Simple Powerful Manual Tune Induction Heater

Induction heating is the fascinating ability to wirelessly heat metal or graphite objects without the use of an open flame and with minimal wastage of heat to the surroundings. It is not a new phenoma and has been around for more then 100 years. It is widely used in the smelting and automobile industry because it can easily be controlled, and scaled.

Having  read the tutorial and built a phase lock loop (PLL) induction heater based on the work of Jonathan Kraiden (mindcallenger.com), I wanted to come up with something straightforward and easier to build for beginners in hobbyist electronics. As I am also an electronics newbie I felt that this project would be fun and give good insights into the fundamentals of how induction heaters work.  In addition to the PLL heater mentioned above, I have also built many  Mazilli or Royer type induction heaters including cheap Chinese made Mazilli drivers available from Ebay and Alibaba. I found that although these are good and easy to use, they are prone to failure as they are limited by low voltage range of operation and relatively low power for what I wanted. They are also difficult to control in terms of the amount of heating. I wanted to make something that would be more reliable and controllable and not result in a pile of blown transistors! Basically, I wanted a manually tuneable setup that would run off rectified mains instead of expensive switching power supplies. The basic construction would be to take higher voltage at lower current (for which MOSFET transistors are good at handling) and convert this to lower voltage at much higher currents of the order of 100s or 1000s of amps and high frequency AC typically 20-100 kHz. This is passed through a coil of a few turns of copper (referred to as the “work coil”). If a piece of iron is placed in the work coil, eddy currents are induced in the iron piece (also referred to as the “work piece”) in such as way that the work piece acts as a shorted 1 turn primary coil. Due to the ratio effect of a transformer, huge currents of the order of many hundreds or thousands of amps flow in the workpiece which results in heating of the work piece due a combination of internal resistance (IR2 heating) and hyteresis (due to effect on the random array of magnetic dipoles in the iron work piece changing direction many times per second at high frequency). Due to the high frequency, current flow takes place preferentially in the very surface layers of the work piece and work coil also known as the “skin effect”. This further increases the effective resistance of the workpiece resulting in even more I2R heating. As the skin effect is also taking place in the work coil, there is energy loss in the work coil in the form of heat in the surface layers of the coil. Thick copper tubing which has outer and inner surfaces or Litz wire (multistranded insulated wire with each strand having electical insulating coat) increases the effective surface area of the work coil reducing energy wastage as heat loss. Litz wire is employed in induction cooktops for the same reason. Higher frequencies have more skin effect with more surface heating better for heating smaller work pieces. Lower frequencies have less skin effect and are better for  heating larger work pieces.  This heating can take iron, for example, from room temperature to  red heat to orange heat to bright yellow heat beyond the Curie point (the point at which the iron or steel work piece looses heating from hysteresis due to loss of its ferromagnetism). In order to be able to achieve further heating and melt the iron work piece (including non ferromagnetic metals such as copper, silver, gold and aluminum), it is necessary to attain much larger currents to overcome the lack of hysteresis. The aim of this project is to build such a simple small tuneable induction heater that would be able to melt small quantities of these metals.

With PLL induction heaters referred to above (for a good detailed tutorial on these see work by Jonathan Kraiden at http://inductionheatertutorial.com/), they are self tuning to a point. The PLL works from a voltage controlled oscillator. As the frequency of resonance of the induction heater changes as a workpiece is placed in it. This would result in loss resonance and corresponding loss of heating as maximum heating occurs when the tank circuit of the induction heater is in resonance. With PLL, the voltage across the tank is fed into the voltage controlled oscillator of the CD4046 PLL chip to maintain maximum tank voltage. However, typically when a metal such as iron achieves the Curie Point, the change in resonant frequency is beyond the range of the PLL and the circuit goes out of resonance and heating is lost. Some electronics hobbyists in the fields of induction heating (for example Jonathon Kraiden at mindchallenger.com) have gotten around this by using a microprocessor to keep the circuit in resonance as well as to periodically detune the circuit to lower overall current flow in the power transistors  saving them from destruction in the event that the current increases beyond their maximum tolerance. Many people have helped advance the fields of solid state induction heating – these include people such as “Neon John”, Bayley Wang at MIT, Jonathan Kraiden, Richie Burnett (UK), and many others.

Completed Project-Small Powerful Desktop Sized Unit:

If you plan on building an induction heater such as the one shown here, a word of caution first: This project utilized unprotected mains, high voltages, and high currents with serious risk for injury or worse if not performed in experienced hands. I take no responsibility for injury or worse resulting from any work described here. The work described herein is for academic and scientific interest only.

