2kW Induction Heater / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

2kW Induction Heater

Move Thread

LAN_403

Wilson

Mon Apr 24 2006, 02:16AM

Registered Member #78 Joined: Thu Feb 09 2006, 11:27AM
Location: Sydney, Australia
Posts: 133

I have been working on this project for almost 1/4 of a year now, and i finally have a barebones setup which i'm not too embarassed to present.

Here's the Setup

At the top right you see my isolation transformer, made from the primaries of 2MOTs stuck onto the same core, as well as a 400V 35A bridge mounted on a heatsink and a filter cap.

In the centre is a full bridge of IRFP460s, to the left of which is my standalone TL494 based oscillator, adjustable from 19khz-220khz, in push pull mode, driving 2 UC37321 gate drivers.

I'm currently employing the LCLR topology outlined in Richie Burnett's site. The work coil is put into parallel resonance with a cap bank, and is connected to the bridge through a series inductor.

The current work coil is made from 6mm Cu pipe, 5 turns and roughly 2 inch dia.
If you look carefully, i will notice that its water cooled.

The cap bank used for parallel resonance is 12 x 1500V 0.047uF caps from RS
These are connected to the work coil through crimped 4AWG wire.

The series inductor is admittely rather cheap...
Homemade litz wire (7 x 0.5mm strands) wound onto a core from a monitor's SMPS.
(It was vibrating alot, so i clamped it, and accidentally broke it by clamping too hard

)

Enough with the details, here's some pics of it operating:

Hot Iron Pipe

This pic was taken a while ago, while i was still using my old 3" work coil. Fres at around 80khz, power at roughly 1kW (couldn't go higher was i was using IRFP250s at the time)
Temperature roughly 800-900 degrees. The pipe is made of softish iron.
This work coil was replaced, as the coupling was far too low to heat all but the biggest objects, which are unweildy when they are at 700+ degrees above ambient.

Here's a pic of my latest 'successs'
Glowing 20c

Being heated is an Australian 20c coin. A cupro-nickel alloy with 75% Cu. Melting Point of around 1100-1200oC.
The variac was turned slowly up to 240VAC in, the setup drawing 2kW, upon which the coin had started to melt. Although the breaker tripped b4 i got a pool of metal :(. This was done at 190khz.

Here's the resulting coin

And the underside

Current aim: To melt pure aluminium, either by raising the freq. or increasnig the power. At 1kW, Al gets hot, but nowhere near glowing temperature.

Improvements needed:
New series inductor- The current one gets rather hot, especially at 2kW
New Cap to coil connection- The 4AWG wire gets hot enough to melt the insulation on it....dodgy skin effect
Better cap cooling- the current bank gets hot to to touch at 2kW.
Self resonant oscillator- it gets annoying having to twiddle pots every 2mins.
I've tried using PLL, but i couldn't get a clean output waveform.
Find a place to buy mini crucibles.

Comments as welcome, along with any suggestions on improving efficiency, etc.

Enjoy

Wilson

Dr. Shark

Mon Apr 24 2006, 01:27PM

Registered Member #75 Joined: Thu Feb 09 2006, 09:30AM
Location: Montana, USA
Posts: 711

Very nice, makes me want to work on my induction heater again. I like it how you are working at 2kW, which is probably just enough to do some serious heating. A simple propane torch is in the same ballpark, so less just wont cut it.

It looks like heating losses by skin effect is your biggest problem at the moment, but have you figured out a way to minimize these? How did your litz-wire inductor perform, litz-wire wise?

For the feedback I imagine you could just use a 1:1000 current transformer on the resonant circuit like people do with TCs. No need for a PLL here, or am I missing something?

Wilson

Mon Apr 24 2006, 02:09PM

Registered Member #78 Joined: Thu Feb 09 2006, 11:27AM
Location: Sydney, Australia
Posts: 133

Skin effect is most definitely a problem. However, i am rather surprised at how well my litz wire inductor worked out, considering that i used some old recycled wire from a MOT. At 1kW there are no problems, its rather warm, but it was passing 5A. At the end of my 2kW run, the enamel had started smoking, and the core was also quite hot (presumeably from heat conduction).
I'm planning on buying some decent wire soon, and winding some massive litz wires. Skin depth is roughly 150um at 180khz, so 0.3mm wire will do nicely.

The most lossy component at the moment is the workcoil itself. The first time i ran it, without water cooling, it got hot enough to start discolouring within 3mins, and was beginning to sag from the looks of it.

To tell the truth, i cannot think of any solution which would have minimal waste in the work coil, without having a massively thick multistranded conductor. The circulating current is calculated to be in the order of 170A peak. 4AWG wire at almost 1cm dia is rated for 110A DC...
So far, the only feasible solution i can think of is to wind a 'sheath' of 0.3 enamalled wire around the work coil, using both the workcoil and the wire as the conductor. Any excess heat would then be transported away by the water cooled tube.

A novelty solution would be to use a super cooled conductor :P

In regards to the feedback, i chose to use PLL as it seemed to be the most popular and well documented. I have yet to build a SSTC, so i'm not too well aquainted with the technicalities behind it.

EDIT: Just thought of something: How about a large, but flat Litz wire braid wound into a multilayered spiral with the workpiece in the centre?

Hazmatt_(The Underdog)

Mon Apr 24 2006, 06:39PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

When I have the spare time, I am planning on using some of my silver-teflon 10AWG wire for the inductor and see how that goes.

hvguy

Wed Apr 26 2006, 08:42AM

Registered Member #289 Joined: Mon Mar 06 2006, 10:45AM
Location: Conroe, TX
Posts: 154

Current aim: To melt pure aluminium, either by raising the freq. or increasnig the power. At 1kW, Al gets hot, but nowhere near glowing temperature.

