How do I make my induction heater get 200 degrees hotter?

[shortbus=]

I ask the question once, what do I need to make a high power induction heater and got no reply. I guess no one knows the answer. I bet a 20KW industrial industion heater does not have 50 mosfets in parallel.

There you go, you answered your own question. Watt's are the key to heat. So many watts = so much heat.

 

[Tony Stewart]

Your reduced coil experiment lowered the load impedance , when you should be raising it.


I estimate your half bridge impedance based on 640 mm coil to centre from your photo the 10AWG wire length appears to be 15" feeder & 25" coil = 40" ( ~1 m @ 0.00328 Ω/m)
= 32.8 mΩ or 17 mΩ per half bridge.

How much power is lost in the + 8uH supply shunt ? This gives a high common mode impedance which limits power.

Something is wrong with your schematic resonant frequency for 40,000 uF 1 uH is ~1 kHz ?? Far too high uF for inductive resonance. and too high CM choke impedance after tthat is fixed. This is wht MOSFETs smoke with only a couple turns less. THere must be some parasitic resonance not characterized in the layout and components.

So what impedances do you get?

I get per half bridge.;

R copper = 17 mΩ
L coil = 0.75uH total or 0.375uH per leg
Zcoil = 0.375uH *2πf @ 64 kHz
Zcoil= 17mΩ +j 151 mΩ
Xc = 1/(2πfC)
thus @64kHz C=0.04F
Xc = 62 uΩ
SRF= 1/(2π√LC)
64 kHz Zload= 17mΩ +j 151 mΩ
Q = L/R

You ought to be driving the MOSTFET gates with rectangular pulses of width is << 1/2 of the Self Resonant frequency and resonating the coil, but non-linear R//Diode cross coupling drives the gates with quasi sine peak pulses and leads to poor efficiency, it also leaves the MOSFETS open to parasitic oscillation. According to my calculations 0.375uH + 0.04mF resonates at 1300 Hz not 64kHZ

Another clue to parasitic resonance is that the resonant frequency should drop significantly when Iron core is inserted as the coil inductance should increase SIGNIFICANTLY compared to an air core and not simply ~ 20% lower

 

[ChrisP58]

Something is wrong with your schematic resonant frequency for 40,000 uF 1 uH is ~1 kHz ?? Far too high uF for inductive resonance. and too high CM choke impedance after tthat is fixed. This is wht MOSFETs smoke with only a couple turns less.

The 40,ooouF of electrolytic caps are filtering the AC line ripple. The resonant caps are the 8 0.47uF film caps adjacent to the coil.

 

[gary350]

Your reduced coil experiment lowered the load impedance , when you should be raising it.

That is exactly what I was thinking when I did it. I was thinking it might over load the mosfets but I need to see what happens. Now I know the coil must stay above 1 uh. Each problem solved is another learning experience.

The original circuit calls for 2 uh choke, that worked fine when max current was 8 amps. Higher current blew up all the power supply capacitors. 4 uh choke helped and 8 uh was better.

Maybe next experement should be a 2 uh LC coil.

 

[Tony Stewart]

oops, That's more like it.
3.76 uF @ 64kHz = 661 mΩ

Still about 2 or 3 decades below the frequency it needs to be for low conduction loss.

Consistent with what I have seen at Perkin Elmers factory has a similar size coil capable of > 1kW operating at 27MHz or between 1~500MHz
https://docs.google.com/viewer?url=patentimages.storage.googleapis.com/pdfs/US8263897.pdf

You will want to use best plastic caps for low ESR and reduce significantly. Adjust current limiter before full load test.

But you may not get these MOSFETs to run more than a few Mhz , but in any event, high f is much better here and larger diameter increases L impedance so that a higher voltage can be used. Gate R must be reduce to 5 Ohms with a bipolar buffer.

Thats my 2 cents.

 

[Tony Stewart]

Without the iron core you should get a Q of several hundred 500 tops. But frequency will drop when loaded with iron more significantly, which is why a fixed drive frequency at the loaded L rather than free air L is better than a self resonant circuit one as Q amplifies the voltage and may exceed the cap ratings. A VCO with pushpull complementary bridge driver IC will do the job for high power than the one being followed. 64kHz is too low Z and drives up conduction losses. Too high may cause eddy current losses and skin effect losses. But increasing at least x10 would be my goal. which may scatter into AM band, which is why license free f of 21? MHz is used for industrial RF power.

