) are usually "conduction cooled", meaning you need water on (or in) the bus bars they're bolted to. And some conductive heatsink paste wouldn't hurt.

Tim

It is a mistake to attribute the problems of molded case dry tantalum capacitors to the wet slug type. Only wet slug tantalum capacitors are permitted as bulk capacitance in circuits requiring the highest reliability (notably satellites) because no other type of capacitor available in similar values comes close to the high reliability delivered by the wet slug tantalums. The extremely high price is likely mostly due to meeting mil-spec for high reliability; a lot of testing and documentation expense for a very low volume product. AFIK there is no significant commercial market for them.

Apparently they're also very low leakage (one broken in). I recall a Jim Williams / LT nanovolt appnote using them.

Tim

silly, except as some hack retrofit, or to fill a catalog.

they've been making inverter grade (hot new term for "computer grade"?) caps with a bottom stud forever. The bottom is designed to be a heatsink and a single stud must save serious assembly time over those silly 3 footed clamps.

Any idea why the heatsinks trashed the life of the tubes?

Nothing definite. My best guess(tm) is that the thermal connector spring, between the glass envelope and the outer shield, blocked the normal convection air flow. There's no way I can measure the heating at this point as I don't own anything that uses vacuum tubes or have any such head sinks.

Doesn't sound like you've ever tried to integrate a stud-mounted component into hardware.

RL