Good capacitors

a Y5V cap makes a nice cheap temperature sensor.

I just finished a job on a customers smps. They had a lot of output ripple, so had tacked 9 x 10uF 16V X5R caps in a big stack on top of the original cap. I measured the output ripple, which was about 10x bigger than expected, so I measured diode current. There was a total of 8uF at

And that was without the caps getting hot. I didnt bother to do it, but a heatgun would have made the output ripple much worse :)

Cheers Terry

In article , John Larkin wrote: [....]

Unfortunatly unless you have a large market, you won't be able to prevent the maker from "improving" the capacitor so your design quits working.

Some also have a huge temp-co. You can make a surprising temperature difference to electrical power converter based on this. You charge them, heat them up and then discharge them.

Slick! CV is constant but C*V^2/2 goes up. I wonder what the conversion efficiency might be like.

Sort of reminds me of the rubber-band air conditioner.

John

Cell base-stations have problems with the linearity of *connectors*. Atomic corrosion layers or bad plating can cause unacceptable channel interferance.

John

That's how my 200 THz diodes work--20-25 angstroms of nickel oxide between nickel layers. Metal-insulator tunnel junctions are *quick*.

How can that rectify? Where's the asymmetry? DC bias? When can I buy an imaging 200 THz stud finder?

John

there was a paper in the last IEEE trans. power electronics, on a capacitive generator. charge it up with plates far apart, mechanical energy forces plates together, discharge it.

Cheers Terry

Good questions.

There are two sources of asymmetry: asymmetric junctions(*) and dc bias.

The I-V curves of these junctions looks vaguely like

V 0.1*tanh(V/100 mV) I(V) ~ ---------- - ----------------- 100 ohms 130 ohms

Given an I-V curve, you can calculate the responsivity in a few lines of algebra, using the Taylor formula:

d^2 I / dV^2 dI_out/dP_in (V) = ------------ 4* dI/dV

where one factor of 2 comes from the 2! in the Taylor formula and the other from integrating sin^2(omega t). For our junctions, this peaks at about 80-100 mV of bias, reaching ~1 A/W, which is quite respectable if it's still true in the IR. (Indications from the literature and our other measurements suggest that it is, but I can't prove it yet.)

We should have some complete waveguide integrated devices to test this week, assuming Murphy remains on vacation, and if so I'll post the results here.

Cheers,

Phil Hobbs

BTW: You'd be much better off with a 200 GHz stud finder. 200 THz is too close to daylight (green light is 600 THz). It's too close to daylight for me too, sometimes--metals are much better behaved at lower frequencies.

(*) The asymmetry comes from having different barrier heights at the two metal-insulator interfaces--for our junctions the splitting is about 20 meV out of 200 meV, due basically to the fabrication history--you put down some nickel, oxidise it in a very weak O2 plasma (like 8 seconds at

For some reason, this reminded me of an article in PopElec about 1000 years ago, about "Ovonics", named after this guy:

He apparently invented some device with insanely fast switching speeds, that consumed no power, or something like that. I saw a drawing of his device - a littel flake of amorphous Si between two wires. I have no idea how it worked, but seem to remember it had to be cooled to superconductor temps.

Or, I could be thinking of the "Josephson junction" - I'm sure someone will enlighten me on that. :-)

The thing that kind of surprised me at that website is that it seems he claims to have invented the NiMH battery!

And what ever happened to the cheap, long strips of amorphous Si photovoltaics?

Thanks! Rich

The other way around sound more logical. Charged plates attract, pull them apart takes mechanical energy and adds electrical.

This is the oldest generator principle used I suppose...

Thomas

That's how a Van de Graeff generator works--except that one of the plates is a rubber band.

Cheers,

Phil Hobbs