After seeing some industrial applications of induction heating, I thought that it would be an interesting experiment to construct a small scale induction heating unit. This process is a means of heating conductive objects by passing an electric current through them. The part being heated is made to form a shorted secondary turn of a power transformer (usually air-cored). Electric power is coupled into the primary, and the resistive dissipation in the secondary causes heating. In steel parts, magnetic hysteresis losses also contribute additional power.
The ability of this technique to concentrate large amounts of power into a small space gives the potential for quite high temperatures to be reached. In a conventional oven or furnace, the temperature of the flame or element governs the maximum attainable temperature. However, when the 'element' is the actual object being heated, it will continue to heat up until the radiated heat loss matches the input power.
In order to achieve good coupling between the primary and secondary when using an air cored transformer, it is necessary to use a much higher frequency that the 50 or 60Hz used for power distribution. So, a high power, high frequency AC source is required.
To provide this, I used a one transistor power oscillator, with transformer coupled positive feedback. This type of circuit is commonly used for applications such as running fluorescent lights from low voltage DC supplies. Colin Mitchell has got a good explanation of this circuit on his site. (I have built Colin's lamp inverter design, it works well)
The circuit is quite simple really. The exact component values should not matter too much, provided components with appropriate ratings are selected. I built up the circuit point-to-point on the workbench, as it was only required for a quick experiment. The transformer primary, or "work coil", along with C2, forms a parallel resonant circuit in the collector of Q1. This allows the current in the coil to be many times the supply current, which translates directly into more current flowing through the object being heated.
The switching transistor Q1 was an MJ15003. This is rated at 140V, 20A, and 250W. The DC safe operating area allows 20A at Vce=10V, 8A at 30V, and around 4.5A at 50V. It is in a TO-3 package, and is bolted to the bottom of the large heatsink in the picture.
The "work coil", L1, was wound with 8 turns on a piece of 32mm diameter PVC pipe, over a length of about 50mm. It is important to use thick wire for this coil, the wire I used had a cross-section of around 15mm^2. While this may seem excessive, it will have to carry large currents, and some of its cross-section may be unusable due to skin effect. It would have been better to use a former that could withstand higher temperatures, though it must of course be nonconductive.
As for the coil, the resonant capacitor C2 must also handle a large current,and should have a low dissipation factor. I used 10x 100n X class MKT capacitors connected in parallel. If you build this circuit, check that the capacitors do not get too warm. If they do, you will need better quality ones.
The feedback coil L2 was simply a few turns of hookup wire around the work coil. It is necessary to get the correct phasing here, otherwise the oscillator wil not work. R1 is used to adjust the transistor bias - a resistance substitution box was used here. The circuit was run from a bench supply capable of giving 50V at a few amps.
General view of experimental setup
Results from the circuit were quite promising. An M4 machine screw held in the centre of the work coil was heated to orange heat in approximately 60s. Current consumption was around 3.5A at 50V. The heatsink warmed up slightly after a few runs, but the dissipation was by no means excessive. This indicates the efficiency is quite reasonable, and would point to a heating power in excess of 100W into the part.
4mm machine screw at orange heat. Note colours are somewhat inaccurate due to the camera picking up IR radiation.
The voltage across the work coil was monitored using an oscilloscope, and was found to be approximately sinusoidal, at 39VRMS and about 130kHz. The capacitor will have a reactance of 1/(2*pi*f*C) = 0.81 ohms. The current flowing in the capacitor (and the work coil) can be found by V/X = 39/.81 = 48A. This clearly shows the current magnifying effect of the parallel resonant circuit, and the need for the thick wire!
I didn't really experiment much with the coil geometry, part size, or operating frequency. No doubt, performance cold be improved by optimising these factors. However, it certainly gives a good demonstration of the principle of induction heating, and I think getting parts to glow orange is quite impressive from a few junk box parts thrown together in half an hour.
Another application for this circuit is wireless power transfer. I made a coil out of tinned copper wire, and soldered it to a 12V tail light globe. Inserting this assembly into the work coil while the circuit was running caused the globe to light. Further experimentation showed that it was possible to light a 12V 50W halogen globe at full brightness from nothing more than a jumper lead wrapped three times round the coil.
Tail light globe and coil.
Light globe illuminated by work coil.
Halogen lamp at full brightness.
The circuit probably isn't practical for real-world applications in its current form. The transistor is running quite close to its ratings, and there is no protection circuitry of any kind. I eventually blew up my transistor after playing around with the circuit for a while. However, hopefully this page will serve as a basis for experimentation if you are interested in building an induction heater.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.