Simple to build isolation transformer that consumes less power than it gives out

[forest]

In this case winding direction does not matter, only the polarities which you can easily find out by swapping the second coil. In the junction of the two coils polarities must be the same at the same time instant.

Jack, my english is not good enough. I don't understand this sentence :

"In the junction of the two coils polarities must be the same at the same time instant"

Please give me a clue.

[Jack Noskills]

Think in terms of poles S and N, polarities. When current goes via coil one end will be S and the other will be N. If you put two coils together and winding direction is the same, you get S-N-S-N. In fact each turn in a coil forms S-N and when connected they form one S-N electromagnet.
Idea is to connect those coils so they form S-N-N-S electromagnet and you connect load to either S-N or N-S. Easiest way is to try the two bulb experiment, first L-coil-bulb1-bulb2-N, then one coil parallel to bulb1. When bulb2 is not lit or dimmer than bulb1 then coil is connected correctly.

What is the permeability of your core and what is the DC resistance or you coil ? Can you compute AC resistance (impedance) at your operating frequency ?

[Jack Noskills]

I am trying to make a simple tank circuit but just cannot get it to work. I am using bulb after tank to detect resonant frequency, when it goes out tank would be tuned. Is this correct way ?
Now I am wondering accuracy of inductance calculator, if inductance changes with drive frequency, how can calculator that does not need this frequency as input ever work ? For example, this calculator

http://www.mantaro.com/resources/impedance_calculator.htm#toroid_inductance

gives certain L value and when combined with 1 uF cap it says resonant frequency 7084 Hz. When I do a sweep from 7000 to 8000 there is no sweet spot there. I tried brute force method, search over 500 - 20000 Hz using 2, 5 and 10 turns with 10 nf, 220 nf, 1000 nf and 2000 nf caps but nothing showed up.
Can someone give any help how to detect resonance (without using any meters) ?
My idea here is to make tank circuit, then put same amount of turns in the second coil and then take power, possibly putting a series cap between load and second coil. Three possible locations for series cap exist so could test all positions with little effort.

It was a struggle to get a step up trafo working at high frequency, normal winding did not work. I had to use layers of bifilar coils, odd layers CW and even layers CCW, then connect those together to maximise output voltage. Now it can give out around 190 watts even at 20 kHz while normal wound gave me only about 25 watts. I measured it by using watt meter in the wall before audio amp and shorted the secondary. It was a nice learning experience, I actually solved a problem lol.

[forest]

Think in terms of poles S and N, polarities. When current goes via coil one end will be S and the other will be N. If you put two coils together and winding direction is the same, you get S-N-S-N. In fact each turn in a coil forms S-N and when connected they form one S-N electromagnet.
Idea is to connect those coils so they form S-N-N-S electromagnet and you connect load to either S-N or N-S. Easiest way is to try the two bulb experiment, first L-coil-bulb1-bulb2-N, then one coil parallel to bulb1. When bulb2 is not lit or dimmer than bulb1 then coil is connected correctly.

What is the permeability of your core and what is the DC resistance or you coil ? Can you compute AC resistance (impedance) at your operating frequency ?

I cannot measure permeability of core  or impedance. No tools. I'm also in first stage - preparing isolation transformer. I dismantled one transformer and added 500 turns for each primary bobbin, converting step down transformer into 1:1 but results are not satisfactory. Probably internal shortcircuit beccause resistance only rised by 6 ohms. Anyway , I have to find coupled capacitor.

[Jack Noskills]

I was able to make tank circuit and tune it. I started to use 12 volt 5 watt halogen so there was no need to step up. I used 11 turns on each coil (separate coils) on 80000 perm toroid. With only one coil connected I found resonance at 1956 Hz, 1957 Hz was little bit off. Then I connected the second coil and resonance point moved to 2169 Hz. With this setup I was able to replicate the two bulb test, my watt meter showed power was consumed though. This conflicts earlier results at grid frequency, or power is being pushed back. No idea how to check what is happening here.

Another test was a sine sweep starting from 300 Hz upwards using the same two bulb setup now with 50 watt halogens. At low frequency the current limiter bulb was lit while load bulb was not. As frequency increased light moved to load bulb. This shows how impedance increases in coils and power begins to go to load bulb instead of current limiter bulb. Watt meter showed the same power consumption during the sweep.

One thing I noticed was that my amp made clicks when signal was turned off even when watt meter showed 20 watts was consumed when I was testing plain 1:1 trafo, I quess it comes when power is being switched off and some safety mechanism is activated. But when I used the modified version, there were no clicks even if watt meter showed 100 watts. Maybe this is nothing special.

Finally I tested step up version, eleven turns in one coil, second coil 10 meters bifilar winding except coils were connected from the end to form center tap, about 150 turns total. So I got this as shown in the picture:

L - 11 turn coil - bifilar start1 - center tap - bifilar start 2 - N

all coils in same direction.

220 volt bulb connected parallel between center tap and bifilar start 1. 12 volt halogen connected serial between bifilar start 2 and N.

This setup gave full brightness to 220V/40 watt bulb while 12 volt halogen was not lit at all. Then I added second 40 watt bulb and also full brightness and now little glow in halogen. Again watt meter showed power was consumed and when signal was cut off no click. No caps here as it worked at all frequencies. Again behaviour of the bulbs similar to low frequency trafos, but conflicting watt meter result.

If this is OU, it is not very powerfull. Something is still missing. I tried parallel capacitor between center tap and bifilar start 2 but I was not able to find resonance point between 300 - 20000 Hz. Load light went out completely and core started to warm up. All the power went through the cap instead of coil. Maybe this tuned cap would be the solution to get more power out, or thicker wire in bifilar so capacitance is increased. I think it is worth investigating but I have run out of ammo for now and help is needed from more experienced forum members.