If we had an extremely wide area, extremely thin gap capacitor, its capacitance would be huge, but maybe the resistance of the dielectric in between would become sufficiently reduced to allow noticeable conduction to occur even with very high resistivity?

Yes this increases the capacitance, and it's even the only way to make capacitors usable at low frequencies.

The first limit, when making the dielectric thinner, is that it may break down at a voltage too low for your application. This happens before the insulation resistance gets too low (and insulators don't behave linearly anyway, so one can't really define a resistance).

If you use the capacitor under a low voltage, say 1V or 2V, then you can use a really thin dielectric, but current will flow through it by tunnel effect.

Both effect (breakdown and tunnel) are routinely observed, as dielectric layers really are that thin to make capacitors miniature. Not with a laminated plastic film is such cases, but for instance with oxide layers grown by electrochemical means on Al, Ta or Nb.

On silicon, tunnel effect was a limit to thinner gate insulators. See "High-K dielectric".

To increase the surface, electrodes can be thin and stacked or rolled. On may also sinter an electrode made of tantalum powder and then corrugate each grain further by electrochemical processes; after growing the insulating oxide, an electrolyte makes the other electrode.

In C = K*S/d, K is only 8.8e-12F/m, but we have supercapacitors of several F the size of a hazelnut. Have a look at what it means about S and d, it's surprising.

Thanks for the information. Also thinking, at high frequencies, the impedance should also decrease inversely with capacitance and frequency as well, right?

This is right at reasonable frequencies.

The best leaded HF capacitors have 4nH stray inductance, the best surface-mount capacitors about 0.5nH, so at a few 100MH or GHz they get inductive instead of capacitive, unless the capacitance is chosen very small (few pF).

Losses also increase with frequency. Most aluminium capacitors are more lossy than reactive as soon as 1kHz.

As of now, the EEESTER seems to be the most promising super-capacitor technology for auto application... over 40 Farads, 3500 volts in it's present form, with up to 10,000 volts of charge allegedly possible via the technology. I’ve followed the eeester story in passing for a couple of years… eeester was pretty quiet, and last I had noticed NASA had interest in the technology for possible use in space probes (obviously, an intersteller space environment would render operating temperature ranges moot.) Then, it was recently announced that a car, the (Chinese-built?) Zenn, will be powered electrically using only the EE Ester, and I thought “Wow, it’s for real.”

But a few questions come to mind.

Assuming the eeester works as claimed, what’s irked me all along was the power conversion strategy they intended to use (I only found that out recently in reading about the Zenn.) It’s a buck-boost converter.

The buck-boost converter would seem to be the most simple, direct, and straightforward approach to the problem of adjusting output from a capacitor charged with great pressure, yet this strategy would have a poor duty cycle due to parasitic resistance… thus, a degree of inefficiency would seem unavoidably inherent to such a design. Now, to get even nominal performance out of even a compact car will require between 50 KVA and 100 KVA of power (converting from horsepower, assuming efficiency to be fairly high.) If a 12 volt DC motor were used, very large currents would be necessary to produce the electromotive forces required by a compact car.

The alternative to such would seem be a higher-voltage AC motor (perhaps 120 or 240 VAC, 60 Hz) with the conditioning/inverting power supply consisting of either (1) a power transistor array, or (2) vacuum tubes. Either option could be microprocessor-controlled to invert the DC capacitor output to AC, while improving system efficiency while continually monitoring loads, yet the second option impresses me as possibly being more practical, even though vacuum tubes have become antiquated for most applications.

First, I’ll summarize the possible obstacles of using a power transistor array, as I see them. The most robust semiconductors commercially available can handle about 1 KVA. Thus, it would take (at minimum) 50-100 of the largest power transistors readily available- operating in tandem- to meet the aforementioned drive requirements of a small car. Constantly varying high loads, in addition to harmonics which would be introduced into such a system during city driving (in addition to power regeneration strategies proposed for braking to recapture the car’s momentum as current during braking) produce the kind of conditions that make semiconductors vulnerable to damage and failure. Longevity or reliability could initially present design challenges. Such a power conversion system would necessarily be complex, it would require cooling (probably liquid) and it would likely be relatively expensive to produce.

Suppose vacuum tubes were used. The disadvantages would include perhaps 70% efficiency, and a need for water-cooling (likewise for a semiconductor array… similar to a radiator and water pump as on gasoline engines.) 2 EESTERS could form a +/- 3500 VDC supply, the control grids could be semiconductor controlled via a microprocessor to provide 120 or 240 VAC with supplemental conditioning components, for instance, which could drive an “off the shelf” motor of high-reliability using far less current. Perhaps with an array of 2,4,or 8 pentodes, the power being directed to the pair most compatible with the charge-state of the capacitor. The microprocessor could be interfaced to monitor loads and adjust grid potentials, and which tubes were powered to maximize efficiency.

The advantages to vacuum tubes in this application, as I see them, are:

1. Even high-power tubes are compact and light.
2. Due to their design simplicity, they can be manufactured very inexpensively.
3. Tubes are virtually impervious to damage from shorting, overloading, and harmonics.
4. Tubes can handle 10 megawatts or more, with those rated over 10KVA water-cooled.
5. The technology is simple, and has already been developed to an advanced stage, with the operation theory very well-understood.
6. Even very high-power vacuum tubes are compact, light, have few internal parts, are reliable, and can last up to 10 years.
7. Manufacturing methods, tools, and instruments for commercial production already exist and have been highly-refined.
8. Their physical designs can easily be made highly-durable… for example, a high-strength polymer enclosure with inner ceramic or metal cylinders, in a rectangular enclosure including channels for coolant flow would be one possible configuration.

Anyone have any thoughts on the matter?

3D-IC is a natural direction for the future. However, I find issues with the logic-memory combination. One would have to have both types of chips ready at the same time and built to the same TSV packaging standard. Also, there is also a conflict of interest posed, e.g., between Intel and Taiwan's motherboard makers as well as the various memory manufacturers. So the barrier to 3D-IC is, for the most part, political.

Are you all kidding?

An inverter bridge with single transistors (Igbt in this case) handles about 6MW, which is the power of a locomotive.

And that's not in the future, it's right now - and even since quite a few years.

No one would consider going back to valves for power conversion. They are still used in microwave ovens, but less and less in transmitters for radiocomm and radar.

------------

Just to get a sense of what is feasible:

The Itaipu dam produces 12GW in three-phase AC which are converted to DC for the transport (1000km) and then back to three-phase AC in Sao Paulo. All with semiconductors. And this was done more than 20 years ago.