Magamp oscillator (2023 Update)

The 'set' and 'reset' reactors are, in their extreme control settings, biased to allow current flow in one direction only. They require a non-reversible current during their control phase, or in a coupled control winding which is only really practical using a synchronous switch or rectifier.

One maintains unidirectional current to keep the part in saturation prior to application of reverse voltseconds, the other applies restting voltseconds only, to avoid conduction in the forward phase.

Energy is derived for oontrol either from the source input, or the (un)regulated output - for wider control (or blocking voltsecond) range, both control methods can be applied to the same part.

Kotlarewski (sp) in Ottawa published a definitive article on these in the 90s.

A certain amount of self-maintained bias can be induced by placing hard magnetic material in series with the flux path, but these tend to be mechanically difficult to construct, short-lived,lossy and of low relative permeability.

RL

A rectifier, with an AC source, produces DC EMF; but any magnetic field that produces in a winding a DC EMF, must have

d(flux)/dt ~= dotproduct( (area) ,dB/dt ) >0

in a net average over time. That's a ramp in B that never every decreases: you can not do that, over any sufficiently long period of time the B field will exceed your magnetic material limits (or current supply capability, or material strength limits).

The best you could do, is with some moving-parts switching to reverse a winding direction (the commutator on a DC generator reverses the winding polarity); so an AC motor/DC generator or other 'rotary converter'. That doesn't rely on the dB/dt to be always-increasing, it changes the orientation of the 'area' and flips the sign in the dot product so that the product is never negative.

NT

Pre-tube, big radio transmitters, hundreds of KW things, used negative-resistance arcs to make true oscillators.

Not to be confused with spark gap transmitters, that just made ringing oscillations.

Am not so good at designing with nonlinear components. Know of the trick using nonlinear inductors in a discrete L,C PFN to get faster risetimes; not sure how helpful that maybe in this case. First i need to find nonlinear inductors/transformers that do not need excessive power to act that way. And the play with them. I do know that as the excitation is increased on the Ma Bell ring generator, that there is some point that it starts to become unstable, and finally (as an oscillator to run at 20Hz) settle down in the preferred mode; that info is not helpful. If you have other clues, please let me know.

Humph. "Maggie" was the term used in radar as a short term for the magnetron..

Saturating magnetics can do cool high-power stuff that would destroy semiconductors. Like the nonlinear delay lines you mention. I've seen some biggish eximer lasers that used saturating magnetics to make the big pulse drives into the discharge tubes. Very rugged.

Do you mean the old central office ring generators? There is very little online about them, but they seemed to be rotating motor-generators mostly. The CO stuff did all ultimately run off 48 volts DC.

Do you know of a non-rotating, non-semiconductor, DC powered ring generator? Bell was pretty good with magnetics.

Even after tubes were available, the spark-gap oscillator was a standby of diathermy/induction heating. Crude, effective, and in a welding shop you need to ventilate against other gas buildup, so why not against ozone?

I think you're correct. I saw this: which seems to say that the subharmonic mag-amps were energized by utility 60Hz power and DC rotary convertors handled power outages.

US patent US2452239 shows a local application but again the ringing generator is pumped at 60Hz.

piglet

Many thanks for clarifying - makes mag-amps seem less terrifying when I can relate them to varactor para-amps!

piglet

Right. Core saturation creates a nonlinear element that can do things, but it needs an AC pump.

Here is a diode acting like a subharmonic oscillator. Same sort of idea.

Quick proof-of-concept hack. Could be optimized.

That happens in active peak detectors too, if you get the time constants wrong.

Cheers

Phil Hobbs

You brought this up some time ago when I asked about passive circuits generating lower frequencies from a single higher input frequency. (The equivalent of what a BBO crystal does with light.)

But this doesn't work. What you're really seeing in this circuit is just the switch-on transient response of the tank, and which dies out pretty quickly.

Jeroen Belleman

Phil - do you know of a source for tunnel diodes? Preferably less than $100.

Hul

Phil Hobbs wrote:

In active circuitry, sure. I'd like to see some passive circuit that does this. Non-linearity is not enough. You need a bifurcation in the phase space trajectory.

Something with a tunnel diode would work.

Jeroen Belleman

You can get back diodes, which are similar but not exactly the same as TDs for oscillators and triggers.

My stash all came from eBay. Most of the old-style TDs had gross amounts of capacitance, like 200 pF. If you want speed, you have to find diodes with 100-200mA peak currents. Most of mine are only a few mA.

Cheers

Phil Hobbs

You don't think that a 1N5819 and a tank circuit is passive?

Cheers

Phil Hobbs

Lots of people make "back diodes" which are tunnel diodes with low Ip currents. They are really designed to be RF detectors, although they do have a negative resistance region.

Somebody (GPD I think) was making the original GE Ge TDs for a while, the fast switchers, but they may be gone now. I can't find them.

These folks sell TDs, but I don't know if they are old stock or new manufactured.

I don't understand your point. The waveform is periodic with twice the period of the fundamental, but not at the fundamental frequency. How is that not a subharmonic?

Degenerate parametric amps (pump = input) can also generate subharmonics--you can buy packaged "frequency halvers" that just use varactors. See e.g.

Cheers

Phil Hobbs