Powerful Yet Simple Induction Heater

21/06/2014: Updated power supply section.

This is my first instructable and I'm not really good in explaining things, so forgive me if it sounds like too much.

As this whole thing is a work in progress make sure to check for updates, which I will add at the end of each step.

In case there is need to change the original step I will include a seperate note at the beginning of the changes.

You can find quite a lot of schematics and instructions on how to build yourself a nice induction heater,

but I noticed most of them simply lack the understanding of how everything actually works together.

If you tried it before you might have noticed your capacitors getting so hot that they pop or despite a good power supply it takes forever to get a small nail glowing red.

That is the reason why I wanted something better...

Please excuse that I repeated myself a few times, but it is only for the thing that are important.

The basic circuit is based on Mazzolli's flyback driver, so thanks to him for the initial design.

Part 1: How does it actually work?

Induction heating works basically like a transformer inducing so called Eddy currents into a conductive workpiece.

This works at all frequencies but in the ragne of 20-250kHz ferrous metals heat the best.

Every time the magnetic field of the work coil changes the resulting Eddy currents will heat the workpiece.

Considering the size, shape and mass it quickly becomes clear that for bigger stuff you need more power or it will take forever.

I don't want to go too much into the details as Wiki can supply you with all details about it ;)

Part 2: What are the dangers?

Induction heating is surprisingly safe as it is a form of non-contact-heating, meaning your workpiece never has any contact to the insane powers in your workcoil - no touch.

Also the heating stays localized inside the coil, so longer workpieces can be held in your hand.

But! The voltage produced inside the tank circuit will go as high as Pi multiplied by your input voltage.

This is one reason why I limited thisversion to only 30V input voltage for the time being.

Currents go the way of least resistance and unless you touch the workcoil and some grounded earth at the same time nothing at all will happen as it is much easier for the currents to go back into the circuit.

Of course I recommend to insulated the coil with paint or a silicone sleeve to be on the safe side.

Part 3: Consideration for the selection of parts to build

I fried quite some caps and killed a bunch of mosfets on my way to succes, so I will try to help you avoiding these downfalls.

Many tutorials give you completely different values for the parts used.

The circuit in question is a so caller "Royer Oscillator" which is basically a flip-flop that gets the timings from the attached tank circuit.

There are several ways to make sure the circuit resonates properly and does not kill your mosfets, so I decided to go the most simple way with some modifications.

If you have access to electronic scrap from hard rubbish collections or similar you don't have to buy any parts, except maybe the copper pipe and a water pump for cooling.

The mosfets are not eally critical as long as they meet or exceed these specs:

At least 150V, 10 or more amp and a low internal on resistance below 0.15Ohm- IRF540 or IRF840 are commonly available for cheap otherwise check the datasheets of the mosfets you could salvage from power supplies, TV's and similar.

You need two identical mosfets.

Either small heatsinks with a fan or big passive heatsinks are recommended to mount the mosfets.

The caps seemed to be of great importance to me in the beginning of this project.

I constantly had them heating up or even explodig on turn on.

So going for better quality was my first order of business until I realised that even expensive pulse caps and caps designed for induction cooktops won't solve the overheating.

It was only I perfected the work coil setup and the electronics when I realised there are several hundred amps going through the caps I used.

Running time of the heater was limited to a few seconds which meant a new way had to be found.

The ideal capacitance for a work coil following my desing is between 2 and 4µF.

The safest way to get there is by using a lot of smaller caps connected parallel and directly to the exteded pipe of the coil.

So you end up with a coil that has two long "legs" with a lot of small caps between them.

MKP snubber caps used for mains filter work great in the range of 100-330nF - the higher your input voltage the smaller and more numerous the single caps should be.

Don't try to solder them onto some cable or copper sheet to connect them as I can tell you even 6quare mm burns trough in a few seconds.

For input voltages over 15V you will need thicker copper pipe, I used 5/16" below that 3/8" will do fine.

To get the power needed to run I removed the HV side from a MOT and rewired it with thick cable to get two windings, one for 15V and one for 30V, running through a 25A rectifier and 7200µF caps for filtering.

Part 4: Hints on making it easier.

The mosfets usually have a metallic back which is connected to the drain, so if you want to avoid big heatsinks you can braze a copper sheet onto the pipe, mount the mosfet on it and have a perfect connection plus water cooling for the mosfet at the same time - plus you get it all a bit smaller :)

All connection to and from the tank should be oversized!