This can take a lot of power. At 2kw I would not expect much out of Al; it is notoriously hard to couple power into. If youâ€™re looking to heat a large quantity of aluminum (2â€ x 2â€ block to 1220F or liquid) you are probably looking at >12kw. Also, aluminum does not change color when liquid, it just becomes shiny.

I would stick with ~100khz and up the power. Good luck...

Wilson

Wed Apr 26 2006, 09:37AM

Registered Member #78 Joined: Thu Feb 09 2006, 11:27AM
Location: Sydney, Australia
Posts: 133

Dang...the problem is, i live in an apartment, so finding a way to draw more than 2.4kW without burning down the place is a challenge. I could either draw power directly from the swtichboard, or from 2 outlets connected to 2 different breakers.
The IRFP460s are already running close to their limits at 2kW....it turns out that the reason why the breaker tripped on my 2kW run was that a mosfet(s) died, taking out the rest on the bridge. I'm still trying to figure out what caused the failure, as there was no heating (or bangs), and the gate voltages were clamped...I'm suspecting HV spikes between DS atm...

Steve Conner

Wed Apr 26 2006, 01:00PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

woot, nice job!

Water cooled copper tubing is what all commercial induction heaters use for the work coil, so I guess it must be near optimum in terms of bang for the buck. I would be tempted to just dunk the cap bank and matching inductor in a plastic tub of water and hope it doesn't conduct too much current. After all, you've already broken the water/electricity taboo by water-cooling the work coil. Some ice cubes might help too :P

I had trouble driving an induction heater with my Mk.1 PLL circuit. It doesn't track the resonance exactly. It gets close enough for a Tesla coil, but I saw some very dirty waveforms with bad ringing when driving a heater. The Mk.2 should work much better since it has an integrator for the phase error, but then it's much more complicated.

As for only being able to get 2.4kW... are all your breakers 10 amp or something? You can probably draw a fair bit more for short periods before the breaker pops. Here (in the UK) our wall outlets are on ring mains with a 32 amp breaker. You're only supposed to draw 13A from an outlet, but they don't seem to mind 32A for short periods, so you can get about 6kVA

Wilson

Wed Apr 26 2006, 01:59PM

Registered Member #78 Joined: Thu Feb 09 2006, 11:27AM
Location: Sydney, Australia
Posts: 133

whee^^ 6kVA. I'll probably need to parallel 3 IRFP460s for some safety margin and design a monster of a gate driver. The breakers are rated for 30A. I was just worried that the wires in the walls would start smoking, as its a rather old apartment, and i'm dubious as to whether the wiring is of the same standards as it is now.

In regards to self resonance, i was thinking on turning the output from the current transformer into a square wave somehow, buffering it, and then sending it to the gate drivers, and implementing some soft start circuitry. Seems to be the easiest way providing that i can detect the zero crossings accurately.

Dunking everything which is hot into water is a novel idea

Coffee anyone? And if i eventually put this all in a case, i can say to friends "I'll just fill 'er up" lol

I know all commerical setups use water cooled copper tubing, just shows that they couldn't find a better more economical solution.

Power lost in the coil means power not delivered to the workpiece. Not a major concern for commerical setups operating at 100's of kW, but i think it is for our ametuer setups:
Heat Content = mass x specific heat capacity x change in temp. (H=mCdelta T)
My mini water pump operates at a feeble 1L/40secs. Assuming the change in temp on the 2kW run to be around 3-5oC. Specific heat capacity of H2O is 4.2J/g
Heat lost in coil = 1000 x 4.2 x 3 J = 12600J
This energy was lost over 40secs, so power lost = 12600/40 = 315W. With a 5oC change in temp, the loss goes up to 525W

- A significant portion of the input power....<sigh> Maybe a plasma arc furnace would be more efficient in melting Al, even if its 'elegant', or an iron crucible.

This project is currently on hold. I have a large amount of schoolwork which i need to catch up on before the start of term, and i need to build another full bridge to replace my dodgy old one. I *just* have enough 460s for a second.
I'll post the results for 2kW on some Al next time i get a chance.

Thanks for all your input eveyone,
Wilson

Hazmatt_(The Underdog)

Wed Apr 26 2006, 09:18PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

You can draw a significant amount of current through a breaker before it trips. I try not to advocate this because its conditional and dangerous. Inside a structure I would be careful about doing this. Now in the garage I have 220V service at 30A and I think its wired through with 8 AWG. Now I was only using 1 phase of it for 110V, but I pulled horrendous currents from the breaker before it popped.

I was doing a 'streamers rule' sorta thing, where I had my GE WWII surplus dry type magnetron transformer with a Jachob's ladder on it. It was about 30' inside and we were all far away from it, operating from the breaker itself. The transformer, A 3180V 50lb beast, drew 220A at 110V for 2 seconds with a 14" flame flaring from the ends.

So whats the point? Well, you could probably draw 40A for a long time from a 30A breaker before its element trips the breaker. Or if you want to live dangerously, 200A for about 2 seconds.

Today there are plans on board right now for a power supply. 220-110 5KVA stepdown, 220V 7.3KVA Staco, Enormous 660V 100A rated switch (might be German, I dunno), Need meters, welded frame, Hampden 100A binding posts, but no more foolishness for me. I prefer to live safer.

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