Rectification

Perhaps, but why should anybody care? There are Gunn diodes which do odd things at very high frequencies

It's not so much a theoretical question as a theory-free question which demonstrates that you don't know enough about electronics to be able to know which questions are worth asking.

I would have thought, if the load was resistive, you would just see the A.C. waveform, as the self-capacitance of the diode swamped out all the other effects. If you have a smoothing capacitor following the diode, that would effectively be in series with the diode's self capacitance and the two of them would act as a capacitive divider which progressively shorted out the power supply as the frequency increased.

Yes, indeed, I'm sure a tank cap would be a short at relatively low frequencies so I didn't envisage one for this particular mind experiment. I'd like to try it in Spice but I don't think the available diode models are up to the job.

Bear in mind that there are a lot of different types of diodes and some are used as mixers up to SHF, so you might have to simulate low infra-red before you noticed an non-diode behaviour with them. Something like a 1N4001 would be more manageable.

Eventually the ESL of the diode will series resonate with its capacitance, and beyond that it's an inductor.

By that time, it's not much of a diode. In a PN silicon diode the carriers can't move very fast so it starts to look ohmic at high frequencies. PN diodes have both reverse and "forward recovery" time delays.

I've used SiC diodes for fast high-voltage things and they are much better than silicon.

It is an interesting problem, how to make a power rectifier at very high frequencies. It's been proposed to put solar arrays in orbit and beam the power down as microwaves, into "rectennas" on the ground. They must have some sorts of diodes in mind.

The step recovery effect is cool too. Look up "drift step recovery diode" aka Grekhov diode for some other interesting effects. I made one thing that forward biases a power diode at +50 volts for a while and then reverses it at a couple hundred amps. I didn't sell many but it was fun.

You can get very nasty spikes from reverse recovery SRD effects, even at 60 Hz, or 50 Hz in 3rd-world countries that have electricity.

Cool. Like everyone else here I suppose, I have thousands of diodes of all kinds, some more easy to identify than others. I have the capability to characterize them for their different high speed attributes, but doing so *properly* would take up huge amounts of time and sadly I just can't spare it. Life gets shorter every year it seems.

I've seen the 1N4001 used as a varactor diode. It's not tightly specified for the role, but it didn't need to be in the application, where it balanced a capacitor bridge following the small alternating changes in a capacitance generated by small alternating changes in the pressure across a capacitative pressure gauge in a vortex shedding flow meter.

It would take even longer for you to learn enough to be able to classify them properly.

Step recovery diodes are fun, as John Larkin points out, but even if you spend the money to buy one that is properly characterised for the task you have to wait a while for the snap-off edge that you want and the stored charged that works the magic isn't all the well-defined.

Pretty good for generating a comb, though. I would guess there are more varieties of diode than any other discrete component there is. Quite a remarkable little device and - usually - trivially cheap.

"Generating a comb" relies on the fact that a Dirac impulse - of finite area and zero width - would generate all the harmonics of its repetition rate up to infinity. Step recovery diodes generate narrow pulses and the upper limit to the harmonics they generate is set by the width of the spike.

A few people still make official SRDs. Metelics/Macom for one. Some of their parts are under $1. Mostly used as RF multipliers. The really fast impulse generators nowadays are NLTLs, shock lines.

For making steps, nowadays one can buy ECL gates and comparators with ballpark 35 ps edges. There are some cheaper laser drivers that will make a fierce differential output with 25 ps edges.

I wonder if it would make a noiseless parametric amplifier for audio frequencies?