The second feature of the scheme is winding L3 to which we will apply short pulses through the switch VT1. Energy of these pulses are relatively small and can provide only 5-10% of the power load. In the initial state of L1 and L2 do not affect the L3, so with her help we will make a parametric change of inductance of TV1, and the winding of call management (the collector). With L4 will remove power through a diode bridge VD2 to serve the load HL1-HL4.
You can come up with your own TV1 as described under methods. The frame, which will be wound coil, is divided in half. Clockwise, until the mid-L1 is wound, changing the direction to "counterclockwise" until the end of the frame L2 is wound. Thus, we obtain two oppositely wound coils. L3 and L4 are wound on one and the other half on top of the frame, respectively, L1 and L2. The direction does not matter. If the winding length of L1 + L2, for example, 12 cm, then the length of the winding coils L3 and L4 — 6 see as the core of a used ferrite or transformer steel.
The author's data were as follows. The frame under the ferrite rod with a diameter of 10 mm. winding Length L1 + L2 is 12 cm, the wire diameter of all coils is the same — 0.6 mm.
The load in the circuit can be applied incandescent bulbs to 12V, 5 to 10W (two or three, connected in series). Good results showed watt led matrix GBZ-10W is designed for 12V power each. They need to connect the 4 pieces in series, as in the diagram.
The General approach is: the higher the load, the greater must be the capacity C, and the more voltage.
Transistors for the output keys can be any powerful high-speed MOSFET-s, for example, IRF3205 or IRFP460. The maximum voltage drain-source on-they must have not less than 55V.
The VD1 diode — Schottky, with a maximum reverse voltage of at least 100V; example — MBR10100.
This stage is the most interesting: we will need to catch the moment when at the same power output, the input will be less than 35-40%. But first you need to determine the resonance frequency of the circuit CL1L2 (above in the diagram). To do this, turning off when the switch VT1 is necessary to achieve the maximum amplitude on L1L2. The peak will be very sharp, but it does not matter, since the total rezonansa then the curve will be much flatter.
To get a relatively short pulse to the control coil plug VT1, and the fourth generator GG lock the top contact of the switch SA4. The combination of switch contacts SA3, which shifts the phase of the signal on the VT1 relative to VT2-VT3, we will have to get a pic of this pulse at the top of the sinusoid at L1L2. We need to achieve the maximum emission of svetodiodov HL1-HL4, and in the second maximum. First, at a lower frequency belongs to the contour CL1L2, and we have to tune a little higher. The brightness of the bulbs does not change, but the power consumption of the scheme will be reduced by 35-40%.
Another indicator of the separation of the first and second resonant maximum when the frequency setting may be the appearance of a sharp vysokochastotnogo squeak: while spinning the slider of R2 in the direction of lower frequency from him received the first maximum in the direction of greater frequency — we need a second maximum.
After the above settings, you can switch SA5 translate into a lower position, which will allow to smoothly change the brightness of the LEDs by the resistor R7. It changes the duty cycle of pulses in a packet that allows you to adjust the settings of the control winding. By the way, on this winding are relatively low-power pulses that control a much more powerful load that can be interesting both in theoretical and in practical terms.
Successful from the point of view of efficiency and effectiveness of load control was below, and the option to include parametric transformer TV1. This inclusion differs in that here the coil L1L2 steel detachable coil, and L4 inductor.
Attention! You need to be careful when conducting such experiments. In some cases, when applying a control pulse, the ferrite does not just breaks and explodes and shoots like a Gauss gun.