Auroratek demonstration from Bill Alek at TeslaTech conference

[TinselKoala]

Both channels of the scope are AC-coupled.



Is there a scoposcopist in the haus?

[MarkE]

Well Mark you can have high coupling factor indeed!

But first I want to point out, measuring magnetic circuits employing ferrite is a b*tch: changes in temp, slight mechanical shock and differing drive level (changes in applied H-field due to changes in number of turns) add much experimental error.

Now on to my little saturable reactor / magamp. Note that, for this circuit, tight coupling was intended for experiments exploring parametric variation (such as frequency conversion and AM modulation of an RF carrier). The tighter the coupling the less control winding current is needed to cause a change of inductance in the counter wound secondary circuit. Fundamentally the circuit is a variable load inductance programmed by a voltage controlled current source. All that said, the magnetic circuit closely resembles the primary half of Alek's transformer. At a later date, I'll put up some measurements of a full equivalent circuit, when I can find a CCM to use for the other half of his secondary arrangement.

What's interesting about this circuit is that the primary is shared across two separate and isolated cores, which is exactly how Alek's primary circuit is wound. The only major difference is that of winding style and obviously the lack of the CMM secondary arrangement--which I still feel doesn't contribute to large secondary inductance as you propose.

On to the measurements and Pics. First I want to say that both methods of determining k (shorted secondary and measured mutual inductance) were undertaken with the utmost care, all values are derived from measurement and not calculation, unless required. As is generally know to be the case, the mutual inductance method gives erroneous values, this is due to the greater number of ampere-turns that excite the core which shifts the incremental permeability to differing values, in this case greater u_i. Thus k is calculated as being greater than one, which is an invalid result. The arguably more useful shorted secondary test gives a a very high coupling coefficient of 0.99879.

I am good with determining K by measuring the inductance of a given winding first with all other windings open and then successively shorting one winding at a time. 

.999 coupling is a very high number especially with as few turns as are visible in your photos.  0.96-.99 seems more plausible to me.  Your structure is quite different than Bill Alek's.  In the case of your structure all windings closely hug one or the other of the toroids.  In Bill Alek's case, each winding extends far beyond any of the given cores.  I expect that to result in substantial leakage inductance and therefore K way below .99 that is achievable by tightly wrapping all windings tightly around a single core.  Flux transfer between the various cores is by way of induced currents in each of the windings, including through the copper losses of each.  To me the resulting reluctance will be much higher than where the flux couples onto a single closed high uR core such as a toroid of appropriate material.  The only way to really know for certain is to either build up 3D field model, or measure the device properly.  I submit that if you were to lay your two toroids flat with the openings facing up and wind one winding on each core and a third winding shared between them that you would be much closer to Bill Alek's arrangement, and your leakage inductance would sky rocket.

Concluding, I think we can safely say his primary circuit isn't a major contributor of leakage inductance. However, large leakage in the secondary extraneous circuit is still on for debate. I'm thinking it's not going to be very large though, seeing as how the split secondary fields oppose one another and have the same turns ratio, they will likely have very low (possibly <200uH) leakage inductance.

A few tens of uH in the clipleads and WW resistors at 3kHz is still small compared to the stated winding resistances.   I think that is potentially an error source of a few degrees but do not believe that it is the elephant in the room.  I disagree with your conclusion that his windings are tightly coupled for the reasons stated above.  I expect that the coupling is weak and that is the major source of phase shift.  Of course Bill Alek could measure his device properly and then we would know what the situation is without dispute.

[TinselKoala]

@Garrett: Isn't a coupling constant in excess of unity one of the ways that "OU" can sneak into a circuit? By dismissing this result as invalid you are semi-circularly rejecting the possibility that the measurement that leads to the "anomalously" large K is actually telling you that the FE is coming in here.



Great work by the way, mucho respeto.

[TinselKoala]

"I am far less concerned by the clip leads than others. "

Ah, but you are a professional, or at least do a good job of acting. SO you have professional-grade clipleads. Many cheap Chinese clipleads look great on the outside but if you examine them closely they may have only one tiny strand of wire inside a big sheath of plastic insulation, and a crimp that doesn't even bother to strip the plastic, so it can be high-resistance or even intermittent. The tiny strand of wire probably has higher inductance than, say, a fully populated length of #20 stranded copper of the same length.

And the alligator clips suck, especially after you step on them a couple of times.

[MarkE]

"I am far less concerned by the clip leads than others. "

Ah, but you are a professional, or at least do a good job of acting. SO you have professional-grade clipleads. Many cheap Chinese clipleads look great on the outside but if you examine them closely they may have only one tiny strand of wire inside a big sheath of plastic insulation, and a crimp that doesn't even bother to strip the plastic, so it can be high-resistance or even intermittent. The tiny strand of wire probably has higher inductance than, say, a fully populated length of #20 stranded copper of the same length.

And the alligator clips suck, especially after you step on them a couple of times.

I try to keep the clip leads away from the stampedes.