The project described here can be divided into 3 sections: 1) Variable Frequency Oscillator, 2) Half bridge inverter with either MOSFETs or IGBTs and 3) Tank circuit.

Here is the components list for this project by section:

VARIABLE FREQUENCY OSCILLATOR:

1. 8 pin IC socket x 3
2. 14 pin IC socket x 1
3. 1kOhm 0.5W resistor x 2
4. 10nF, 50V ceramic cap x 1
5. 1 nF, 50V ceramic cap x 1
6. 20k 10 turn pot x 1
7. 100nF, 50V ceramic cap x 4
8. 47uF, 35V electrolytic cap x 4
9. 1 uF ceramic, 50V cap x 4
10. 1000uF, 35V electrolytic cap x 1
11. LM7815 x 1
12. LM7805 x 1
13. small heatskink for LM7815 x 1
14. Aluminum project box (optional) x 1
15. 120VAC to 19-26.5V stepdown transformer (radioshack) x 1
16. full bridge rectifier 50V 2A x 1
17. 1 – 1.5 inch diameter green ferrite toroid x 1 (for GDT)
18. perfboard x 1
19. thin solder (fluxed with rosin) x 1
20. soldering iron 20-30W x 1
21. UC37321 x 2
22. UC37322 x 2
23. NE555 x 1
24. 1N5819 (shottky) x 4
25. hookup wire x 1 roll
26. 74HC14 hex inverter x 1

HALF BRIDGE INVERTER AND BUS CAPACITORS:

1. IRFP260N x 2
2. 6.8 Ohm, 2W resistor x 2
3. 1N5819 (shottky) x 2
4. 1N4744A (15V zener) x 4
5. U860 ultrafast diode x 2
6. 1.5KE440CA bidirectional TVS (or unidirectional) diode x 1
7. 5uF, 450V polypropylene snubber cap x 1
8. 100kOhm, 2W resistor x 2
9. Aerovox RBPS20591KR6GNZ 1 kV, 2uF snubber cap used for the DC blocking cap (available from Eastern Voltage Research, NJ USA) x 1
10. 400-450V 1500 uF electrolytic caps x 2
11. 20-30Amp toggle switch x 1
12. 20-30Amp shunt ammeter x 1
13. Ferrite Toroid (large) for coupling transformer x 1 (or more)
14. 16 gauge insulated stranded wire for 22 turns on coupling transformer
15. large aluminum heatsink x 1 (Ebay)
16. 35-40 Amp, 400V full bridge rectifier x 1
17. 30-40 Amp fuse with fuseholder or 30-40A breaker x 1
18. terminal screw connectors x 4 (for easy removal of IGBT mosfets)

TANK CAPACITOR – 6 TURN WORK COIL:

1. 3/8″ soft copper tubing roll from hardware store x 1
2. 3/8″ copper tubing sweat connectors x 2-4
3. fountain pump x 1
4. latex tubing to connect pump and copper tubing x 1 roll
5. plummer solder and flux x 1
6. copper tube cutter x 1
7. 12 x tank capacitors 1200V, 0.33uF (available here: https://www.ebay.com/itm/10PCS-New-BM-Capacitor-MKPH-0-33uF-630VAC-1200VDC-for-Induction-cooker-P-30-5/272271654633?hash=item3f64a7bee9:g:PJ4AAOSw9eVXXQsg) as well as other Chinese sites on Ebay
8. 47kOhm resistor x 1
9. green LED x 1
10. UF4007 ultrafast diode x 1

In this project, we are using a 555 timer chip (Figure 1) in astable mode to generate a square wave 50% duty cycle signal. This is achieved with a variable resistor (0-20kOhm) to generate variable frequency ranging from 35kHz to 132 kHz. This range is great for a variety of different sized induction heaters. Although the 555 produces a square wave output, it is not a “clean” square wave. In order to clean up the square wave, the output of the 555 is fed into a 74HC14 hex inverter which outputs a nice clean square wave signal. This is fed into the inverting and non inverting inputs of UC37321/22  MOSFET driver chips. These chips are powered at 15V for a 15V output. Because the chips are run in continuous mode, instead of pulsed mode as in solid state Tesla coils, they tend to heat up. To reduce overheating of these chips, 2 of each of the chips are stacked in parallel by soldering their legs together. Small strips of aluminum can be glued onto these parallel connected chips for even more cooling.  The output of the chips is passed through ceramic capacitors which function as DC blocking capacitors. Typically 1-2 uF is all that is necessary (Figure 1).  The caps should be rated for at least 50V. A gate driver transformer is wound on a ferrite toroid which is wound 1:1:1 with 10-15 turns of trifilar wire. One of the trifilar windings is the primary of the GDT connected to the output of the gate driver UC chips. The two remaining windings  turn on the the gates of the MOSFETS; one of the 2 is turned on while the other is turned off. This is achieved by reversing the connections of the output of the second of the trifilar windings. Before connecting the GDT outputs to the MOSFET gates, check the waveforms of the GDT secondaries to make sure they are square waves or as close to square waves as possible. This may involve using a different or bigger GDT or increasing or decreasing the number of windings on it. There are many different types of ferrite material available. The green ferrite toroids work best. The yellow or light green powdered iron toroids found in computer power supplies give a very poor gate drive signal and are unsuitable for this purpose. The only disadvantage of using a 555 timer chip in this manner is your duty cycle is fixed at 50%. As the 555 allows the possibility for both MOSFETS to be turned on at the same time known as “shoot through”, there is a potential for both MOSFETs to be turned on at the same time shorting out and destroying the half bridge. However, despite this possibility with the 555 chip, I have never observed the effects of “shoot through” after hours of operation of the induction heater!

