LCLR Induction Heater / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

LCLR Induction Heater

1 2 next

Move Thread

LAN_403

Zum Beispiel

Fri Jan 25 2013, 02:01AM

Registered Member #514 Joined: Sun Feb 11 2007, 12:27AM
Location: Somewhere in Pirkanmaa, Finland
Posts: 295

Hey all, long time no see. Been busy with stuff, so I haven't posted much (read: at all). I've got a bunch of projects done, though, and some of them may be trickling down here soon(ish). Here's the first one, my attempt at an induction heater. Hey, everyone needs to build one sometime, right?

Basically I'm running a fixed frequency fullbridge of IRFP460's, all driven with a UCC37322/1 and GDT setup such as one would find in a garden variety SSTC. Power comes from a salvaged Finlandia Interface rack mount (and mil-spec) DC power supply, capable of 10A on the 48V rail. We'll get to running from mains once everything else is sorted out.

The induction head itself consists of 10T of 1/4" refrigeration grade copper tubing, calculated at 2.9ÂµH. The resonant capacitor is 20 pieces of 0.24ÂµF Wima MKP10 in parallel, measured capacitance 4.72ÂµF at room temperature. F0 comes to around 43kHz unloaded. The series chokes are 2T on mystery ferrite E-cores. Tank current is monitored with two series 33:1 CTs (1089:1).

Have a video. First test of the new head with 24V, circa 240W DC input.

So far things seem to be working quite nicely. Nothing has caught fire yet and the inverter doesn't get warm with passive cooling.

The induction head is pretty much complete. Next up will be building a better inverter and proper self-tuning driver circuit. Water cooling the work coil will come after that, I just need to find a nice small radiator first.

Comments? Suggestions?

(I wish I had a proper lab. Living in an apartment building sucks.)

Sigurthr

Fri Jan 25 2013, 10:53AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

That's mighty fine looking work for an apartment dweller! I have the same issue and it really disturbed progress on my TC building.

Josh Campbell

Fri Jan 25 2013, 09:46PM

Registered Member #5258 Joined: Sun Jun 10 2012, 10:15PM
Location: Missouri - USA
Posts: 119

Great setup, apartment or no.

Question: Is that a current transformer just behind the work coil? If yes, is it wound around a ferrite core or is it a rogowski coil?
If no, is it a coupling transformer, possibly when you were trying out series LC driving vs the currently set up LCLR?

Zum Beispiel

Sun Jan 27 2013, 06:38PM

Registered Member #514 Joined: Sun Feb 11 2007, 12:27AM
Location: Somewhere in Pirkanmaa, Finland
Posts: 295

It's a current transformer, 33:1 wound on a Ferroxcube 3E25 core, to be exact (the GDT uses the same core). It's coupled with another 33:1 CT for a 1089:1 ratio, like people like to use with DRSSTCs.

I'm now working on a new inverter, the PCB is done and I'm just waiting for some parts (MUR1560's) now. It should be pretty nice.

There should be more soon.

Josh Campbell

Sun Jan 27 2013, 07:59PM

Registered Member #5258 Joined: Sun Jun 10 2012, 10:15PM
Location: Missouri - USA
Posts: 119

Watch out for that, it will overheat. It must be getting at least warm at the power levels your running now.

Another method that works well is to put a smaller capacitor in parallel with your tank capacitor and measure the current through it. If your capacitor is 1000 times smaller the current will also be 1000 times smaller.

Here is a real life example with pics.

I have a 10uF tank capacitor in this setup. In parallel with that I have a .01uF (Cornell Dubilier PVC1611) capacitor, film/foil & non-inductively wound type. My measurement (or parasitic as I'll call it) cap is 1000 times smaller than the main tank cap so my current measured through the .01uF cap will be 1000 times smaller than the current through the 10uF capacitor.
I have a 1:100 current transformer on one leg of the .01uF cap with a 100R burden resistor. So one volt on the scope will equal one amp flowing through the .01 cap which will equal 1000 amps flowing through the 10uF cap.

There is another method that produces the same value as the measurements taken from your parasitic tank cap. This formula:
Tank I = V*2*cake*F(in hertz)*C(in farads)

Ok here is the real life example:
I fired up an IH and let the work piece heat up past curie then lowered the power level to give me a stable frequency, tank voltage and tank current. I measured 130V on the tank capacitor, I have a .5V sine wave on the parasitic tank capacitor and it is oscillating at 62.5KHz.
So according to this measurement we have 500 amps (.5V * 1000 = 500A) of oscillating power between the tank capacitor and the work coil/work piece (which is a 3/4" RR Stake in an 8 turn 1.5" diameter 3" long work coil).

