3 mosfets in parallel on Induction heater.

Status
Not open for further replies.

gary350

Well-Known Member
3 Mosfets in parallel did not work very long but none went up in smoke or caught on fire. 1 mosfet pulls 30 amps for several minutes so 3 mosfets in parallel should pull 10 amps each for a very long time. 50 amps on 3 mosfets for about 45 seconds suddenly amp meter pegged at 100 amps & started humming like crazy. I tested the circuit 1 mosfet went bad it is an open circuit I wonder if that make the circuit stop oscillating? I had this working good yesterday 1 mosfet at 30 amps I used it for several hours to heat bent & damaged tools red hot so I could straighten them out good as new. I still need to repair the crow bar that has the end broken off it needs to be red hot so I can hammer a new tip on the end. The crow bar is 7/8" diameter solid steel 3 ft long. This was a fun project that turned out to be very useful. Circuit drawing the 21 VDC is a 100 amp transformer filtered DC power supply. 12 VDC is a filtered transformer that produces 10 VDC. This circuit also runs good on a car battery. The circuit drawing is original with a few changes the transformer filtered power supplies, 6 capacitors = 102 KHz and 8 capacitors = 89.9 KHz.

Yesterday I heated up a bent socket wrench handle red hot in 45 seconds with 1 mosfet at 3o amps but today 3 mosfets in parallel running 50 amps did not heat the same wrench red hot in 45 seconds. It was not even starting to turn red in 45 seconds with 50 amps? What is the deal with that?

Just in case someone wants to know why there is nothing soldered to some of the wires, I removed the power supply so I could take this outside in real sun light to get a good photo.

**broken link removed**

**broken link removed**
 
Last edited:
Check if you have any shorts from your Transistor tabs to the heat sinks.

Check if your diodes are still good. They only have a max reverse voltage of 40V. You may exceed that with the inductive kick when coils are present and high current.

What frequency are you measuring on you heating coil?
 

All the tabs are pin 2. All pin 2 are in parallel. Tabs are all connected to the heat sink. Heat sinks weigh 3 ounces each they are large enough not to need a fan and they never get hot. Frequency with 6 capacitors .47uf each = 102 KHz and 8 capacitors = 89.9 KHz. RF voltage on the coil is 70 VAC. I unsoldered all the parts they all test good expect for 1 mosfet. It acted like the circuit stopped oscillating & 1 bank of mosfets were stuck ON pulling 100 amps. These mosfets are rated 55 amps each.
 
I've used a very similar circuit before, even with the split supply so the gates to the mosfets are happy. One thing I see with your parallel mosfet approach is that your entering the line at different "tap" locations. You have to consider with this circuit the voltages produced across the coil, and that even the straight pieces count as part of the coil. With the parallel capacitors it isn't so bad, but there should only be only one entry point for the transistors, not 3. Doing the same thing to the parallel caps will also improve the overall "Q" of the resonant circuit. Simply construct the parallel transistors and capacitors as a BLOCK with only one S,D,and G for the transistors, and only two nodes for the capacitors. The BLOCKs can be copper or aluminum bar. Otherwise you have phase differences that can lower the Q and/or kill a transistor prematurely because it fires slightly early or late which could also cause an avalanche effect.
 
How many times have you ran a thread on this design, and every variant of it you have came across, only to be told that it does not work for high powered applications, just as you find out because it blows itself up every single time?
 
How many times have you ran a thread on this design, and every variant of it you have came across, only to be told that it does not work for high powered applications, just as you find out because it blows itself up every single time?

I have not worked on this circuit for 2 years. Last thing I remember I was told put several mosfets in parallel to get more power. Someone posted a picture of 30 mosfets in parallel rated 3000 watts. 2 days ago I get the circuit working again and it works good on 1 mosfet per side. Today I tried 3 mosfets on each side and it does not work as well as I was told 2 years ago. In the short time I tested it today I could see 1 mosfet is better than 3. I am trying to learn why? It seem logical that it should work. This is really a horrible circuit but it is a fun learning experience that is what I do these days I am retired now I have time to screw around with things like this. LOL.

I started to build a stereo amp but lost interest very quick I have no use if it I have not listened to music in 25 years. It was suppose to be a fun project but how can it be fun with no motivation to want the finished stereo. I need something else fun to build???

I have already built an excellent TV antenna it is over kill to the point no matter how hard it, rains, snows, hails, wind, lightning, storms, tornadoes, signal is crystal clear all the stations are 40 to 45 miles away. Another town 70 miles away crystal clear signal.

At the moment I need to repair my broken crow bar so I am motivated to make the induction heater work. It would be easier to buy a bag of charcoal to heat the crow bar red hot but not as much fun as & induction heater project.
 
Last edited:
Yes, I recall some of it and it comes down to the fact, that for the power levels you want to work with, you need to go to a much higher input voltage and control it with a H-Bridge and dedicated driver circuit given these simple self oscillating push pull type circuits just don't scale up very well due to the inherent limits that the sizes and types of semiconductors you are working with have.

They are fine for a few hundred watts and that's it. Beyond that power level it's necessary to go to a higher voltage input and a full H-bridge in order to handle the added power while also having better control of things as well.
 
