DIY: Phase-Shift Controlled Variable Induction Heater

DIY: Phase-Shift Controlled Variable Induction Heater


I cannot be held liable for any use/misuse of this information, the resulting design/assembly/operation/maintenance/etc are all strictly your own doing.  This is a dangerous machine and my plans were thrown together on the spot from sources gleaned on the internet and through the actual assembly process – most of the actual circuit diagrams were drawn up after the machine was built for the purpose of this tutorial and may not even reflect the actual circuit – though I’ll try to keep it simple and straightforward – use common sense: if you don’t know what your doing don’t do it.


I began some experiments requiring custom vacuum tubes a couple months ago, and after checking into the costs of having them made – decided to make them myself.  I will be posting tutorials as I complete various aspects of this project that may be of use to others – this one is on how to make a phase-shift controlled variable power induction heater – I’m building it to fire getters within tubes and to degas the surface of the metal portions, though I might end up utilizing it in the creation of custom alloys as well, since it does provide a clean and efficient method of smelting small quantities of metal relatively quickly.


I used this as a rough basis for my induction heater – I HIGHLY recommend reading through this before starting as it goes into very good details about different methods of creating induction heaters, the physics behind it and the like – though it is very light on details of control circuitry so I made this tutorial as a result:

Power Source

I’m using 3 power sources for this project, one is 5VDC 12VDC/350mA and the other is appoximately 4VAC/360A – a bit of overkill but it will be easy to expand later if I need to.

5VDC/100mA Power Supply

The 5VDC  power supply is dedicated to the signal generator and phase shifter for the sake of reliable signal generator.  Power is provided via a 120VAC to 6VAC/100mA transformer with an attached LM7805 5V voltage regulator after a bridge of diodes with a relatively large (~3600uF) capacitor to allow for stability.

120VAC/16A Power Supply

I found a 120VAC/2KVA oil-filled transformer on eBay for $50 and went with it even though my MOSFETs are only rated for 500V/14A – just have to add a couple turns to the secondary side to drop the amperage and raise the voltage a tiny bit (~143V to hit 14A, 154V for 13A just for paranoia’s sake).  The power is then rectified with a 1KV/90A bridge rectifier (just because it was cheap and not under the power rating needed- though I’m not opposed to over-engineering) I may or may not smooth this supply as my (possibly incorrect) calculations for the capacitance needed for the circuit to actually be smoothed are based around the resonant frequency and tank cap of the coil as follows:

376nF [tank capacitance] * 225600Hz [work coil drive frequency] = 84,825,600nFHz [capacitance per second]

84,825,600nFHz / 60Hz [mains frequency] = 14,137,600nF or 14.1376mF [capacitance required to smooth the work coil’s draw when powered by mains frequency]

Ultimately, the worst case for an unsmoothed supply voltage is a ~50% longer runtime as radiated heat is minimal for a given timespan and catching the cycle out of phase with a significantly higher frequency pulling from it won’t split the power any – it just won’t flow out of phase – so the electric bill shouldn’t even be higher than with a smoothed supply.

Work Coil and Tank Capacitor

The work coil should actually be one of the first things you make on this project as you will need to measure it to get a proper basis for your control circuitry if you utilize phase-shift power control of the H-bridge legs as I describe in this tutorial.  I used a work coil that (purely by chance) came out to 1.5uH of induction (matching that described in the source URL above) so I used the same capacitance he described of 376nF (implemented with thin metal film high frequency AC capacitors ordered on eBay – whether they pop has yet to be seen, though if they do I’ll post a solution in an update on this page).  When I plugged this into a network analyzer I was able to spot the resonant frequency at 225.6KHz with several harmonics thereafter – I wouldn’t recommend utilizing the harmonics as they will result in more waste energy and you might have trouble finding MOSFETs that can even switch the power required quickly enough if you choose one too high.

Control Circuitry

Once you have the resonant frequency of the work coil/tank capacitor component you can build the signal generator and phase shifter.

Signal Generator

For a signal generator I used a MAX038CPP IC – I would recommend going with this (despite the fact it’s been discontinued they should be pretty easy to find on eBay) or something similar as it takes very little time to build and tune.  The circuit diagram for my signal generator implementing the MAX038CPP IC follows:

Phase Shifter

The phase shifter is a PitA to design for this frequency – unless you want to amplify the signal from the signal generator and run it through a very large coil (~655m in length – so about a cubic foot in volume) with a slide attached to it for 180 degree adjustment of the phase.  I was playing around with LTspice IV for several hours trying to make an analog phase shifter for the sake of precision and finally gave up as I couldn’t find a mechanism to do so without requiring at least 2 variable capacitors that are themselves fairly expensive – if anyone reading this knows of a method for analog phase shifting with either a single var cap or a single variable resistor I’d be happy to hear it.

The following are images from LTspice’s simulator and the circuit diagram for a phase-shifter that will output a signal for 8 power settings (inclusive of full power, 9 if you count “off”) for one of the H-bridge legs (the other one being attached directly to the signal generator):

Note: not all outputs are in the correct order – from left to right on the circuit diagram power settings are as follows with 8 being the highest power setting and 1 being the lowest: 8, 6, 4, 1, 7, 5, 3, 2.  The 180 degree phase angle is not included as it would be more efficient to simply shut the device off (no power would flow at 180 degrees out of phase – disregarding NPN/PNP transistor switch time differences and marginal phase shifts introduced from digital inverters which ultimately implement them to split off one of the H-bridge legs) to avoid uselessly heating the high power transformer.  I’m not too worried about the ripple in the pink trace since the outputs will need to be converted to a digital signal before being sent to the MOSFET power switches of the H-bridge at a 50% duty cycle, but if you happen to know how to get it without dropping into the nH range of inductance on the pink block of the circuit diagram I’d like to hear it (as opposed to dropping capacitance from 10nF to 5nF as I did to achieve that particular phase angle – inducing the ripple as a result).

More to come as this project continues.

About the Author