Lets check our original tank current measurement with this equation:
Tank I = 130 * 2 * 3.14159 * 62,500 * .000,01 = 510.51 Amps

Pretty damn close, probably my crap scope here and the approximate readings taken at the scale it's set to. It's not the burden resistor it's a 1% and I picked one that measured exactly 99.8R.

For fun here is the RR Spike during this test (first pic) then the spike and scope tracing with tank current at 1000A (5uS / .5V scale):

If you happen to try the parasitic cap method make sure to put the cap on alligator clips so you can move it around. Having a multi cap array like that the reactive current will likely be higher near the work coil.

Another good formula to have is the one for calculating the KVAr the tank capacitor(s) are processing.
KVAr = VÂ²*2*pie*F(in kHz)*C(in Î¼F)*10-6

So in our above example with 500Amps of measured tank current:
KVAr = 130Â² * 2 * 3.14159 * 62.5 * 10 * 10-6 = 66.36

Or since we know the tank current we can just calculate the VAr like so:
VAr = Tv * Ti
VAr = 130 x 500 = 65,000 / 1000 = 65KVAr

Pretty close again. Definitely some values off a bit, I read the tank voltage off the scope at a 50V scale and the frequency at from a 5uS scale so that's probably not helping either, but you get the point.

It took me a little bit to get to this point, having measurements and equations that always work for tank current, KVAr and such, so hope this helps! Just make sure to keep the cake and pie separate.

Zum Beispiel

Sun Jan 27 2013, 11:19PM

Registered Member #514 Joined: Sun Feb 11 2007, 12:27AM
Location: Somewhere in Pirkanmaa, Finland
Posts: 295

Cool. I'll be keeping that in mind and I'll be watching out for overheating when I crank up the power, though I haven't noticed anything yet.

But why should the CT overheat in this application in the first place? A similiar setup is widely used in DRSSTCs for primary current monitoring, where they can see over a thousand Amps. Now, I know that it's pulsed, but the average power can still be several kilowatts. How is this different?

Josh Campbell

Sun Jan 27 2013, 11:41PM

Registered Member #5258 Joined: Sun Jun 10 2012, 10:15PM
Location: Missouri - USA
Posts: 119

My guess is that it's the continuous, high reactive power. The low duty cycle on the primary of a DRSSTC is reducing the average power seen by the CT toroid in comparison to an IH. On my first induction heater I melted my AMPPROBE clamp on meter, it pegged out at 300 amp (maximum measurement) and was only on the tank for about 10 sec. I later found that there was around 550A on the bus between the tank caps and work coil. Of course the clamp on meters have iron laminate core's, that was a dumb idea in hind sight. I later tried ferrite UI core's with the same result. Even very large ferrite cores, (type 3C90, 3C85, and one unknown) heat up very quickly in the resonant tank of an IH.

Avoiding any inductance in the path between the tank cap and the work coil is a main design criteria for most high power tank assemblies.

Those are my thoughts, anyone else?

Zum Beispiel

Sun Jan 27 2013, 11:57PM

Registered Member #514 Joined: Sun Feb 11 2007, 12:27AM
Location: Somewhere in Pirkanmaa, Finland
Posts: 295

Interesting. I'll do a stress test when I get the time. Just let it run at ~300W for a while and then check what is getting hot.

Any chance that your clamp meter had a laminated iron core? Because I'd expect something like that to heat up pretty badly in the high frequency magnetic field. Even a single turn, such as in a current clamp, would induce some hefty eddy currents in iron, whereas ferrite isn't nearly as conductive.

Shame about the Amprobe, they're a nice meter. Was it one of the analog ones? I have one from the 1970's, I think. I still bring it out every once in a while. Good piece of kit.

Josh Campbell

Mon Jan 28 2013, 12:10AM

Registered Member #5258 Joined: Sun Jun 10 2012, 10:15PM
Location: Missouri - USA
Posts: 119

I updated my thread after you had started your reply, yes it was an iron laminate core. I figured I could jump on quickly before much heat built up but ended up getting distracted as sometimes happens when the power starts flowing. It still works, just a little more roughed up. I only use it where there is a possibility of loosing the meter since the Fluke 365 is easier to read, more accurate and (mainly) more expensive.

It's an old model RS-3 my father gave me. Takes a lickin'!

Zum Beispiel

Mon Jan 28 2013, 09:56AM

Registered Member #514 Joined: Sun Feb 11 2007, 12:27AM
Location: Somewhere in Pirkanmaa, Finland
Posts: 295

Hehe, toasty.

I've got a RS-3 too, inherited from my dad as well. I'm proud to say that mine is a little bit more banged up. Still works like a charm, though.

1 2 next

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