4KW Induction Heater / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

4KW Induction Heater

previous 1 2 3 4 next

Move Thread

LAN_403

Tonskulus

Sun Dec 04 2011, 02:05PM

Registered Member #1223 Joined: Thu Jan 10 2008, 04:32PM
Location:
Posts: 133

I wonder how would it be possible to heat up and melt non-ferrous metals using this circuit?
It looks like when in idle mode (empty work coil), there is not much tank current circulating there.
Its not frequency issue, my halfbridge igbt induction heater will melt non-ferrous coins or similar in seconds! Input power was around 2kW and frequency 50-70kHz.
aluminium coin:
http://www.youtube.com/user/tonskulus?feature=mhee#p/u/34/STVJOB7o9dI

I know, non ferrous metals require much more work coil current compared to ferrous metals like iron.
Difference is that iron has both, hysteresis losses and circulating current (resistive) losses while non-ferrous has only circulating current losses. Like aluminium, it is quite good conductor so lots of current is required.

Marko

Tue Dec 06 2011, 11:32AM

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

Hey guys

Tonskulus: Unlike series resonant or LCLR induction heaters, a current fed royer oscillator tends to hold constant voltage on the tank circuit, equal to pi*supply voltage. Since impedance of the tank is dominated by it's reactive components, the circulating current in it will approximately be equal to this voltage divided by the tank's characteristic impedance. And this is by no means a small figure, I seem to get close to 150A in my little royer induction heater just when it sits without workpiece.

When non-ferrous workpiece is inserted it tends to both decrease inductance and increase the loss in the tank, so current will go up slightly as well as get more into phase with voltage. I don't think this increase is very significant though compared to no-load current in the coil, even if the workpiece completely shadows the coil I don't believe it ever reaches 1.5 initial current or so.

It's true that non-ferrous metals (especially highly conductive ones) will represent a less of a load when put into work coil, but how fast you can melt them really depends on how much power you put into system! I was limited by quite smallish transformers I was using for my power supply.

The good thing about current fed oscillator is that it can't run away when not loaded with any workpiece - can be pretty much ran open loop unless you need special power regulation.

By the way, after some research I think it would be possible to implement an IH like this with big 1500V HOT's. Due to slowness of devices it would require high tank capacitance, on order of 100uF.

If someone is willing to donate me one of those 100uF 500A eurofarad caps which I can't get from ebay, I promise this will be the next thing I'll try

Marko

Tonskulus

Tue Dec 06 2011, 03:15PM

Registered Member #1223 Joined: Thu Jan 10 2008, 04:32PM
Location:
Posts: 133

Yes, that makes sense.
I never measured the actual current on my tank circuit, DC bus voltage was around 300V so ofcourse there will be much more tank current compared to a royer operating around 70Volts.

One solution would be heavy matching transformer, there will be power losses but some commerical induction heaters use this method. Yes, tank circuit is at primary side of transformer. Work coil is connected directly to the secondary.

Some vacuum tube oscillator type heaters used air cored matching transformers (I did that too). But it is totally different...
but the point is that matching transformers can be used anyway. Requires heavy windings/water cooling usually.

I need to try out that royer circuit if I can find a method for heating non-ferrous metals with minimal power losses :)
Maybe a challenge for us!

RogerInOhio

Tue Dec 06 2011, 03:44PM

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

Marko, after doing some searching I think HOT must stand for horizontal output transistor and the ones I have found have kind of a low current rating. It seems like a tank circuit with a 100uF capacitor would have a low impedance even if the frequency was low. Would this be a Royer Oscillator or some other type of IH circuit?

On the subject of eddy currents, I have noticed that the work coil itself has significant eddy current losses. If I run my IH without water going though the tubing the work coil gets hot but the tubing going to the work coil stays cool. I guess that a work coil that used something like litz wire would be more efficient.

Marko

Tue Dec 06 2011, 08:08PM

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

Hi guys

Yes, HOT's stand for TV horizontal deflection transistors, and I thought their ratings would be too low until I saw some beefy ones for high-definition TV's ;)

And yes again, they would be used in a royer oscillator, in a classic configuration with feedback winding!

This one is rated 25A continuous, 50A peak and 1600Vce which should make it robust enough for the application.

2SC3998 is another nice one... if anyone finds a better transistor let me know!

Ofcourse there's a large chance this idea might not work at all, and I'd be just as happy if someone else tested it instead of me!

To keep the frequency low enough for operation of HOT's, I realized that quite large tank capacitance would be needed, on order of tens of uF... and those eurofarad caps just came to my mind as they seem useless for most other things (apart for decoupling on a monstrous boost or buck converter?)

