+10KW induction heater / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

+10KW induction heater

1 2 3 4 next

Move Thread

LAN_403

Linas

Sat Jul 07 2012, 06:09PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Ok, here is the deal, I need to melt some material (magnetic) and reach 2500C.
so I guess I will need more than 10kW for this to happen.
So I have got 3x SKM200GB125D and one for the final version SKM400GB125D

+-12V power supply for a +-24V Gate driver with 2:1:1 ratio

Here is resonant controller which doesn't work as I would like, a lot of noise and it even doesnâ€™t work as it should.
controller is MC33067

Capacitor bank for bridge power supply 3phase capable with middle point for matching transformer

IGBT GDT driver is made from Full-bridge IRFZ34N iRF9Z34N mosfet's and TC4422 drivers.

LC have 30kHz Fo , with 12.25uF 450A (+ Celem 1uF 550A 500V) capacitor bank sandwiched with two 5mm copper plates.

water cooled coil, noting very interesting 6mm diameter

For galvanic insulation I will use a matching transformer which is also capable to reach around 30kW (from CT scanner flyback transformer)

3 Phase rectifier for the final version (60KW capable rectifier)

Uge looks like this, and I am quite happy with that (later I can change it to +-15V )

I have a few questions:
Should I use parallel LC circuit or in series?
And what kind of driver should I use? I need only power control

radhoo

Mon Jul 09 2012, 01:58PM

Registered Member #1938 Joined: Sun Jan 25 2009, 12:44PM
Location: Romania
Posts: 699

given your experience with electronics, you should post the answers not the questions :) btw, yet another great project

Steve Conner

Mon Jul 09 2012, 02:29PM

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

Looks good! :)

Linas wrote ...

LC have 30kHz Fo , with 12.25uF 450A (+ Celem 1uF 550A 500V) capacitor bank sandwiched with two 5mm copper plates.

It looks like you've used ordinary steel bolts to mount the capacitors. They may get very hot especially if the RF current is allowed to flow through them. I used brass bolts for my IH tank circuit.

However, I've used steel bolts for connections in Tesla coil tank circuits, and it seems to work provided the bolts are just clamping a direct copper-to-copper contact.

The LCLR and series-feed tank circuits behave quite similarly from a drive point of view.

Linas

Mon Jul 09 2012, 03:51PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Steve Conner wrote ...

Looks good! :)

However, I've used steel bolts for connections in Tesla coil tank circuits, and it seems to work provided the bolts are just clamping a direct copper-to-copper contact.

The LCLR and series-feed tank circuits behave quite similarly from a drive point of view.

Yes, they just clamp massive brass plates to copper ones.
My question is, if I want to melt steel, I need to use current or voltage LC circuit?
The only reason I think of putting in series, is that they can use more from HV capacitors, and since I have quite a low voltage I should use a parallel LC circuit
Or is it something i don't know? Which circuit is most suitable for melting steel?
For now, I will try to fix my MC33067 current feedback controller

Dr. ISOTOP

Mon Jul 09 2012, 03:51PM

Registered Member #2919 Joined: Fri Jun 11 2010, 06:30PM
Location: Cambridge, MA
Posts: 652

I've used steel bolts in my IH with no ill effects.
Since you're using a transformer, you can safely use a series-fed topology, since the transformer can perform impedance matching between the tank and the inverter.
A simple way to control power is simply to detune the driver. This works, but is lossy. Alternatively, you could try a thyristor-based rectifier feeding the bus, and just modulate the bs voltage.
Oh, and nice Celem, BTW :)

Linas

Mon Jul 09 2012, 03:57PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

bwang wrote ...

A simple way to control power is simply to detune the driver. This works, but is lossy. Alternatively, you could try a thyristor-based rectifier feeding the bus, and just modulate the bs voltage.
Oh, and nice Celem, BTW :)

Yes, MC33067 works that way, it goes to into higher side from Fo, and it keeps zero voltage switching as well, so no loss here ( as far as I understand or is it just at max power, i don't know)
and best thing, it has 0 phase difference since it doesn't use any feedback for oscillation whatsoever
Here is the problem
MC33067 is set up to limit 60kHz and with a current control oscillator and a 25k resistor gives 21kHz oscillation frequency to an error amplifier output when E/A is low (a high one gives 60kHz oscillation frequency)
Problem is, that the current limiter doesn't work, when I turn on the power very low , it jumps to 60kHz, and if I set just a bit more the oscillation goes to 21kHz, and I can't get anything in between.
Here is the principal circuit of MC33067:

1341917792 1143 FT141346 2

and here is my feedback part:
1341917698 1143 FT141346 Untitled

I use bleed resistor after 1N5819 rectifier so I don't have to deal with Vf
CT set to 10A/V so I should limit the power at 51A with my Vref which is 5.1V

Brass , aluminum, steel
Steel washer just cut my crucible in half so I have to stop melting. it's only 1/5 of the original power :)
and here is problem with CCO

Melting copper, still no luck with steel :(

Linas

Sat Jul 14 2012, 05:00PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

still stuck with feedback problem, can't get it working on MC33067,
clearly enough i get same problem with PLL driver
If some one point out problem so my feedback will run as i would and clear power regulation via CT and current set potentiometer is achieved, i can donate Celem cap for that person

Right now problem is that low power regulation is working, and getting closer and closer to Fo i get lot of noise and eventually it jumps below Fo,

Bridge is solid, can deal with 220V ,blowing 16A fuses with no sweat

Steve Conner

Sat Jul 14 2012, 05:10PM

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

If you ever allow the frequency to get close to f0, it is likely to fall under f0, at which point it will get stuck at the lowest frequency. This is because the feedback in the current control loop changes sign and becomes positive below f0. At f0 itself the loop gain is zero, and I think this is why you see the noise.

The solution (IMO) is to increase the lower frequency limit so it is above f0 under all conditions, and increase the DC bus voltage until you get your desired power level.

You can also run in the leading region below f0 by swapping the terminals on the error amp and resetting the limits. IGBTs may run more efficiently in this mode.

Linas

Sat Jul 14 2012, 05:19PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

good idea, i can try that.
below resonance i get a little bit of noise but its around 5-10% bus voltage.
But in that operation i will have hard switching , not so good

update

get same results

Steve Ward

Mon Jul 16 2012, 06:38AM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

So i think the problem is with your current feedback circuit. The RC time constant of 30 ohms and 1nF is only 30nS, so this is not effectively filtering. After this you have 1k and 330pF, only a 330nS time constant, again very fast compared to the 20-60khz signals you are trying to average and regulate.

You could either 1) insert another diode (and incur its forward drop) after the burden resistor so that you just capture the peak, and feed it into a much larger RC filter, or 2) make your capacitors way bigger in the existing circuit to achieve a smoother averaged signal.

I think your regulator must be going crazy trying to keep up with a feedback signal with so much ripple.

EDIT:

After looking at your error amp circuit some more, i have some questions. Is the E/A amp a voltage to current amp? If you used a regular voltage type op amp, it looks like it would just behave like a comparator.

Another thing im not clear on is how the error amp drives the frequency. Right now it looks like it starts out at a high frequency and then drops the frequency as the current reaches the target level. This would be the opposite of the control i would employ, which is to have the controller default to Fres, and have the frequency increase as the current reached the target level. This increase in frequency should provide inductive switching for the inverter, which i believe is preferred over capacitive load switching.

1 2 3 4 next

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