With mosfets mounted directly to the pipe this leaves the connection to the choke coils (more on that later!).
The reason for this is that if there is a problem with the resonace your mosfets might pop but also your soldering connections resulting in quite massive sparks.

Using thicker cable can help avoiding this.

A lot of the available tutorials will state the use of Zener diodes and parallel resistors from Gate to source on the mosfets to keep the gat voltage at around 12V no matter what the input voltage is.

This can work quite well but I found that the use of a 12V regulator to drive the gates is much better as it allows for much more current and therefor higher and more accurate switching times.

I came to this conclusion after frying a few 5W zener diodes...

A small heatsink is recommended for the regulator to avoid overheating.

Part 5: Getting it all together....
Please check the schematics carefully before proceeding!

You can use any power supply that is capable of delivering at least 10 amps at 30V - lower voltages down to 12V are possible but reduce the output power.

I used no circuit board but direct wiring with standard house installation wire as it is the easiest way, especially if you use the heatsinks to connect the mosfets to your tank.

Again: make sure all connection to work side (what is connected to the output of the mosfets) is done properly!

Too thin wires or bad connection will result in massive arcs!

Leave enough pipe on the coil so you solder the caps directly between the two pipes - this will drastically reduce the need for copper sheets or finger thick wires to connect the cap bank to the coil.

Part 6: Ok, I have it all connected but it looks like it is one big short and can't work....

Well, in one that is true, but you have to keep in mind the circuit is self resonating.

This means both mosfets switch when the amps are at zero.

Due to the high frequency the workcoil, caps and choke coils oscillate in resonance.

The impluses are extreme short and it will work fine.

Part 7: Troubleshooting...

As I mentioned I killed a few parts on the way of developing this heater, so here is what to look out for.

The caps should be MKP types rated at 275 or more volts, Wima caps are ideal.

If your caps overheat (and you got the right type) your single values might be too high, try 220nF or 100nF caps.

The mosfets should only warm up with big loads in the workcoil, if they get hot with no load you might have to use better choke coils - I used yellow ferrite cores with quite thick magnet wire.

The diodes should be really fast, there are many ultra fast recovery diodes available - as long as they are rated for 200V or more you should be fine, if they are too slow your mosfets heat up and you have poor performance.

All looking good but no output power at all? Check if the 12V supply to the gates is ok ;)

Last word of advise: Never hold small workpieces by hand, a drill bit gets red hot in seconds, which can be quite painful for your fingers ;)

Since not everyone is familiar with mosfets, soldering and pipe bending I will add some more info here.

First off the coil:

You see the finnished coil and how it looks with the capacitors and wires for the mosfets attached - not the thickness of the wires!

I do recommend to use quite thin piping and not the 1/4 inch I used here.

But as I plan on using a ferrite mantle (there will be a tutorial on making proper ferrite for this project at a later time) and a higher input voltage I used this thick pipe.

For short term use you can do without water cooling, especially for smaller workpieces but I recommend to include a little water pump to avoid overheating, which can actually cause your solder joints to fail with the high currents.

Use annealed copper pipe, this is the stuff that comes in a coil anyway, the straight pices of pipe are not good for our job.

There are several ways of bending pipes, most involve tools like benders or springs.

Other ways are using salt or fine sand to fill the pipe before bending.

They all work but I don't like any of them as you first have to straighten the pipe to fill it and than tap it hard to comapct the sand/salt, tools cost money...

I simply leave the whole coil as it is and fill it with water - you need a good seal and a bit of pressure so all the air comes out.

Put it in the freezer over night and do the bending the next day.

If you don't have the option to weld a little bar on a piece of steel pipe to hold your copper pipe in place:

Put the coil over your steel pipe and straighten the inner winding a bit - about 40cm is good.

Now use both hands and bend both the straight bit and the still coiled up copper around your steel pipe.

You will see as soon as you complete the first turn the bend around the pipe with start to follow the shape of the steel pipe.

Now hold one end and wind the other so you end up with a straight bit long enough for your caps and connections.

The number of tunrs affect the output power and frequency - 3 turns minimum - 8 max, I opt for 5.

Hint: You might notice when you bend the pipe a few times that is becomes hard to bend and might even get cracks.

Heating the section of pipe until glowing red and letting it cool down fixes that problem and softens the copper.

Update! :

Heating vs melting, coil size and pipe diameter.

I did some test with my previous coils to compare output power levels.

My intial thoughts on coil size and magnetic flux density have been confirmed.