I wouldn't likely be able to drive them to their full voltage ratings - but with 500A rms they would surely be a lot of fun anyway :)

Marko

Inducktion

Wed Dec 07 2011, 11:44PM

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

How come no other type of oscillator has been used in induction heating?

Hartley, Colpitts and other classic oscillators come to mind. I know they're used in Vacuum tube IH'ing, but what about solid state induction heating?

Marko

Thu Dec 08 2011, 12:44AM

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

Well, to my understanding, IGBT's and MOSFET's were never really popular in industrial scale for self-oscillating circuits because they are prohibitively hard to bias properly into linear region and stay such with variation of temperature!

They are not designed for that, unlike tubes and bjt's, and should really be controlled by a PLL circuit or something.

On other side bjt's might be a much better for this and much more similar to tubes. I bet you should be able to find at least a couple of patents for a self oscillating BJT based IH, from the era before other devices evolved enough to replace them!

You could look into some bjt based hartley or whatever oscillators and try to scale them up if you ahve courage.. I chose to stick to mazzilli just because it's most familiar to me and hence the least chance of blowing up :)

Marko

Marko

Fri Dec 09 2011, 04:14AM

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

Hi guys

I've done some tests regarding this idea (finally).

I didn't have much of useful HOT's - I found two of 5 and 10A rating and tried the circuit with them... and at just 50 or so volts in the 5A one blew instantly without any clue why... I didn't use a DC blocking cap and I suspect it might be due to saturation of the transformer with DC due to unequal transistors :(

I did some mini scale experiments with BUT11A's instead. Those are pretty tiny transistors and served for proof of concept only, bolted to a symbolic heatsink. I used a 100kHz tank circuit of my mosfet royer IH and since those are very slow devices there wasn't much sense in trying to push more than few tens of watts.

One prediction was right, the bipolar transistors do bias wonderfully in this circuit. The circuit can't latch up disastrously like IGBT or MOSFET one; instead it just turns off in case the tank is overloaded, floating the devices to low-dissipation linear region.

I made a quick video of the circuit, with upper scope trace being the collector and the lower base voltage.

The interesting thing is that circuit starts oscillating in linear region, with very low voltage applied! I suspect this is due to current dependent Hfe of the transistor imposing negative feedback sort of like the light bulb in Wien bridge oscillator.

With increasing voltage the circuit jumps to relatively normal class C royer operation. At first I used only 5 secondary turns on the transformer and the waveform looked quite nice, but when I added more some really nasty ringing started to appear. In any case there are some interesting dimples in both base drive and collector voltages which get far more interesting when I load the heater by a piece of steel! It could get small pieces of metal only to like 100 degrees C before heatsink would become uncomfortably hot, and it wasn't the point now anyway to seriously heat or melt anything.

The specs of the circuit -

30 turn primary, 10(ish?) turn secondary, 3 turn feedback winding (going between bases with 50 ohm series resistor), two 18k pullup resistors on both bases, 100uH inductors and my old royer IH tank...

In overall, I'm not suure what chance would this have to work with hot's, but would be fun to try I guess. I would need to make some proper waterblocks for them as well as a new bigger workcoil assembly, could do it over holidays I think

Marko

Sun Jan 08 2012, 06:03PM

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

Hi roger, everyone,

The 100uF cap and the HOT's finally arrived - took a lot of waiting and I only had about a day to experiment before having to return to Zagreb.

Long story short, I managed to get the circuit oscillating, but hardly any power output before it blew up (most likely thanks to my error of using way too small biasing resistors).

Low resonant frequency has required a troublesome construction for the step-down transformer with very high leakage inductance, which is pretty unwanted between the switches and the tank circuit.

I couldn't make the circuit oscillate properly at first, but this was solved by adding a small capacitance to the primary side of the transformer (which, hopefully, will process only negligible reactive power).

I'll be back working on this in about a month - I plan on switching to a self biasing feedback scheme (very similar to grid leak circuits in VTTC's) and hopefully finding better inductors than current ones I'm using (PC-PSU PFC inductors)

Marko

RogerInOhio

Sun Jan 08 2012, 10:31PM

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

Hi Marko, I'm glad to hear you where able to get started on it. What is the resonant frequency of this oscillator?

I have been been working on a Royer oscillator for running a modified automotive ignition coil and it runs somewhere between 1 and 2 kHz. I should be posting some details on it in about a week.

previous 1 2 3 4 next

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