These calculations show how raising the impedance and Q is done by doubling the Coil diameter then raising f by x10 .

1" diam ~ 0.25m ( Usually inner radius is used when skin effect occurs. )

https://www.coolcircuit.com/tools/single_layer_coil_calculator/

As F goes up so does transition pulse current so driver must be improved to at least 5% of output current, although only for a short duration with a square drive to 5 Ohm // diode to turn off quickly. Dead time becomes critical > 1us.
https://www.fairchildsemi.com/datasheets/FA/FAN3278.pdf for example is too small. ( 1.5A low)

At these frequencies, RF can be dangerous if not shielded with a microwave oven like screen ( although holes may be bigger.but is useful to torch aluminum rod.

 

[jjw]

That is exactly what I was thinking when I did it. I was thinking it might over load the mosfets but I need to see what happens. Now I know the coil must stay above 1 uh. Each problem solved is another learning experience.

The original circuit calls for 2 uh choke, that worked fine when max current was 8 amps. Higher current blew up all the power supply capacitors. 4 uh choke helped and 8 uh was better.

Maybe next experement should be a 2 uh LC coil.

The original circuit at the beginning of January had 2mH choke.

 

[Tony Stewart]

I dont understand why I can not get a hotter part? I need the part to be about 200 degrees hotter. The temperature I have is slightly too low for metal forming.

1) start using metric ('C)
2) which color or how hot was it ( best case)?
3) what do you need?
4) Do you know about Curie Temp for iron is Fe 1043(K) or -272 = 800'C (1400'F) where Relative permeability of iron core drop abruptly then you are relying on eddy current losses in iron core which should be great. ( & hot)

 

[Tony Stewart]

The original circuit at the beginning of January had 2mH choke.

2mH is definitely too big, but 10uH would be nice with about 2 Tesla

Multi layer coil designer
https://www.datavoyage.com/coilmaestro/coilmaestro.zip

 

[Tony Stewart]

Only cold rolled grain oriented steel (CRGOS) can handle 1.5Tesla without saturating commonly used in power transformers. When it saturates, losses increase up rapidly.

As I was saying around 10 uH would be a sweet spot for me but with 100 uH. @ 30A you can reach 1.5Tesla. With a much higher impedance coil at higher f resonance, you can then get as much higher V across the coil rather than the CM choke and into the iron core.

The 8mH series shunt needs to be bypassed with a large cap and the smaller cap to resonate with coil at least 10x f. or 500kHz Large high V cap but low uF and high V MOSFETs.

As iron reaches Curie point L drops so f must increase otherwise current drops.

https://www.calctool.org/CALC/phys/electromagnetism/solenoid

 

[gary350]

Looks like the fireworks a few days ago caused a chain reaction that killed several other parts. Both mosfets are dead. The 7812 and both its capacitors are dead. Not sure if diodes are good. Power supply seems ok. This started out to be a FUN retirement project. Then it turned into a FUN challange. Then it turned into hope for a useful 2000 watt induction heater tool I can use in my shop. Now it seems like a lost cause, it will never be a good useful industrial quality induction heater. Its not FUN anymore too much like work. I quit.

I might consider building another industion heater if I can get some industrial heater parts to work with.

Next project. Put new heat sink compound on my computer CPU. It does not sound like FUN just boring work.

 

[Mosaic]

Gary:
When dealing with power electronics from an experimental point of view, when you boost the power in one part of a circuit, the next weakest link can breakdown and take the neighbourhood with it.

It always costs a lot of time, $$$ for new parts and frustration.

The lesson here is for certain projects doing an upfront set of goals, stretch goals and budgets can keep you focused and prevent feature creep.
Feature creep is the enemy of finishing a project, there's always a bit more you can do or a bit more power you want.
I do a lot of work in high current, fast rise time, capacitive driven FET pulsing..to the tune of about a kiloamp. I have burned a handful of FETS and made steady improvements. In this field, circuitry construction matters, opto isolation, star grounding and star pulse sourcing matters. Proper wire gauges, time domain reflector based wiring lengths for impedance matching, proper materials, soldering and connectors as milliohms matter.You don't want voltage bounce or the FET nearest the load burning up.

A tool that helps a lot is an FLIR image camera....you can SEE the heat bloom anywhere in your circuit, usually in time to intervene, especially with soft start circuitry and PTC protections.

Time permitting, I too plan on building an induction furnace to try making a graphite lined, stainless steel crucible hot enough to melt gold and alloying elements. Nothing bigger than what you have. I plan to stay away from resonant effects . Water cooled copper tubing. Power from a stack of Meanwell isolated SMPS supplies, they respond well to current limiting as a means to control the temperatures.

 

[gary350]

What is SMPS?

https://www.smps.org/

 

[Mosaic]

https://www.smps.us/power-supply.html

 

[tcmtech]

Well if you want something to work right you have to start with using the right parts which basically means not using low quality undersized components and running them without any real solid design knowledge of what you are aiming for.

So far on your induction heater circuits I see what everyone else sees. A highly questionable poor low quality low efficiency circuit being built from crap parts by someone with very little practical knowledge of parts and process he is working with and trying to design.

Personally if it was me first off I would be putting my money towards getting some good used or surplus stock high power IGBT devices for the switching circuits and using actual dedicated gate driver IC's to control them and powering them with a well built DC power source that is not built from bits N' pieces of low end kitchen appliances and obvious junk chinese knock off parts.

My recommendation would be to find some 300 - 600 amp 600 - 1200 volt IGBT, Mosfet or power transistor modules designed for switching power supply or electric motor drive applications and set them up with dedicated gate drive IC's and building a good high capacity DC power supply to power them built out of something like actual purpose built power transformers with good high capacity rectifiers with a decent sized battery bank as a primary filter and load holding power source.

Granted the initial investment might set you back $100 or so but you wont kill them every time things go wrong in the design.

BTW don't say you cant afford to do it right because doing it wrong is all you can afford. If you can afford to do it wrong multiple times over ending with nothing you can afford to do it right.

 

[jjw]

Did you change something in the circuit before fireworks?

 

[Tony Stewart]

Too bad you don't have a scope. I have 2 I hardly use but couldn't do solid analog work without it.
Having a variable supply with a variable current limiter is a good project next.


When you search for anything in google like SMPS... ignore the first few hits which are ONLY paid ads and often useless.

I know someone who could turn 1/2" iron rod white hot within seconds after a few minutes warmup of a propane with high pressure air running at 90 degree in a nozzle to create a high rpm vortex of the mixed fuel. The result sounded like a rocket but what a torch from a little propane. Air Pressure had to be ramped up slowly with low propane flow rates until a super vortex flame was produced.

Next time you try 1KW induction, try 13.56 or 27.14 MHz. This is most popular for 1kW Induction or plasma heaters.

 

[Mosaic]

Have a look at this for guidance:
https://www.mindchallenger.com/inductionheater/

I have a digital scope I could let go for cheap....got the logic tester interface as well.
Its a Bitscope 100.
https://www.bitscope.com/product/BS100/?m=p

Since that unit I got a couple more Rigols and Tektronix for a total of 6 now.....o well, I like scopes and I could never afford one when I was young.

 

[gary350]

Well if you want something to work right you have to start with using the right parts which basically means not using low quality undersized components and running them without any real solid design knowledge of what you are aiming for.

So far on your induction heater circuits I see what everyone else sees. A highly questionable poor low quality low efficiency circuit being built from crap parts by someone with very little practical knowledge of parts and process he is working with and trying to design.

Personally if it was me first off I would be putting my money towards getting some good used or surplus stock high power IGBT devices for the switching circuits and using actual dedicated gate driver IC's to control them and powering them with a well built DC power source that is not built from bits N' pieces of low end kitchen appliances and obvious junk chinese knock off parts.

My recommendation would be to find some 300 - 600 amp 600 - 1200 volt IGBT, Mosfet or power transistor modules designed for switching power supply or electric motor drive applications and set them up with dedicated gate drive IC's and building a good high capacity DC power supply to power them built out of something like actual purpose built power transformers with good high capacity rectifiers with a decent sized battery bank as a primary filter and load holding power source.

Granted the initial investment might set you back $100 or so but you wont kill them every time things go wrong in the design.

BTW don't say you cant afford to do it right because doing it wrong is all you can afford. If you can afford to do it wrong multiple times over ending with nothing you can afford to do it right.

I didn't know I was playing with a micky mouse toy but you did. You knew about IGBTs the whole time and kept it a secret. I had FUN that was the original purpose of this project.

 

[Mosaic]

Gary:
Have a look at the tutorials in the left column here:
http://www.w5dor.com/W5DOR-Links.html

This is interesting, but is not really applicable to your induction heater. It matters for longer distance transmission or UHF signals.