It's worthless to parallel mosfets when using a self oscillating gate drive circuit like you are using, as only one mosfet will actually get turned on. This is because each mosfet has a slightly different gate-source turnon threshold voltage. With your circuit, once the first one fires the drive voltage drops, so the others, with slightly higher turn on thresholds, never turn on.

The way to drive parallel mosfets, is with a gate driver that hits all the gates with a voltage pulse greater than the G-S turn on threshold, typically 12 volts. That voltage pulse needs to have a low source impedance, and be able to source and sink enough current to charge and discharge the total mosfet gate capacitance quickly. That way you know that they all turn on and off at exactly the same instant.
 
One observation about the operating voltages in your circuit.
You have a 10 Volt supply in series with a 21 Volt supply. This puts 31 Volts at the left end of the 8mH inductor and, in a DC sense, the same 31 volts at the center tap of your work coil.
Now, I'm sure that you know that a center tapped work coil kind of acts like an electrical see-saw. When one end is pulled to GND by one mosfet, the other end flies up to twice the center tap voltage. And the other end of the coil (62 Volts) is connected to the drain of the other mosfet (60 Volts)
And those are just the nominal DC values. When you've got switching currents flowing through inductors, you will have transient voltages on top of the DC voltages.

What kind of voltage spikes do you see on your oscilloscope when you probe the drain points?
 
A huge problem comes when you start the circuit without a steel tool in the coil. Then he coils are air core and about 1% he inductance of when an iron bar is filling most of the space. This higher inductance lowers the oscillation frequncy oscillation and energy transfer to the workpiece is greatly reduced at 10kHz - (usually inductive heating is at 30KHz to 150kHz).

What else happens at lower fequency? The inductive reactance is much lower at low frequency. But your wave shape will change if your cap bank drains too quickly (before a complete half-cycle. At that point, you have pure DC current flowing (100+ amps) based only on the dc resistance of the coil's copper wire and nearly zero inductive reactance.

Use an H-bridge as tcmtech suggests to eliminate the presence/absence/cross-section diameter/alloy of a work piece from controlling oscillator frequency, energy transmission to the workpiece and heating.
 
Last edited:


That pretty much how I see it. With a full H-bridge set up with the induction coil in series with the capacitor bank in the center it's very hard to kill that type of circuit plus with minimal extra components it very easy to add full independent variable frequency and PWM control that can be simplified down to just two pots which would make the circuit easily tuneable to whatever the load requires.

There are a number of circuit built around the LM494 and older LM3524 inverter driver IC's that can do wide frequency band operation with wide independently controlled ~5 - 95% PWM duty cycle output. Add a pair of high/low side driver IC's like the IR2110 or such to properly control the H-bridge switching devices and that's basically it.
 
Also, how are you keeping an iron crowbar perfectly in the center of your coil without touching the coils? I don't imagine there is any enamel left on the superheated coil wire to insulate it from being shorted out if the crowbar does short out several turns of the coil.
 
Good point and just one more example of how these overly simplified induction heater circuits do not scale up well.

When induction heaters get over a few hundred watts they really need to have hollow tubes for the coil sets so that a non conductive coolant of some kind can be pumped through them otherwise they will go into thermal runaway and melt right along with the material they are heating up.
 


As mentioned above, his diodes are only rated at 40V peak reverse voltage so not a good pick with the high inductive kick he will see there.

So many things wrong with this picture.
 
TV manufactures cut pin #2 off the mosfets then use the heat sink tab as pin #2 so I tried that & now I can pull 50 amp from these mosfets. Before the mosfets burned up at 35 amps. Very interesting learning experience. This 3/8" bolt turned red hot in a few seconds. 8 capacitors work better than 6 capacitors. I think I have gone about as far as I can on this circuit time to play with something else. Someone mentioned the H circuit, I need a circuit drawing for the H circuit using mosfets? Someone mentioned higher voltage I can rectify 120 VAC = about 170 VDC and voltage doubler = 340 VDC. I know I have a few 1500v 5a mosfets & might have some in the 400 or 500 volt range.

**broken link removed**

**broken link removed**
 
it would be better if you used a crystal controlled oscillator. the ideal situation would be to run the heater within one of the HF ISM bands. within the ISM bands, induction heaters and wireless battery chargers are subject to no restrictions of output power, but there are limits on the field strength of the radiated signal (including "conduction" radiation from AC power wiring in the building). if the device is operated outside of the ISM bands(6.780Mhz(+/-15khz), 13.560Mhz(+/-7.0khz), 27.120Mhz(+/-163khz)), you might get a visit from your local FCC officer.

you would be surprised how far the RF from just the coils and workpiece can travel.
 

I think a solid iron core will convert any signal to heat by the time you hit 150kHz. I don't know if any induction heaters running in megahertz. Why do you want to go that high? Just for getting the RF police to knock on your door?
 
Go with a PWM controller H bridge and a good MOSFET driver for proper FET triggering to kill off the parasitics and frequency dependencies on the work piece. A TL494IN with a thermistor feedback loop can protect your FETs from cooling failures (like a fan fail).

I deliver around 800A pulses via 3 MOSFETs with that approach, an added active snubber dissipating backEMF power into an incandescent lamp allows the FETs to run cool and stay in the SOA as per spec, and not avalanche.
 
Status
Not open for further replies.
Cookies are required to use this site. You must accept them to continue using the site. Learn more…