All this reactive power stuff is becoming a real interesting subject. if we look back to the post linked below we see the top left
shot is the fluro traces, and the bottom right shot is the tank traces feeding it.
http://www.overunity.com/14443/quantum-energy-generator-qeg-open-sourced-by-hopegirl/msg414843/#msg414843

Volt x amps in the tank is 84.8 x .922 = 78.18 VA and power factor is (192 cosine) = -0.978,
so 78.18 VA x 0.978 PF = 76.46 VAR.
So 78.14 VA - 76.14 VAR = 1.72 Watts missing from the tank.
With 3.8 Watts to run the AC generator
and a load power of 0.5 W
while the fluro is struck, lit up and conducting out of phase current as well as receiving a spike of positive power and some.

The 3.8 Watts input minus the 1.72 Watts missing from the tank = 2.08 Watts generating losses running the tank at 78.18 VA.
The 0.5 Watts output divided by the 1.72 Watts missing from the tank = 0.29 or 30% efficiency.
But the light is lighting like it's powered by almost 10 Watts = doesn't add up. 

Very interesting.

My setup includes the cost of generating the HF AC power from stored potential "energy". We could think of the
output tank as the grid wall outlet.  And the primary circuit as the HF AC generator, with the DC as the fuel.

A complete system to analyze.

The QEG mob and the other reactive power OU claimants plug in to the grid, which is equivalent to the output of my tank.
They don't take into account the cost of generating the AC in the first place from the stored energy. But they get billed for
the costs in their bill. That's cheating. That cost should be included in the efficiency of running the device.

I can simply use a battery for the 12 volt supply. And charged by a solar system as well. Our grid power is free as well
since our solar array feeds my power supply. Suck it up claimants and include the real costs of running the devices.
Calculate it out or generate your own AC power from stored energy.
..

EDIT: I need to clear up some things on how I confused some people unintentionally, 

1) In the video I show the drawing but I indicate I'm going to scope the tank current when I actually scoped the current out of the
tank to the load. The other two resistors on the board to the left of the right variable cap are for scoping the tank currents.
I moved the reference grounds so as to scope the load to a different spot, but "electrically" the same reference point.

2)On the drawing I use color coding with yellow for the current probe placement, when on the scope the current trace is actually
the blue trace, and on the drawing I use red to show where the voltage is scoped and the voltage trace on the scope is actually
the yellow trace.  Power is the purple trace "V2", the "shown value is divided by 0.1" to get VA.

Video clip of the test referred to. The shots are from a previous test but the values are pretty much the same.
https://www.youtube.com/watch?v=vFOHk_0IDZg

I hope I got all that right.
.

Yeah ok, fair enough, but how does the fluro give off so much light with 0.5 Watts power consumption ? Even if it was consuming
the entire 3.8 watts I think it was still too bright for that input power from the supply. The V x A in the tank is either almost all
active or almost all reactive. But there is real power there in the "Watts range", I think, gone from the tank.

The tank does remain having sine waves but the current trace gets jagged when the fluorescent lights "in any way", so the light
output has a reflection to the tank current. It doesn't distort the tank current or voltage trace because the oscillating power
is so much more in magnitude than the load power I think, If I load the tank with a large capacitor or an inductive load then
the trace can be distorted if the inductive load is not tuned to the tank. 
.
Resonant rise.
With 12.3 volts input I have 85 volts across the load.
Primary turns = 8
Secondary turns = 40
Transformation ratio = 1:5

EDIT: Coupling is not 1  , but I do not know what is, the distance between primary ends and secondary bottoms is
about 12 to 15 mm. End to end coupling distance 12 to 15 mm. I estimate 0.2 to 0.3 "k" to each side, maybe a bit more.


12.3 volts x 5 = 61.5
85v minus 61.5v = 23.4 volts resonant rise still on the secondary tank and load.

The resonant rise is what "strikes" the fluro and the declining Q diminishes V to an equilibrium with the load and input.
The different factors determine if there is resonant rise left applied to the tank and the load or not.
.
The primary reflecting capacitor gets charged to over 40 volts each cycle and that calculates to 7.4 joules per second is put
on the primary "reflecting" capacitor that's across the switch somehow. Exactly double the input. Should be the other way round.
..
It's bizarro world today, everything is back to front and upside down.
.

P.S. Isim if there is more any information/data at all that I can provide to help with a simulation just say and I'll get it as quick as
possible, due to our time differences I need to make an effort to be "in time" for you.

Circuit values measured -
The drawing shows the secondary coils have 112 uH but they turned out to be 119 uH each measured. 1 mm Magnet wire. 40 tuns.
The frequency is now around 420 Khz.
The measured capacitance for resonance with the fluro load in place is 957 pF. Parallel plate air variable capacitor 1350 pF max.
 
Values by estimation and logic. 
Calculator says 119 uH should need 1207 pF for resonance at 420 kHz.
so inter-winding capacitance must be between 200 to 300 pF I estimate. 
The primary inductance now that I've spaced the primary turns does not read on my meter.  2 x 1 mm wire in parallel 8 turns.
The primary switch has a 20 nF capacitor across it and it seems fairly close to resonance so primary should be
in the 5 to 7 uH range.
Fluorescent Lamp is a - F10T8CW also has 40H written on it. Info sheet attached.