Figure 1 (555 Timer based driver circuit- NB if there is too much heating of the gate drive chips, try reducing the value of the ceramic capacitors between the output of the gate driver chips and the GDT from 2 uF to 0.1 uF ). 7HC14N is a typo error and should read 74HC14. CLICK HERE FOR HIGHER RES IMAGE

The negative rail of the driver section in Figure 1 should also be grounded to mains Earth ground as there would be fluctuations in the output as you move your hand or other objects near to the driver section during manual tuning if it were not properly grounded as I have experienced. The driver section should be isolated from the high power portion of the circuit preferably by an aluminum box. The box should also be grounded to earth ground. Important- note that the 15V LM7815 voltage regulator MUST be heatsinked otherwise it would overheat and shut down. Heat sinking can be achieved either by adding a screw-on heat sink to the regulator or more conveniently by screwing the LM7815 into the wall of the project box. As the metal back of the regulator is the negative rail, it allows easy grounding of the negative rail as well as the box to earth ground simply by attaching earth ground to the outside of the box.

The half bridge inverter (Figure 2) is powered by rectified AC from a variac. It will also work with rectified AC without the large smoothing capacitors if there is no need for a voltage doubler.

The rectifier can be wired as a voltage doubler allowing a maximum DC voltage of 120 x 1.42 x2 = 340V across the inverter or 170V on each leg of the half bridge. You want to use at least a 2 or 3 KVA variac. However, will work with a 500VA variac but with reduced power.

Figure 2 (Half bridge inverter with protection diodes, snubber cap,  and 6.8 Ohm gate resistors) CLICK HERE FOR HIGHER RES IMAGE

The IRFP 260N MOSFETS are good for 50A continuous and 200A pulsed drain current and 200V drain to source voltage giving them tremendous power dissipation (which definitely needs a heat sink).  For even more power, the IRFP260N MOSFETS can be replaced with FDH44N50. As all of these MOSFETS have inherent gate capacitance which when coupled with the gate drive transformer secondary inductance can lead to severe ringing oscillations on the gates during switching leading to damage to the gates. This ringing is dampened with the 6.8 ohm gate resistors. Residual charge on the gates of the MOSFETS is quickly drained by the Shottky (1N5819) diodes connected in parallel with the gate resistors (See Figure 2). If the gates are run at 15V, they are operating in the plateau region of the voltage current curve where heating of the MOSFET is reduced. This is further reduced by ZVS or zero point switching which is achieved with correct tuning (see later). The opposing zener diodes across the gate and the source prevent voltage spikes across the gate greater then 15V and therefore protect the gates from damage due to voltage spikes. The ultrafast free wheeling diodes (U860) across the source and the drain of the MOSFETS protect them from reverse voltage surges. The snubber capacitor rated at 450V 5uF and the the TVS typically rated 400-440V (transient voltage spike) diodes do just that in protecting the bridge from voltage spikes. Large electrolytic capacitors help to smooth the rectified AC and act as a charge reservoir. They also function to allow the setup of the voltage doubler from the full bridge rectifier. The rectifier and the IRFP260 transistors should be heatsinked with addition of a fan for added cooling of the heatsinks. The output of the inverter is fed into a coupling transformer preferably made of ferrite and with about 20-22 turns of 16 gauge insulated stranded wire wound around it. These windings act as the primary windings for the tank circuit (see Figure 3).  The tank is essentially a 1 turn primary  in series with the work coil (in this case 6 turns x 1.5 inch diameter of 3/8 inch copper tubing) and in series with the capacitor bank (Figure 3). The capacitor bank consists of 12 parallel connected 0.33 uF 1200V MKP capacitors (rated for use with induction heaters and Tesla coils). The use of good quality MKP or other polypropylene capacitors that can handle high current, high voltage, and high frequency is essential for the induction heater to work. If the capacitors are wrongly rated, they would heat up and explode and there would be little or no energy transfer to the workpiece in the 6 turn work coil.