For ferrous metal and normal heating the coil diameter is not really that important, meaning it is easier to use 1/2" pipe than 5/16" not only for the cooling.

But when it comes to non ferrous metals or melting you want to maximise the flux as only the eddy currents will heat the workpiece.

So a small diameter coil with little gap to the workpiece is prefered for this purpose.

Since melting is not an option at current input voltages I did not really bother to make different coil diameters to match my workpieces.

To get a figure on the water cooling:

With 1/2" pipe reaching forging temps for small object like a 8mm bolt is no not problem without water cooling, continous use or short cycles however will require water cooling as the copper not only heats up fast but also stores the heat for quite some time.

There is not need for huge amounts of water, a small aquarium pump or similar will do, you only need a slow water flow at these levels for cooling.

Soldering on the pipe....

At this stage you might have encountered the problem that your soldering iron is incapable of heating everything for a proper connection.

There are many was to overcome this problem so I will reduce it to the two most simple.

a) Put the coil in the oven at about 200° celsius for about 15 minutes to get an even temperature throughout - make sure to wear gloves when handling it to prevent burns!

You now have abot 15 minutes to solder the caps and cables on until the copper cools down.

Working fast and on both ends of the pipe makes it easier.

b) Use a blow torch :)

Heating the coil or the pipe ends with a blow torch will heat the copper enough for proper soldering, just make sure to keep the flame away from the caps!

This is also good if you used a) and the coil cooled down too much to finnish the soldering.

Be aware that oxidization will happen and that normal flux solder won't penetrate this layer, so you might need to give it a quick rub with some stainless stell wool.

You need to make sure every cap is soldered properly to avoid sparks or melting connections during use of the heater.

It pays off to first put a layer of solder on the pipe and to add the caps one by one while adding enough solder for a proper connection.

If you have enough of these terminal strips as used on lamps or to connect two cables, you can remove the plastic around them and braze them onto your copper pipe so you can use the srew connection for the caps.

This is only useful though if you want to experiment with a lot of different caps for the tank or have the need to change the working frequency without soldering.

You can use whatever you available to power this circuit as long as you stick to voltages suited for your mosfets.

As you can see I used an old microwave transformer that I rewired.

I used a angle grinder to open the welding lines on the sides and removed the secondary, high voltage coil.

The wire used for this was standard house wire, but please don't rip your walls open to salvage it, find other means of supply! ;)

For testing purposes I have a center tap to allow the usage at 15V and 30V - currently running on 30V.

There is no use to try low amp supplies as you need at leat 5amps and for bigger coils and workpieces you might go as high as 15amp - so choose wisely.

As you can see I also used this wire for the connections to the rectifier and filter caps - although here the term filter caps is not correct.

Rectifier is a standard 25A type mounted on an old CPU heatsink - for higher voltages a fan is required for cooling during long use.

To get proper power levels you will need a stable DC supply voltage that can handle the fast switching of the mosfets.

Without big capacitors your power supply might get into trouble in terms of supplying a stable DC.

In my case 2 7200µF electrolypte caps from the voltage doubler stage of an old amplifier are used.

Anything below 5000µF will reduce you output unless a proper power supply is used.

PC (ATX) power supplies might work but I have not tested them yet.

Update! :

After some feedback from someone in my forum who complained about low output power I did some tests and would like point out some important things.

From the electrical grid we get 50 or 60Hz depending where you live.

In the current setup with 2µF tank caps I run at 75kHz.

Unless you have an overrated lab power supply you might run into the same problem with low outpout power levels, so here is how to fix it:

To make sure the mosfets can deliver the full current several thausand times per second the DC supply must be able to handle it.

In the rectifier stage you need caps with high µF values and high energy levels.

In my pictures you can see the massive size of my two caps, which are designed to store high energy levels as the original use was in a voltage doubling stage of an amplifier.

Using the standard, small, electrolyte caps you find in standard power supplies and DC filters you won't be able to provide the energy needed to reach max output power.

Same, by the way for the cap on the 12V outpur side for the gat drive - go for at least 1000µF rated for 25V otherwise you might have problems driving the gates properly.

As an alternative for expensive huge sized caps on the rectifier stage you could use lead-acid batteries.

For example to run on 24V you use 2 12V batteries (about 4-8AH) and a regulated power supply at 26V.

This is of course only a last resort option or for those that only have power supplies with low amp ratings.

Check old amplifiers from the rubbish collection or scrap yard - the bigger their pysical size the bigger the energy levels available.

You will see what I mean once you compared the normal caps with those ;)

Update 21/06/2014:

I use now IRFP260 mosfets and an old stick welder for the power supply.

The use of the 260's was necessary for the higher supply voltage and more power output.

As my former modified MOT was constantly overheating I did some more checks on the actual power usage with the welding transformer.

Using my big coil as used in this Instructable the current going into the rectifier at close to 40V is already 9.5Amps.

With that in mind I strongly suggest to use a power supply with suffient reserves as bigger workpieces will need more current.

It also shows that mosfets with a high enough rating should be used, even with my oversized heatsinks the mosfets do warm up with longer use.

As they are supposed to only switch when the current is zero it nicely shows that the circuit is doing it best to at resonance, otherwise the mosfets would just go up in smoke.

Once the ferrite coating is complete I will upload another video showing the power levels and heat up time using the workpieces from previous videos.

To do it properly a technician will tell you to use gate driver IC's and some logic for the frequency control.

Although this will be a very safe way to protect your mosfets, the additional electronics actually make it more complicated than necessary.

You can see I used a simple 12V regulator with a 1000µF electrolypte cap, the small cap is only for filtering and the diode for general reverse protection - not a must except for the big cap.

The use of 12V Zener diodes instead of a regulator might seem easier but in my experiments I had a lot failures with them and as a result killed the mosfets.

Using a 12V regulator capable of delivering 1A solved this problem for good.

The mosfets are mounted on oversized heatsinks, you can use smaller ones, I had these at hand and was too lazy to look for smaller ones.

Although a bit hard to see in the pic, the resistors and fast switching diodes are soldered directly to the pins of the mosfets.

The thick blue and red wires are the connections from source of the mosfet to the negative output of the power supply - they must be quite thick as a lot of currente runs through them!

Update! :

I'd like to point out that using a good sized cap at this stage is vital to make sure the gates of the mosfets are driven properly!

This is especially true if your caps on the rectifier stage are insufficient to deliver the power you need!

1000µF is the lowest you should use here, go for 2200 or 4700µF if possible.

Update 21/06/2014 :

In some schematics for the Royer Oziallator or similar circuits used as a flyback driver you will see the use of 10kOhm resistors parallel to the Zener diodes.

They are supposed to help achieving faster switching times.

I tested different values between 7.5 and 20kOhm and advise against their use.

The circuit often fails to oscillate properly resulting in blown mosfets if you don't act fast.

This is more dominant when high power mosfets are used, with IRF540 or IRF840 mosfets the circuit swings fine but I could not see any benefit at all with the added resistors.

As I also use a welding transformer for the power supply now I realised with the added power the caps on the DC side of the rectifier really need to be quite big - both in value and in size.
If your DC supply breaks down during the times when the AC supply is in the zero range for the current, your caps need to be able to provide the power for the mosfets to switch properly.

A good sign of failure is that the resonance frequency fluctuates and that one or both mosfets heat up too much.

If you are only working with a 12V supply you can use these caps used for the audio equippment in your car.

Don't like the use of batteries but a ggod sized one parallel to your power supply should eliminate all problems until you have a proper power supply.

With about 400 to 600W currently on the input side we are looking in the 2-5kW range inside our tank circuit.

Bigger is better, same for the choke coils.

Check the size and color!

I used yellow toroid cores as they are rated for the frequencies we use - standard black ones or othe colors can work but I recommend the yellow ones.

Most people won't have the means to check for the inductivity of their choke coils, so again: more/bigger is better!

The choke coils should be as close as possible to your tank (the coil with the caps) !

To get the connection of thick wires, choke coils and my pipe I had to use hose clamps as my soldering iron could not supply enough heat.

The wires and coils are soldered onto the thick part of the clamp and slides over the pipe where it is thightened up good for a proper connection.

A word of warning for those that already built a heater based on the standard Royer Oscillator and a coil with center tap:

The choke coils for the normal Royer only need to filter the high frequencies, in this setup they also have to store a lot of energy and "block" the direct connection from the mosfet drain to the positve rail of your power supply!

You need a higher inductance for this project!

Why do the choke need to be of such high inductance?

Well, it all comes down to resonating.

The caps in the tank store the energy which comes from the mosfets and the the work coil.

This work really well in resonance as the resstance of the coil drops to zero.

As the work coil also releases energy into the choke coils we want to re-use it.

Take a normal relay for example.

