4KW Induction Heater / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

4KW Induction Heater

1 2 3 4 next

Move Thread

LAN_403

RogerInOhio

Sat Nov 26 2011, 04:38PM

Registered Member #1034 Joined: Sat Sept 29 2007, 12:50PM
Location: Chillicothe, Ohio
Posts: 154

This Royer induction heater achieves a moderately high power level by using groups of 4 IRFP260 MOSFETs in parallel. The circuit is similar to Marko's Royer oscillator shown below except that each side uses four MOSFETs paralleled together with each MOSFET having it's own gate biasing circuit. The B+ voltage is 70 volts. I found that if I go much higher than that the efficiency starts going down and the MOSFETs become more prone to failure. Since I am using a higher B+ voltage I have changed the resistors in the gate biasing circuits from 470 ohms to 800 ohms.

The resonant frequency runs in the range of 40 to 60 KHZ depending on what work coil I am using. The tank circuit uses 24 .22uf capacitors in parallel for a total of 5.28 uf. They are Cornell Dubilier polypropylene capacitors rated for 1600 volts. I can leave the induction heater on for long periods of time and the capacitors will just get a little bit warm.

The power level depends on the work coil and the size of the object being heated and I am measuring input power here. With nothing at all in the work coil the machine consumes 700 Watts and if you load it much beyond 4KW it will stop oscillating and bad things can happen to the MOSFETs. Only ferrous metals will heat up red hot and other metals just get hot to the touch.

1322325485 1034 FT0 1313589228 89 Ft1630 Royer Ih

Marko

Sun Nov 27 2011, 02:26AM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Haha, so you boxed it up... I honestly never intended this circuit to be used as a worktool as much as a curiosity for newbies (who always kept doing weird things with flyback cores and never getting it rigtht).

The obvious problems are the power supply as well as the instabilities in the circuit... diode voltage drop along with switching device forward voltage makes switching off very troubleosme at high voltages (I tried it with 1200V igbt's, and after some 150V it blew up)

Perhaps an IH like this could though eb used as a handy add-on for an arc welder.

Roger, ave you tried using a smaller work coil? That way you might be able to melt something instead of just heating huge pieces of metal red.

Also, you seem to have wiring gratuitously hanging everywhere... I'm usually not a parasitic inductance grandma, but your connections between mosfets and tank circuit could be having inductance significant in comparison to workcoil... make them coaxial or at least twist them together, if you can't shorten them... same applies for the gate drive section as well!

marko

RogerInOhio

Sun Nov 27 2011, 02:15PM

Registered Member #1034 Joined: Sat Sept 29 2007, 12:50PM
Location: Chillicothe, Ohio
Posts: 154

Marko, I have been thinking about making a smaller work coil. It would be cool to at least be able to get things white hot instead of red hot. There is a reason that I'm building this beyond the curiosity stage. If all goes well it will be used by some machinist for heat treating and silver soldering little pieces of carbide to steel bars.

As far as the wiring is concerned, I did try having more direct connections to the sources on the MOSFETs once but it wouldn't run that way and when it doesn't run everything goes up in smoke. I had to go back to connecting the source leads to the ground on the biasing boards. This circuit is very finicky about some things but once you get it right it is vary reliable.

Another thing I learned the hard way is not to use MOSFETs other than the IRFP260 or IRFP250 types even if they have higher ratings. At least not unless you really know what you are doing. They can be made to work but they won't work as well.

Marko

Sun Nov 27 2011, 02:59PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hi roger

well, If I for any reason just had to do a high power this way, there's no way I'd do it without overcurrent protection. A simple shunt, LM311 and a NE555 could save countless mosfet lives not only in prototyping stage but as well as when the end user decides to place a too large workpiece into it so the circuit latches up and explodes.

Marko

Jrz126

Mon Nov 28 2011, 01:11PM

Registered Member #242 Joined: Thu Feb 23 2006, 11:37PM
Location: Erie PA
Posts: 210

Looks great Roger.
Have any waveform pics? Where'd you get that boat anchor of transformer?

Sulaiman

Mon Nov 28 2011, 01:28PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

RogerInOhio
a zvs running in zvs mode produces PI x Vsupply peak across each switch
so it's not surprising that your transistors heat up with greater than 70 Vdc
as 3.1416 x 70 = 220 V peak , IRFP260 Vds = 200 Vpk !!!!
i.e. you need higher voltage rated transistors.

