CLASS E Magnetic Induction Heater

Discussion in 'The Projects Forum' started by Plasmahunt3r, Feb 16, 2014.

  1. Plasmahunt3r

    Thread Starter New Member

    Feb 6, 2014
    I am working on a circuit, using a CLASS E Power Amp for Magnetic Induction Heater. The circuit works... for 3 or 4 minutes. Then the MOSFET fails (Fails "ON" which is a short, then sometimes catches fire). Here is the schematic for my circuit:

    The Power supply is built from a rewired microwave oven transformer. I produce 13.2vdc regulated (to the breadboard) and 22vdc unregulated to the Choke coil and Class E Amp.

    You can see the .68 mH choke coil with the auxiliary coil. It gets the unregulated 22V power. The red and black wires twisted together. The red is the primary coil and the black is the auxiliary coil. The aux coil is connected to a 20A 1000v Diode (anode on case) which is connected to the negative supply. The combined red/black twisted wires connect to positive in. The single red out goes the the Class E Amp stage.

    The Class E Amp is on the lower right and all it's components fit onto a 1/75 x 1.75 inch PCB. The Induction coil is connected to the PCB.

    Here is the pic of my Class E, Magnetic Induction Test Setup:

    I still have a problem with the circuit. The Amp load keeps rising. Eventually, the Transistor catches fire. I suspect that the Voltage is rising also, but I don't know how to measure it. Maybe I am getting RF feedback from the Induction coil. Don't know the solution for Class E.

    I have tried a temporary solution by limiting the current that can be provided by the rewired Microwave Oven Transformer. I placed a 30-36 UF 250v Start Capacitor on the AC in Hot Wire (Black), which limits the current to the Primary using Capacitor Reactance:

    Capacitor Reactance in Ohms = 1 / ( 2 * Pi * Frequency * Uf / 1000000)

    So a 36 Uf cap on a 60Hz circuit would be:
    1 / (2 * 3.14 * 60 * 36 / 1000000) = 73.72 Ohms

    120Vac / 73.72 Ohms = 1.628 Amps to the AC primary.

    So the 36 UF circuit limits the current to the primary coil to 1.628 Amps. With the Class E running, the secondary current topped out at 8.93 Amps. Limiting the current available to the Class E is a temporary solution for now. 9 Amps at 22 volts is about 200 Watts.

    The current ratio between the secondary and the primary is about 5.5 to 1 (8.93A / 1.628A = 5.485). If I had 10 amps available to the primary, the secondary would produce 55 Amps. More power than I need. So, limiting the current available is probably a wise precaution when using powerful transformers.

    The MOSFET still fails after 3 or 4 minutes. When the MOSFET fails, it remains in an "ON" state (which is a short). I have added a fan to the heatsink thinking of a thermal failure, but it has no effect. I have tried a 180A/100V MOSFET, 20A/500V MOSFET (irfp460), and a 20A IGBT. They all fail after a few minutes.

    Does anyone have any idea how to stop the MOSFET failures???
    Last edited: Feb 16, 2014
  2. kubeek

    AAC Fanatic!

    Sep 20, 2005
    first of all, why do you need a gate drive transfomer, and why don´t you drive the mosfet directly? Did you check the gate waveform and Vds for spikes? You should also measure the current through the mosfet.
  3. Plasmahunt3r

    Thread Starter New Member

    Feb 6, 2014
    I wanted to use a gate drive transformer because I intend to up the voltage on the Class E stage later to 150V and up the frequency to 13.56Mhz. This is needed for RF Plasma.

    I am testing the circuit principles using Magnetic Induction Heater, because the process is the same as in RF Plasma. The lower 32.8KHZ frequency allows me to test and get the bugs out with $4 MOSFET's instead of $90 RF MOSFETS.

    I am using Zener Diode to limit voltage to the MOSFET gate. Zener's should handle the spikes. I tried Gate Protect Zener's using Dual 14v Zeners (Cathode to Cathode) and using a single 15v Gate Protect Zener. The output from the Gate Drive Transformer, using Voltage Doubler is 16VDC.

    Also, without the Gate Drive Transformer, the voltage from the MOSFET Gate Driver Chip is 6.5VDC. A MOSFET needs 10V to be fully on. So I went with the Gate Drive Transformer to up the voltage.
    Last edited: Feb 16, 2014
  4. Experimentonomen


    Feb 16, 2011
    Check out how Metcal did their induction heated soldering station rf generators.
  5. Plasmahunt3r

    Thread Starter New Member

    Feb 6, 2014
    Thanks Experimentonomen. I found the Metcal MX-500 Schematic online and it will take awhile to understand it.

    I see how they are creating the frequency using the Hex Inverter 74HC04. They are tying 5 inverters on the chip to drive N-FET VN0109N5, which in turn runs a Gate Drive Transformer for MOSFET IRF530. I have got frequency creation down. I can use Hex inverters + crystal or I have bought some 13.56 MHZ oscillators from China at .75 each.

    Frequency is not a problem. My problem is in the MOSFET power Amp. The Gate Driver, or voltage rise from inductors, or current rise from inductors are the problem. The Shunt capacitor on Class E limits Voltage rise. I have tried up to 1UF Shunt Capacitor with no change. The Auxiliary Choke coil is supposed to limit Current Rise up to 50% duty cycle. I don't know where my problem is. My circuit works, but for only 3-4 Minutes, then something zaps the MOSFET.

    The IRF530 Power Amp in the Metcal MX-500 is the main power driver. It has a RF Choke, but this MOSFET is running another transformer in series with choke. The MOSFET is not driving the induction coil directly, but is using a transformer to drive the inductor coil. That should stop feedback from the Induction Coil. Interesting.

    I need to Study this section. The gate resistors to ground and gate inductor to ground is interesting. These may be providing Gate Protection. The IRF530 is also using two 330nf decoupling Capacitors on the series transformer instead of a shunt capacitor.

    Thanks. It gives me a working example to study.
  6. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    Could you have a problem in your re-purposed transformer?

    Check the voltage levels there from start to 'problem' and watch closely for any build up of DC bias levels. Perhaps the transformer has slow building magnetic saturation problem leading to unexpected current flows elsewhere in the circuit
  7. Plasmahunt3r

    Thread Starter New Member

    Feb 6, 2014
    Problem Solved. Thankyou for your assistance.

    Experimentonomen telling me about Metcal MX-500 13.56 MHZ Induction soldering station gave me the clue. I found their circuit on the internet and I liked how they used the MOSFET to power a transformer (in series with the choke). This was "NOT" the solution, but it lead to the solution.

    I made up a transformer using a 38mm Toroid and wired it just like a Joule Thief toroid. This was exactly how how I created the choke coil with the auxiliary coil, but instead of the Auxiliary connecting to ground (via Diode), the auxiliary coil became the secondary.

    Here is the updated circuit:

    I ran this configuration for 5 minutes, then I saw the failure. At 5 minutes, the 2.2 UF Mylar capacitor "CRACKED" and I saw arcing inside the crack, then the MOSFET blew. It looked like the capacitor was the cause.

    I tried two 1 UF Mylar Caps in parallel, and it still failed after 5 minutes. I could hear a arcing sound in the caps just before the MOSFET blew.

    I surmised that Ceramic Caps would be better for high frequency and high heat applications. I had three .1 uf 1KV caps in my parts box, so I tried just using the ceramic caps (.3 UF total). The MOSFET instantly blew. (too small)

    I went back to the two 1 UF Mylar caps, and added the three .1 UF 1 KV Ceramic caps in parallel with the MYLAR caps. This worked.

    The frequency result was a little different. The frequency started at 65 MHZ without the steel rod in the inductor acting as the load. When the steel rod is placed in the induction coil, the frequency went to the desired 32.8 KHZ. When I pulled it out, the frequency went back up to 65 KHZ. As the circuit warmed up (after about 10 minutes), the frequency eventually stabilized at 32.8 KHZ. I ran the circuit for 15 minutes, occasionally inserting and removing the steel rod. Each time the frequency went to 32.8 KHZ with rod inserted.

    Now that I know the capacitors were the cause, does anyone have any suggestions for high frequency, high temperature caps. Most of the ceramic caps I can find are only .1 UF. I don't want to stack a 20 caps together.
  8. #12


    Nov 30, 2010
  9. shortbus

    AAC Fanatic!

    Sep 30, 2009
  10. alfacliff

    Well-Known Member

    Dec 13, 2013
    is it possable that you have hit resonance in your output coil? that would increase the rf voltage across your mosfet tremendously. and shorted mosfets are a typical sign of over voltage.
  11. Plasmahunt3r

    Thread Starter New Member

    Feb 6, 2014
    Thanks SHORTBUS;

    I am trying to learn this by trial and error. Since the mylar film fails in a few minutes, it seems that the electrons are punching through the dielectric.

    I will order some MKP (Metallized Polypropylene Film) Pulse Capacitors and see how it works. I have the perfect testbed.
  12. Plasmahunt3r

    Thread Starter New Member

    Feb 6, 2014
    The MKP (Metalized Polypropylene Film) Pulse Capacitors came in. I updated the circuit with one 2.2 uF pulse capacitor and the frequency was stable at 32.8 KHZ and it worked like a champ. To think, all the testing and Transistors I blew, was because of improper type of capacitor. Live and learn.

    Now that that problem is solved, I did some experimenting with the homebuilt Autotransformer I am using.

    It is a 39mm Toroid with Bifilar coils wound just like a Joule Thief Circuit. It reads 1.54 mH on each coil.

    In the first test, the CT is connected to the MOSFET Drain. The Current Draw was around 7.49A without the Steel Rod in the Induction Coil, and around 7.58A +/- .4A when the rod is inserted. Circuit maintained a constant 32.8 KHZ, with/without the steel rod load. Here is the updated circuit:

    The second test was done utilizing the same 39mm Toroid, but I swapped the output to use the CT with amazing results. While each coil still read 1.54 mH, the combined coils read 6.26 mH. A higher mH reading meant the the circuit would draw less current from the power supply, but the CT Amps Out would be double while the Voltage Out would be halved. This is more efficient because Magnetic Induction Heating is mainly a function of Frequency and Current, and not Voltage.

    Using the CT as output, the current draw without the Steel rod placed in the coil is 3.28A. When the rod is placed in the coil the current draw on the power supply jumped to 4.65A, the slowly rose to +5A. Since the CT Out is double, the CT Out provided 10A to the Induction Coil, while the previous version only utilized 7.58A. More Amps to Coil made it run hotter, quicker. I think this is amazing because it utilized less power from the power supply but did a better job of heating. Here is the updated circuit using the CT out from the autotransformer:

    Now that I have Class E Magnetic Induction Heater problem solved, I would like to change the circuit to have an Antenna out. I am not quite sure how to do this. Any Radio guys out there know how to configure an antenna as output. It seems like I could use the CT out with the 2.2 uF capacitor and put AC out to a wire antenna. Don't now for sure if that is enough or how to test it. Any Help would be appreciated.
    Last edited: Feb 25, 2014