It turns on and stores some energy, when it turns off this energy is "killed" by a protection diode so it won't harm the circuit driving it.

You can check the high voltages generated from a small relay by holding a one wire of the power supply to the contact while the other is soldered on.

When you remove the wire you can see a little spark, do it very gently and you have a vibrating relay with big sparks on the contact pin.

Same priciple as your ignition coil in the car.

When the mosfet switches off the choke coils release their stored energy back into to tank.

Due to the resonance only tiny amounts of energy are wasted.

This works because the choke coils block the high frequencies while allowing the pure DC to pass.

Update!:

Using 1/2" copper pipe and 5-6 windings you will end up around the 2µH region for your coil.

To make sure the mosfets are save and enough energy is supplied to the work coil, our choke coils must be able handle the currents provided by the mosfets.

Don't be confused with the massive currents inside the tank here ;)

If you make your own choke coil on a donated core you should start with a lot of windings for your initial test if you have no means to test the inductivity.

In my tests on these big yellow cores I killed my mosfets when using only 8 windings of .75mm^2 speaker wire on them.
With 12 windings all was fine and power output good.

Outpur levels decreased once I had over 30 windings on it - so you see the safe region is quite large making it easy to check.

Simply reduce a few windings and check the output level (take the time to heat an object).

Once you notice no significant improvement you are good to go.

Check out how the complete setup works to heat a steel bot of 8mm till glowing red.

The massive noise from the transformer reminds me to clamp the core pack before welding it....

In the background you can see the frequency produce in the work coil in kHz.

Enjoy!

Althoug I stated it before:

Check and double check your connections and soldering !!

Only use quality MKP caps of small size in big numbers, e.g. 20 100nF caps to get to 2µF for the tank.

Use single caps of similar sizes at your own risk as I noticed they all heat up too much due to the insane power levels produced.

If you yre unsure about your mosfets test them this way:

Connect a diode tester (every good multimeter has it) with the negative wire to the source.

Tap the gate for about a second with the positve wire - if it beeps or shows nothing the mosfet might be dead.

Move the postitve to the drain - it should show a quite high value, now use a finger (while still holding the positive on the drain) and place it onto gate and drain - the resistance should go up until it is too high to display - this is good mosfet.

If at any stage your multimeter beeps or shows a short your mosfet is dead.

All connections that carry the currents to and from the mosfets must be oversized!

You don't want the cables to heat up or have a too high resistance, after all we are operating the circuit close to a short anyway (in DC terms).

If you can keep the connection from mosfet to tank as short as possible or mount the mosfets diectly onto the pipe using a brazed on copper sheet - in this case you ust have a proper water cooling to prevent the mosfets from overheating!

Any heat up of the mosfets with no load inside the work coil indicates a problem!

Do your initial tests either with a current limited power supply or for a few seconds only.

Check the temp of mosfets and caps but only when switched off!!

There is still a good chance that some caps fail over time and blow up, a single one is no problem an can be replaced later on, if they all heat up and blow you got the wrong type of cap ;)

Start with small work pieces!

Never hold the workpiece with your bare hands! Especially small pieces can heat up in seconds causing serious burns!

Be aware that with the power level available here you won't be able to melt metal.

Once the so called Curie temperature is reached the steel looses it's magnetic properties.

This is good for forging but bad for melting it ;)

Although non ferrous metals like brass and aluminium can be heated with this setup you should be aware that they have not only a lower electrical resistance but also a much better heat transfer, apart from the problem that they are non-magnetic.

To make full use of the magnetic field and to get even more power out of it my next tutorial will include an update on higher supply voltages and the use of a ferrite mantle on the coil.

Using a higher input of voltage causes several problems:

1. The resulting voltages in the work coil are already considered to be dangerous and having around 1000V in the tank at higher input levels not only requires special caps but also a properly insulated coil and most likely a load transformer to protect the power circuit.

2. Cooling becomes an issue and requires a proper pump and destilled water.

Also insulating the driver part, water supply and coil is a must.

3. Protection of the power outlet needs to be considered. Without this there will be high chance of blowing your fuses all the time :)

So you might "only" see an Instructable on how to make good ferrite mix that won't saturate and focus the magnetic flux trough the work coil.

I want to keep it simple, so before I start a project with more input power a lot of resear will be necessary so I can be sure it can be done safely and by everyone.

Update:

The Instructable for making your own ferrite start here:

Make your own ferrite to improve magnetic flux