RogerInOhio

Mon Nov 28 2011, 03:46PM

Registered Member #1034 Joined: Sat Sept 29 2007, 12:50PM
Location: Chillicothe, Ohio
Posts: 154

Since I went to using 8 MOSFETs I haven't blown any up yet from over loading but maybe I should have better over current protection than the fuses I am using.

The transformer is from an old UPS system that was being scraped. I was lucky enough to get my hands on two of them. I will post a picture of the waveform below.

As for the MOSFETs, I have tried some that have a higher voltage ratting but nothing seems to work as well as the IRFP260. I guess I will have to wait for Marko to pioneer a circuit that uses IGTBs.

I just made a hot new video of the induction heater using a smaller work coil.

Marko

Mon Nov 28 2011, 08:11PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hi guys

Well, I don't think I'll be pioneering this circuit with IGBT's any time soon unless someone pays me especially for it and organizes free time to do so!

Just to illustrate how bad the problem is - to run offline with 325VDC at least 1200V devices would be required. 1200V IGBT's have forward voltage drop of as much as 3-4V. Then consider the feedback diodes, which at this voltage rating might have over 2V drop too which in total is well over gate threshold voltage of the opposing igbt - meaning that it'll never actually fully turn off! I was surprised that the contraption worked at all for some time before blowing up at some 150V DC in or so.

Also, when I tried using UF4007's for feedback they blew up from recovery losses at just 100V in or so, and had to use two 600V 8A fast rectifiers in series which made the voltage drop even worse.

I think this is the same reason why lower RdsON mosfets work better in this circuit; they can turn off the gate of opposing mosfet harder and hence reduce the possibility of ringing from turning it back on (and degenerating into a parasitic oscillation which sends everything to oblivion)

The very topology of a current fed inverter is on the other hand very attractive for induction heater designs, especially at high frequencies where ZVS operation becomes useful as well as unitiy power factor from the inverter obtained with parallel resonant circuit (which isn't the case with LCLR)

The industrial standard for such an inverter would be a full bridge with PLL based controller, although I see no reason why wouldn't it work with push pull topology as long as you respect the device voltage ratings.

I see no reason why a well made direct feedback scheme with phase lead compensation wouldn't work as well, although it would be nowhere closely as simple as a good old royer oscillator and nobody would likely bother building it (instead of just copying kim ladha's or whoever's IH).

Marko

Inducktion

Mon Nov 28 2011, 11:19PM

Registered Member #3637 Joined: Fri Jan 21 2011, 11:07PM
Location: Buffalo, NY
Posts: 1068

Why not try to use those fancy SiC Schottky diodes? Their voltage drop should be really really small, correct?

Or maybe, instead of using diode feedback why not use some active gate driving techniques?

Marko

Tue Nov 29 2011, 05:25AM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Inducktion wrote ...

Why not try to use those fancy SiC Schottky diodes? Their voltage drop should be really really small, correct?

Or maybe, instead of using diode feedback why not use some active gate driving techniques?

Well, it's the device voltage drop, not the diode drop that creates the most trouble.

I have tried feedback from an isolated winding but failed... at low power levels, I also made a circuit that combines diode feedback and feedback windings to reduce the pullup resistor losses, and it worked wonderfully although the diode problem remained.

I'm currently researching direct feedback approach with capacitive voltage divider (basically, making the gate a portion of tank capacitance) but it came with it's own set of problems too - bias on the gate seems to vary and equalizer resistors and zeners weren't as good at fixing it, not sure why. (btw, the idea works wonderfully when gate is connected directly to opposing drain and circuit is fed with small <6V voltages)

I was even wondering if it might be easier to replace igbt's with high power BJT's (like HOT's) since they are biased by current and could be much more easily set into linear region without tendency to blow up instantly (hopefully).

Ofcourse, a circuit like this should definitely be overcurrent protected well with transzorbs to absorb overvoltage when all devices are suddenly turned off... I've never seen people do this though and they just keep crying OMG this circuit sux I blew all 10 mosfets I had and mom give me no money :(

Cheers,

Marko

1 2 3 4 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob