|
Hi Matt,
--Yes, it was a weird experiment, Matt. Nobody seemed to think it was too significant at the time because it was widely believed that either the switching noise was responsible, or the energy output was 'too small' to be bothered with.
I don't know if residual magnetism was needed to get it started. A residual noise current would be required for sure. No spontaneous para oscillations will start without it.
I don't know whether the gain is limited or not. I list three loss mechanisms below that exist in these switched C or L circuits, and if these are eliminated, then there should be good results.
I agree that resonance is important to maintain ultra-efficiency throughout, but this isn't the same thing as para amplification. This kind of reactance power amp. can take place without resonance, in a switched circuit at 60 Hz, but is made much more useful with resonance because energy is recycled and re-subjected to the parametric step.
Resonance is its own highly important thing, and my adiabatic list is grappling with this subject right now. Everyone knows that resonance improves efficiency, but Why?
Yes, not only is noise composed of an infinite number of frequencies, but some sort of broadband tuning seems to be possible. The noise energy is limited by the bandwidth that the noise receiver can absorb. Or, and this is probably much easier, you can tailor a noise source whose energy is concentrated in a certain band. I think circuits with 'tickler' elements are useful here, to excite the diode, or transformer core, or whatever noise source is used, to make it generate a lot more noise, with no net reaction on the ticker current or voltage, since noise is equal in EMF in either direction. For instance, there are probably some noisy diodes that generate a lot of excess noise when they have a reverse bias. Why not 'stir up' their noise levels while trying to tap them?
A later experiment on a switched capacitor quasi-parametric circuit was done in 2001. I didn't take a lot of notes, but this is the upshot. A 555 timer was used to connect and disconnect two capacitors while in a tank circuit. The switching was done with a Hexfet opto-isolator unit with very little switch noise, either in theory or test.
The results were highly anomalous. It turned out that the absolute value of the capacitors and the ratio of their values determined the intensity of the output. Unfortunately, I've lost the discussion of those cap values, since that computer bit the dust. I do still have a brief report on the test results. A 5-? mV ringing pulse with an internal frequency of about a Mhz and a repetition freq of 8.3 Khz was seen across a 100 ohm load, when the tank was resonated at 100.6 Khz with a tank frequency of 50.3 Khz.
According to parametric oscillator theory, we would see oscillations at the tank frequency. We didn't see that, but we did see these seemingly unrelated pulsations at other frequencies.
This only made sense if the ringing pulsation was a weak subharmonic of some very high harmonic-- if that makes sense :-)
He estimated this to be around 33rd harmonic from his knowledge of music.
In a black box thought experiment, an LCR meter will register the same change in C at the terminals of the black box, whether there is a switched capacitor set or a biased varactor in the box. The 'parametric amplification' is the same in either case. But if you put a voltmeter or ammeter at the terminals you will see they're very different. So I theorized from this that parametric change does happen with switching, but that other parts of the process interfere with this, and create losses. In the usual case, the loss is exactly the same as the gain, which is why every switched capacitor or switched inductor circuit is not overunity. The possibility of getting energy from switching came from reading the Barrow paper I'll be posting here today where he shows that a lot of these loss mechanisms can be designed out of the process.
The losses in the switched tank circuit are very large compared to the parametric one. They can be listed from most obvious to most obscure:
1) The disconnected L or C component may actually carry circuit energy in field or charge, and this energy may not be returned to the circuit during the rest of the tank cycle. This relates to your question about residual magnetic energy-- does it take a little bit to restart the process again? This sort of loss can happen especially with the inductor version, because when it is reconnected to its brother inductor, it has lost some of its field energy.
2) The disconnected component may be returned to the tank circuit where its own polarity or current is opposite to that of the tank, thus neutralizing the tank energy. This can be seen when the inductor is reconnected to the tank when it is opposite in flux to the other inductor, or opposite in EMF to the other capacitor. There is a cancellation of energy just like putting two batteries positive to positive.
3) The discharging component has a built in thermodynamic loss that happens when one reactance is discharged to another. This is known from the two-capacitor paradox. If you take two capacitors, one starting with a known voltage, and connect this one to a discharged cap, the final V of each capacitor after charge distribution will be about half that of what it started with, and as a result, the total energy stored in the capacitors is reduced by half. A lot of fancy physics has gone into explaining where that other half went!
Because of these known and hidden factors, the switched capacitor circuit is incredibly lossy. If you took an actual saturable reactor and replicated the L changes in JLN's experiment you would see much more output, but at of course a much higher cost in energy to drive.
I basically think, and centraflow and ION can differ with me on this, that something like this is what is going on in their circuits. I'm going to spend Monday studying both their circuits in detail to make sure I am not comparing apples to oranges.
In doing experiments, remember that parametric phenomena can be finicky to develop in the best of times. It takes patience. (The standard parametric transformer is not hard to replicate). As I mentioned above, I know that in the later capacitive experiments the ratio of the two caps used was not 1:1, but really not sure what it was, or if this is also true in the inductive case.
Sorry for the length of this letter, I've been thinking about this weirdness for a long time, and the thoughts have kind of built up :-)
orthofield
Matt said:
To be honest orthocoil, I'm still very confused and intrigued by the experiment you did with J.L. Naudin. I cannot imagine for the life of me how switching two coils with no input power develops any sort of output. Is residual magnetism required to get this process started? Does this appear to be a gain limited phenomena or are you inclined to think it is scalable?
I will have to do some experiments to get my feet wet, because it's one of those things I'd have to see it first hand to believe it. The way my brain is wired it seems completely counter-intuitive.
I do understand the tank circuit area you listed. I've done a fair amount of study of Dale Pond's Sympathetic Vibratory Physics to know that very tiny oscillations can cause a tuned circuit to begin oscillating with considerably higher amplitude just by being in proximity. When you dig into this rather deeply, what is found is the "noise" isn't pure white noise, it's instead known as pink noise because it contains patterns the resonant circuit is able to tune into.
[/quote]
|
Author
Topic: Parametrics, Noise coherence, and Switching (Read 29084 times)
