Over Unity Lightbulb

[elgersmad]

Could you show us photos? I don't well understand how is connected the quartz tube.
And how did you lead measurements to demonstrate OU? Can you present them?

It's not that hard, all you need is a couple of toroids and magnet wire.  Then you grap a coil winding calculator that gives you cores that you can find or buy on the internet.

At that point, you just work with single inductors.  For a transformer, it's important to load the primary.  So, from that point, you are due to make several measurements using a signal generator and an oscilloscope.
mini Ring Core Calculator this software works well to calculate the primary and secondary windings values of inductance on different cores when the secondary is unloaded.  Don't let the name fool you, a T400 Core can handle 1KW but, if you don't want it to heat up, I wouldn't suggest using it for more than 500 Watts.  Heat and core saturation are the enemy.  If I could afford metglas, I would probably be using it at an operating frequency of around 150KHz to 200KHz.  The core I'm trying to order right now maxes out at 150KW at 250KHz.  Most likely, the only way to avoid overheating the core and changing it's permeability is going to be running it at half power.  Those ratings tend to be made where the wire is getting hot, or the core itself.  They'll push them as far as the curie temperature will allow. Then I'll go to the trouble to order some doorknob capcitors, 3 of them, one for each stage, and a vacuum tuning capacitor, all of which are normally used in DXing, or Amature Radio Transmitters that operate around 1KW to be legal.  But, the parts are often rated clear up to what you would use for a 50KW AM Radiostation.  You can't get away with tiny little leads, and ask them to handle more than a single ampere of current.  I'm making a big lightbulb, not another tiny circuit that's just breadboardered at 50 or a 100 watts just to see the readings or compare it to a simulation that's scaled down to parts I have on hand.  Custom winding large coils, a single large coil by hand can take several hours just due to the stiffness of solid wire, and at this point, I'm not looking at simply theory, and have moved up to a bigger better design.

Where do you think I came up with this equation Pout = Pin * Q * Q * Q?

I tested the circuit, and when every stage is tuned to the same frequency, resonance and Q put that much energy in the final stage. The impedance of a parallel tank circuit, is at it's highest at the resonant frequency.  It's just that a typical simulator does not include an accurate transformer model.  The primary should still act just like a plain inductor when the secondary is left unloaded.  But, when you place a capacitor in parallel with the primary, 10 volts supplied at the resonant frequency does not produce 14.14 Volts AC.  If the transformer were a 1 to 10 step up winding, then you should have an output of 141.4 volts.  You can physically measure that.  This is all first year stuff if you've taken college level electronics.  Oscillators and amplifiers are second or third year if it includes biasing.  Then Bridge H MOSFET circuits.  Typically if you have a low value of XL, you need a driver that has an output impedance of half of that value at the resonant frequency just to get the thing to resonate at the peak voltage during operation.  When the impedance goes up do to the resonant state of the circuit it may be over 1K ohms.  But, until then it's not.  Yea, it will still resonate but, it will just suck all of the power out of the driver circuit as if it couldn't reach the supply output voltage or peak out 1.414 times greater.

Circuit simulators are free, SWCad is out there, and there are even tutorials that include how to correct for transformers on the internet.  LTWiki.  They do not deny that the simulators are not producing real world results from transformers, so you can't see the kind of results that would allow you to choose the right diameter or bundle of litz wire.  The simulators are cheaper if you learn to use one.

[elgersmad]

ltWiki This article defines how to simulate real transformers using LtSpice IV, which is a free circuit simulator.

I use B² Spice, and I have to fix my simulator's models to make them work right too.

Transformer Simulation with B² Spice

They don't write these articles because the simulators work right.  Typically, you're looking at an engineering shortcut, and all that the model transformer does is use a turn ratio to multiply or divide a voltage, and multiply or divide a current.  If the voltage is multiplied, the current is divided by the same factor, and as a result power is always linear.  It doesn't include self inductance of the primary, won't work for a list of power supply circuits, and they do not deny that.  Just read the articles.

[poynt99]

I've done extensive simulations with inductors and transformers (linear and non-linear) and I usually have little trouble.

I can see right away two things from your diagram that will make the simulation unrealistic (and troublesome for convergence). What I suggest is:

1) Add some series (real world) resistance with each winding
2) Air-core coupling is never 1. Depending on how loosely the coupling, you may use anywhere from 0.2 to 0.7 or so.

I use PSpice.

.99

[elgersmad]

I've done extensive simulations with inductors and transformers (linear and non-linear) and I usually have little trouble.

I can see right away two things from your diagram that will make the simulation unrealistic (and troublesome for convergence). What I suggest is:

1) Add some series (real world) resistance with each winding
2) Air-core coupling is never 1. Depending on how loosely the coupling, you may use anywhere from 0.2 to 0.7 or so.

I use PSpice.

.99

That's good.  But, this is not unrealistic.  At first, it seems that way.  But, if you don't go to the trouble to actually build the circuits you won't know.  First, there are several kinds of transformers, and the geometry of the windings and cores make a big difference.  A 1:1 bifilar is locked to a specific value of inductance on the primary and secondary.  As the load changes on the secondary, it never changes the resonant frequency of a parallel tank circuit built on the primary in construction/breadboarded.  The Q changes and the ring value changes, so it uses more energy with a lower load impedance.  Now, if I use a U core or a C core, and wind the primary on one bobbin, and the secondary on the other at a 1:1 turns ratio, the primary and secondary inductances are no long locked to one value.  Instead of Q changing, the inductance of the primary will change as the load on the secondary changes.  But, to know this you would have to build two seperate transformers made from the same core material, and have the same winding cross section or area.  To be sure that your results are correct the inductance of L1, L2, of T1 should be equal to L1 and L2 of T2 when the secondary is unloaded.

In B² Spice, that would make the current transformer model more ideal, or modifying the lossy inductor model to replace the inductors used in that model to accomodate for losses properly. The Lossy Core, doesn't include INductance of the Primary, mutual inductance, or the inductance of the secondary, like the current transformer model does.  So, to simulate a real transformer, you would have to combine the lossy inductor with the ideal transformer model and the current transformer model.

[elgersmad]

There is a way to test a transformer and determin that the reason why the models are not working right is associated to how a reflected load actually works through a transformer from the secondary to the primary.  This inductance meter is very helpful in figuring out how the model has to change.
L/C Meter II

What you do, is hook up the primary to the meter, and the secondary to a potentiometer or resistor decade box.  Then just add a 100 ohms of resistance starting from a short circuited secondary and plot a simple graph.  To scale the graph, you measure the inductance of the primary open and then shorted to get the maximum and minimum values of inductance.  The Spice Model would include Pmin, Pmax, Smin and Smax, and if there are more windings, s1, s2, etc.  If you plot the graph you find out that the change in inductance of the primary is linear in respect to the secondary.  So, it's just a straight line.  But, the turn ratio effects it, and comparing 1:2 to 1:10 to 1:100 shows a real difference in change, and when you reverse the direction of driving it and use the secondary as a primary, which is always possible.  So, what you have in the model would be two seperate values of inductance, that change the value of inductance of an inductor in henrys based upon the load on the secondary.  That way the computer can recalculate resonance.  If you wanted a simple self tuning resonant power circuit, you'd base upon an Armstrong Oscillator design to use the core and feed back from a tickler coil to drive the oscillator.  So, when L of the primary changes, Q changes, and f0 floats over a range of frequencies that are load dependant. But, there's nothing you can do with multiple stages because, the center frequency is all over the place for any changing load.  So, we need a model that allows us to change the primary value based upon the secondary's load.