As already pointed out, the core is tape wound and cut into a type C config using a metglas alloy; most likely type 2605SA1, which is an iron based alloy used in power transformers.

At the moment, I wouldn't be too concerned with the core alloy; though admittedly, the lower loss the core, the more likely you would be to obtain Alek's results (if they are possible in the first place).

I spent a couple of hours last night trying to figure how Alek's device could work using normal circuit theory.

After some refresher reading from Terman's Radio Engineering 3rd Ed. pgs 55-58, it would seem that the complex impedance reflected from his common mode choke segment is causing the phase angle of the primary circuit to shift in the opposite direction than would happen otherwise. Basically, Terman points out that if the secondary current is 30 degrees lagging, it will reflect to the primary to be 30 degrees leading, so that it will partially cancel out the primary's inductance. Since the coupling coefficient can't exceed 1, the best you could hope for is completely nulling out the primary inductance. That is, the primary becomes completely resistive, as you would expect. However, in Alek's transformer, the power factor decreases when the secondary is loaded, which means the reflected impedance is making the primary more reactive--which isn't normal. When shorted, the vector sum of the reflected load impedance and the primary's leakage inductance and ESR add to an angle >90, which means that not only is the secondary paying for the primary's resistive losses its also driving the AC source like a load.

Certainly sounds too good to be true. At any rate, I'm going to throw some functions together and see exactly what load impedance and coupling factor is needed to simulate his results. Which will give some insight into how the common mode choke section is performing in the circuit. No need to build something until you understand how or why it could work.

To be continued...

Hello,

Auroatek measurement failures: see attached screenshot  edited by me.

Even Naudin had to find out the selfinductance of power-resistors.

The two black resitsors used in this demonstration are very old resistors with
resistance-wire applied on a ceramic tube -> selfinductance

Absolutly a no-go for power-measurements in the range of 3.2 KHz

Next: if my assumtion is correct then these two Digimeters at the left
are only designed for AC and never will show  correct values at 3.2 Khz

Why does he not use his scope on the output-resitors ?

So what`s the value of all this demonstration and calculations base on wrong
values measured ? A waste of time

Regards

Kator01

Quote from: G4RR3ττ on August 06, 2014, 02:04:23 PM
As already pointed out, the core is tape wound and cut into a type C config using a metglas alloy; most likely type 2605SA1, which is an iron based alloy used in power transformers.

At the moment, I wouldn't be too concerned with the core alloy; though admittedly, the lower loss the core, the more likely you would be to obtain Alek's results (if they are possible in the first place).

I spent a couple of hours last night trying to figure how Alek's device could work using normal circuit theory.

After some refresher reading from Terman's Radio Engineering 3rd Ed. pgs 55-58, it would seem that the complex impedance reflected from his common mode choke segment is causing the phase angle of the primary circuit to shift in the opposite direction than would happen otherwise. Basically, Terman points out that if the secondary current is 30 degrees lagging, it will reflect to the primary to be 30 degrees leading, so that it will partially cancel out the primary's inductance. Since the coupling coefficient can't exceed 1, the best you could hope for is completely nulling out the primary inductance. That is, the primary becomes completely resistive, as you would expect. However, in Alek's transformer, the power factor decreases when the secondary is loaded, which means the reflected impedance is making the primary more reactive--which isn't normal. When shorted, the vector sum of the reflected load impedance and the primary's leakage inductance and ESR add to an angle >90, which means that not only is the secondary paying for the primary's resistive losses its also driving the AC source like a load.

Certainly sounds too good to be true. At any rate, I'm going to throw some functions together and see exactly what load impedance and coupling factor is needed to simulate his results. Which will give some insight into how the common mode choke section is performing in the circuit. No need to build something until you understand how or why it could work.

To be continued...

His transformer has weak coupling:  K << 0.99.  This makes it look like most of the primary inductance is isolated, resulting in a large phase shift.

Kator01,

FYI, I'm a test equipment nut--seriously--and if you were too, you would know that most DMMs are good up to 100kHz bandwidth... Most go to 300kHz (most hand held Flukes) or even 1MHz (high end HP/Keithley bench meters), so how does your point about 3.2kHz signal frequency have any validity? Further the harmonic content is likely less than 3% (IMD and hysteresis), so implying that there is substantial energy content in harmonics is just inane.

Further, if you knew anything about inductance, you would realize that the smaller the diameter of the former the greater the opposing mutual inductance, counter currents become closer. Also, the greater the spacing between turns the lower the aiding mutual inductance, aiding currents can't couple as well. Ceramic resistance coils exhibit both of the aforementioned phenomena: wide turn spacing, small diameter coil former. Ironically, in most cases, a straight wire has more inductance than a single loop or a loosely wound small diameter coil! Do the measurements!

Point is, the long connecting wires on the secondary side probably have equal or more inductance than the power resistors, but I never heard you point that out... But, point well taken. Inductance of the secondary circuit contributes to the total secondary load impedance.

I doubt the power factor on the load side is a major detractor here. Particularly in in light of the fact that the major component of the complex impedance is resistive. Assume the stray inductance is equal to about 5uH, at 3.2khz this equates to 100 milliohms or 0.1ohm. I'm not seeing how this materially affected the power factor. Now let's say the stray inductance a bit higher at 100uH, we get 2ohms of reactance which might start to actually have significance. When we find the impedance however, 11.39V / 0.459A, we get about 24.81ohms of impedance. Solving for R we get about 24.72ohms... The phase angle (arcsin (X/Z) = theta) should be close to  4.62 degrees... And your point was that our new correct power measurement of 5.2109W_real is significantly lower than the 5.228 as calculated by Alek. Right. Real good job on the math.

Feel free to point out any errors, or misunderstandings on my part.

Mark,

The secondary magnetic circuit is a COMMON MODE choke, meaning it doesn't contain differential mode inductance! That is, the magnetic fields of each winding null out to leakage levels. Assuming I understand how both common mode chokes work and Alek's circuit, your point about k<<1 might be off. Of course it will be lower than a normal transformer because his windings do not completely enclose the primary core section--there is open space. BTW I've built magamps that use very similar construction to Alek's transformer and achieved rather high coupling coefficients. If you want I can take some pics and post the measurement results.'

Also, your point about the coupling coefficient is wrong, in the sense that regardless of its value, the load impedance is reflected back to the primary. Thus primary inductance MUST decrease when the secondary is loaded, thereby lowering the primary impedance and appearing more resistive--however slightly this may be due to weak coupling. Why Alek's transformer does the opposite is the real question.