There is a 1kV 2 uF  DC blocking capacitor connected in between the output of the inverter and the ferrite coupling transformer (see Figure 2 for Inverter circuit). In an initial run I used a wrongly rated DC blocking capacitor (5uF at 275V instead of 2uF 1000v) and here is the result:

Here is replacement with the correct blocking cap (Aerovox) rated 1kV 2uF which has functioned flawlessly in several runs:

The 4 electrolytic capacitors are connected series parallel for the voltage doubler portion of the circuit. Note also the 100 K resistor across this capacitor bank to drain off excess charge when the unit is not in use.

The resonant frequency of the tank circuit can be found by connecting it to a signal generator via a 10k resistor and scoping the voltage across the tank until resonance is attained.

In a typical run, the AC on the variac is set to a low value something like 40V. The driver circuit is tuned starting with a higher frequency ABOVE resonance of the tank circuit and slowly adjusting to a lower  frequency until the resonant frequency of the tank circuit is achieved. At this point the green LED indicator light across the tank will light and there will be an audible humming on the variac. The tuning is continued until maximum current flow is seen on the shunt ammeter (Figure 2).

It is important to start at a higher frequency above resonance and slowly decrease the frequency until resonance is obtained and heating of the workpiece occurs. The reason is if tuning is started from lower frequency to higher, there would be severe ringing voltage spikes between source and drain of the MOSFET below resonance which could lead to MOSFET failure.

It is preferable to start tuning with the workpiece (the piece of metal to be heated) already in the work coil. Absense of the workpiece in the coil will result in very high currents flowing across the MOSFET drains which could stress the MOSFETS. When the unit appears to be in tune (maximum current with the work piece present and indicator light on), the voltage is increased slowly on the variac to achieve the desired level of heating. Retuning is performed as necessary with increasing voltage on the variac to maximize heating of the work piece.

Figure 3 (Tank circuit setup showing indicator light for in-resonance status) CLICK HERE FOR HIGH RES IMAGE

If for example 20A is flowing through the MOSFETS during a heating run, this translates to 22 x 20A flowing in the tank with a 22 turn coupling transformer ie 440A.

Using a clamp meter to measure 355 A flowing in the tank circuit:

With a 6 turn work coil, the amount of current flowing in the workpiece would be 440 x 6=approx 2.6kA! The total capacitance of the capacitor bank consisting of 12 x 0.33uF caps =4uF. With a 6 turn 1.5 inch work coil, this oscillates at 60kHz approx. This is the frequency at which the circuit is in resonance and zero point switching of the transistors occurs. The frequency can be changed on the fly to accomodate a larger work piece etc. Typically a 10 turn 20kOhm trimmer on the driver circuit (Figure 1) would be preferred for finer tuning. When the desired amount of heating is achieved, the power transfer can be reduced and shut down either by detuning, or decreasing the voltage on the variac. The work coil is kept cool by allowing water to flow through the copper tubing in a continuous loop using a fountain pump or simply from a hose connected to the tap. If the ferrite coupling transformer is chosen right, there is little or no heating of either the coupling transformer material or the primary winding of the coupling transformer. The capacitors are connected (Figure 3) in such a way ( I soldered mine onto copper bus strip) such that each of the capacitors in the cap bank contributes equally to the total current flow to avoid excessive heating of any one of the capacitors. Hence the work coil is connected to opposite ends of the capacitor bank as shown in Figure 3. Further cooling of the capacitors can be achieved with forced air cooling from a computer case fan.‭

Lets go into details of construction of the Capacitor bank.

First you would need good quality capacitors rated for use with induction heaters and tesla  coils as pointed out earlier. Mine are 1200V 0.33uF purchased from Ebay (https://www.ebay.com/itm/10PCS-New-BM-Capacitor-MKPH-0-33uF-630VAC-1200VDC-for-Induction-cooker-P-30-5/272271654633?hash=item3f64a7bee9:g:PJ4AAOSw9eVXXQsg):

You would need 2 copper strips approximately 1 to 1.5 inches in width by 12 -15 inches in length by approximately 0.1 – 0.2 inches in thickness. This can be easily bought as a roll of copper strip. I found mine on EBay:

Here is how I soldered the caps and the 3/8 inch copper tubing onto the copper strips for a nice looking setup.