..
Coils are all clock wise wound and I'm using the secondary coil at the mosfet drain side of the primary coil.
.
This attached saved  and zipped web page of a solid state Tesla coil calculator models my transformers input power and
output voltage from one side with the fluro load fairly well.

Solid State Tesla coil calculator
http://www.extremeelectronics.co.uk/calcs/index.php?page=sstc_calc.php
.

The only limitation is the inductance.  The 75mV shunt has lots of inductance and shouldn't be used above mains frequency.  The smaller shunts fro the Kill-a-Watt will be better but depending on their values will probably distort in the high audio range.  The Ohmite WNE's are only about $1.00 each from Digikey and the 0.5 Ohm and 1Ohm will both easily be accurate at 1MHz.

OK no probs, I'm ordering some tomorrow anyway, just curious, those are measuring 0.1 Ohms. I can look with the scope and
see if there is a difference between the power resistors I was using and these from the killaWatt meters.

It's not the price of the resistors Mark it's the postage, I'll have to spend an hour tonight listing all the other parts I want as well.

Now will a Lux meter tell me if a lamp is giving off more light ? Or will it only tell me the brightness, not the amount ? I'll probably
just make one of TK's light meters. I'll need parts for that.

..

Small packages from Digikey cost about $11.00 postage if sent by US Priority mail to anywhere in the USA.  The cheapest FedEx and UPS options are about $21.00.

LUX meters report intensity in LUX.  To get total light, you have to measure in all directions around the source and integrate.  The $10. digital photometers on eBay aren't bad.

I'm in Australia, it cost me over 30 Dollars last order. But it's ok I'll get some stuff.
Thanks.


P.S. Mark can you please check back here for questions while you can, about components ?

What would be the best mosfets and mosfet driver chips to use for switching up to 24 volts in a setup like this at 840 khz or
1 Mhz tops ?

That really depends on the load.  You want the lowest gate charge that you can find that will still yield an acceptable on state resistance.  I would get a good feel for the maximum voltage with transients and then start searching MOSFETs from the voltage and up for the lowest total gate charge at the gate drive voltage of interest.  The gate charge is what is making your drivers hot. Slow rise and fall times that result from the charge and perhaps layout issues is what is making your MOSFETs hot.  At 1MHz and above hard switching losses can be quite a problem.  Resonant and quasi-resonant topologies are more complicated, but those that switch at zero voltage really cut down on the switching losses.

I ask because the TC 4420 gets hot driving a mosfet at 1 mHz and the power consumption for the switching alone seems
very high. At 840 kHz the driver chip is not too bad but at 1200 kHz it gets quite warm "touchy". I'm using an IRF740 but
the only others I have is one IRF540 some IRF840's And some lower voltage IRF1010's.
I've got some Si7478DP-T1-E3 mosfets but they're only 60 Volt parts. Just wondering if you had a suggestion, maybe some
new line of part or something interesting.

Digikey's postage is very fast so holding back won't hurt. I'll order tomorrow after midday.

2) Capacitors for the primary what are the very best capacitors I could use for the primary capacitance ? I'll need them to take
similar voltages to the mosfets I guess. That's an important one the ones I'm using now do get a bit warm.

..

STP19NF20 is a 200V part has a maximum gate charge of 24nC compared to 63nC for the IRF740.  It has an on state resistance of 0.16 Ohms @ 10V drive versus 0.55 Ohms for the IRF740.  So it should cut the driver losses by almost 2/3s, and run cooler itself.

You can cut the gate charge in half again if you are willing to use 150V rated parts:  IPP530N15N3 G.  Gate charge on those is just 12nC which would reduce the driver work load by more than 80%.  The on resistance is under 100mOhms.

Funny, as it happens I was fair dinkum planning to buy this Blackview-JK890-Android phone this week. I decide it was time to get
a smart phone to replace several devices for most applications, my old Motorola Razr flip phone still works like a charm, but it's
about 8 years old now I think. I'll be keeping it working and getting a new sim for the unlocked phone with a different carrier.

http://www.ebay.com.au/itm/Big-5-5-IPS-Dual-Core-SIM-Android-4-2-3G-GPS-Mobile-Phone-Smartphone-Unlocked-/161386679896?pt=AU_Mobile_Phones&hash=item2593654e58

That phone doesn't have a light sensor by the looks of it. I would love any suggestions for a better phone about the same price
or cheaper preferably. I want the bigger type screen though, nothing smaller than 5".


Thanks.

yeah no problem, I think this is it attached. I'll find a tutorial and get myself trained sooner or later. I need to label well any files
I want to keep.
.
I just lit up a 4 foot fluro tube using both sides and it struck up well but it's not quite at the same brightness as the one powered
by the grid in the hut here. Maybe an AC primary circuit will make 1 spike each half cycle and light the tubes better, using only one
secondary causes the tube to be lit from only one end due to the DC primary circuit and the only 1 spike per full cycle. But using
both secondaries the 4' tube strikes real easy. Input was exactly double the input for the single 1 foot tube lit from one
secondary. I need to use a voltage booster on the primary of some kind, to adjust the primary voltage to the load as well.

Adjusting the tuning of the separate sides before full conduction makes those funny looking dark rings go from the sides to the
middle or stand still or one can move and the other stand still, depends on adjustment, curious but that's all.