I used a propane torch and plummers solder to achieve the electrical joins. Be careful not to burn the capacitors with the torch while soldering! The 3/8 inch copper tubing was cut with a copper pipe cutter and joined with copper sweat fitting using plummers solder to seal the join water tight. Copper pipe sweat fitting:

The 6 turn copper work coil was fashioned by first calculating the correct length of 3/8″ copper tubing to give a 6 turn, 1.5 inch diameter coil and then adding an additional 20 to 24 inches to have 10-12 inches straight copper tubing extra on each end of the coil. The required length of tubing is the cut (using a copper tube cutter and not a hacksaw) from a roll of soft copper tubing available from any hardware store. The midpoint of the cut piece is marked with a marker pen, one end of the cut piece is capped and then the cut piece is then filled to the top with sand. The sand is tamped down periodically by tapping the tube on the ground to ensure complete filling with no air gaps. The other end is then capped once the tube is filled completely with sand. Using a piece of 1.5 inch outer diameter PVC pipe or wood held firm with at least 2 vice grips, the mid point of the sand filled pipe is placed over one end of the 1.5 inch PVC pipe or wood dowel and 3 turns are wound on either side of the marked midpoint to give a total of 6 turns with equal length 10-12 inch end pieces on the 6 turn coil. The sand prevents kinking or buckling of the tubing during winding the coil. Winding the coil is not as easy as it may seem and a practice run with a small length of sand filled copper tube may be helpful before you wind your 6 turn coil. The sand is removed from the finished coil by removing the endcaps and tapping it while the sand falls out of it. The last traces of sand in the coil can be removed by blowing them out by mouth or with compressed air.

The driver circuitry was made using regular perfboard and through hole components. For the IC components, IC socket holders are highly recommended! As mentioned earlier, grounding of the negative rail of the the driver circuitry is necessary for stable operation of the driver.

The inverter circuitry for this project uses a half bridge inverter. I used terminal screw connectors to easily change out the IRFP260’s instead of soldering them in place. Although a low inductance laminated bridge with copper bus bar or strip is optimal, it is not necessary for this project as the currents flowing—-

Video of heating up chunks of steel:

Video of a similar larger unit using a full bridge of To247 sized IGBT modules instead of the IRFP260 to heat chunks of iron. Other then the IGBT’s the basic construction is the same. The Al2O3 crucible cracked during this run and the liquid iron started pouring out:

contd

—-through the Half bridge and MOSFETS are not as high as those typically seen in solid state tesla coils (SSTCs and DRSSTCs). The high currents are in the galvanically isolated tank circuit.  Note the nice big heat sink for the MOSFET transistors. As the backs of these MOSFETS are metal and are electrically communicating with their drains,  electrically insulating thermally conductive heatsink pads are necessary to avoid shorting out the MOSFET drains. Plenty of thermal grease is required in between the MOSFET and the front of the pad and in between the back of the pad and the aluminum heatsink. Mica is the best material as it has superior heat transfer and is also a great insulator:

Note the green GDT transformer and the computer case fan. Make sure or the secondary leads from the GDT are twisted and not too long reduce stray RFI interfering with the gates of the MOSFETS:

Ferrite coupling transformer with 22T of 16 gauge stranded wire:

Heating a thick bolt to near white heat. It later melted and started dripping all over the concrete floor:

Shunt ammeter connected to the output of the voltage doubler to fine tune maximum resonance and heating:

Variable frequency trimmer with point of resonance marked. The oscillator circuit is behind the grounded aluminum sheeting:

Using a 40A breaker instead of a fuse. A 30A breaker might have been better:

Driver circuit with aluminum strip heat sinks glued onto the doubled up UC373XX driver chips to help keep them cool. Note the perfboard and the IC sockets:

Heating a large nut. The nut was later melted by cranking up the voltage:

Full bridge rectifier cooled by the same heatsink used for the MOSFETS:

Here is a YouTube video showing operation of the unit including failure of the DC blocking capacitor which was incorrectly rated:

https://youtu.be/nIPuZ1WPtc0

I hope this short tutorial has been helpful for those interested in making a small but powerful tuneable induction heater. This unit has proven reliable without any blown transistor (yet) and has proven easy to operate. For even more power, the IRFP260N MOSFETS can be replaced with FDH44N50  which are beasts in terms of power handling considering that they come in the same small To247 package as the IRFP260N units. This project was satisfying to come up with a small unit capable of melting a variety of metals, and in terms of learning some of the electronics behind these fascinating devices.

If you are interested in other electronics and a whole host of other interesting tech projects, please visit my youtube channel at

https://www.youtube.com/user/skippy38305

and don’t forget to subscribe!