Current limiter for induction heater / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Current limiter for induction heater

previous 1 2 3 next

Move Thread

LAN_403

Proud Mary

Sun Aug 09 2009, 08:10PM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

Hi Tonskulus, if I remember rightly, your system uses a large power triode, or tetrode, I forget which.

A fixed Z match is no use at all if the load impedance is altered,

You need a system which can match your source to varying loads within the range to be expected in induction heating.

Does you system run at about 13MHz? Have I remembered that correctly?

If so, then so much the better than having a much lower frequency, which could make the variable capacitors and inductors needed of unmanageable size.

Ideally, if you have money, I would choose vacuum variable capacitors, but a good tip for the poor like me is to choose big, old fashioned three or four gang variable capaciors (i.e. 4 x 500pF) and then put the capacitor in a tank of dielectric oil. This will - very roughly - double the capacitance, so you will now have a 4 x 1000pF variable capacitor capable of working at some kilovolts. If you can find a wide-spaced transmitting variable and put that in oil, it will withstand tens of kB before breakdown occurs - and if it does, the oil bath makes it instantly self healing.

In parallel with this, selectable by a big ceramic switch, you can put a range of capacitors, to make your range of variable capacitance almost unlimited if needs be.

And you will also need a variable inductor. You don't need full variability, so you can wind a large, long coil, with say a tap every three turns along its length - a little twist of wire to make a small loop from which you should remove the insulation. Then you have a shorting wire, connected to one end of the inductor connected to a crocodile clip. You can vary the inductance by moving the crocodile clip from one tap to the next.

A hot-wire ammeter can show you the RF current going forth to the load, and an SWR meter can show you reflected waves, coming back from the load, and therefore of no use in making heat. (tip: make your own SWR meter, don't buy a cheap CB type, which is about as much use as old cabbage soup.)

I would try the old fashioned pi-match first - a variable capacitor, the variable inductor, and then the large oil filled capacitor I have suggested above. I have no idea of the impedance range to be matched, but this would be my starting point.

I certainly would not have anything but an air core in the inductor, as you don;t want to waste lots of power heating up a load of ferrite.

There are more complex, but still simple, combinations of variable inductance and variable capacitance that are claimed to be able to match anything to anything. I'll find you some circuits if that would be helpful.

The ideas I have suggested are, of course, old fashioned, and 'low-tech' but if you wanted to have a totally automated impedance matching system, with lots of ICs and EPROMs and servo motors you should look at the designs of HF automatic antenna tuners after talking to your bank manager.

PS: I saw the pictures of VP riding bare-chested on his horse - it would be impossible to imagine our own leader riding like this unless it was in a wheel-chair being pushed by psychiatric nurses!

Hernan

Mon Aug 10 2009, 01:32PM

Registered Member #1614 Joined: Wed Jul 30 2008, 03:08PM
Location: Argentina
Posts: 52

Hi Steve,
In my system the current when curie is reached is about 2.5 to 3 times the initial current so it causes a trip of the inverter current limiter to protect mosfets. If I start with lower current before curie the heating time extends to much.

I use a open aluminum C clamp to the moving core.

Like Harry said an automated Lmatch coil is a bit complicated but In microchip page I found AN531 which describes a similar control with PID routine .

regards,
HernÃ¡n

GeordieBoy

Wed Aug 12 2009, 12:42PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

You really don't need to do anything as complicated as a mechanical servo adjusting a gapped inductor to limit Induction Heating current.

You can achieve the same thing electrically very easily. Drive frequency detuning, and DC bus voltage control are the most common methods. However, I have heard that simple burst-fire gating of the RF is effective at limiting current, as is phase-shift control of the two legs of the H-bridge comprising the inverter.

All of these techniques are likely to be much faster than a mechanical servo since they are entirely electrical. If limiting instantaneous peak currents to safe levels is your aim, then I would go for simple Modulation of the drive Frequency. If the LCLR work coil current or voltage rises too high then push the VCO (or DDS) up in frequency until the new desired operating point is reached.

-Richie,

Tonskulus

Wed Aug 12 2009, 02:28PM

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

Harry: I have another thread for my tetrode powered induction heater but its working just fine as it is now. I made this thread because i also have solidstate induction heater.

Frequency detuning is one method for power controlling but as im using self oscillating circuit, it cannot be done.

Zmatch inductor..
When i heat nonmagnetic materials or steel above curie, I use value that keeps input current around 15Amps. Now if I put some really heavy load, like big piece of steel rod in to the workcoil, current drops down to 1-2A. Its way too low power to heat such workpieces.

It is suggested to choose series cap and inductor values so that it is resonant above operating frequency with impedance that allows suficient current to flow. Or tuned to 3rd harmonics to allow full power from powersupply to the tank circuit.

So i think maybe best method would be that 3rd harmonics tuned LC series circuit AND constant current source (or variable voltage source).

Hernan

Wed Aug 12 2009, 04:26PM

Registered Member #1614 Joined: Wed Jul 30 2008, 03:08PM
Location: Argentina
Posts: 52

Hi , thanks all for the advices!| this forum is very useful and inspiring to me
I Â´ve considered freq. detuning since is simple to implement but I want to keep the inverter locked at resonant freq because I noticed that Mosfets stay cool it and worked reliable for several hours melting all types of loads. I need to adapt to diferent size and material loads for example when I use graphite crucibles I open Lmatch to allow current flow because Q is low since it near workcoil in other hand when I melt cooper in a ceramic crucible I open Lmatch a bit because tank voltage rises to high.
Varying DC voltage would be an option...but I'll need more semiconductors complicating the system.
Richie, when you said Burst fire is same as pulse density modulation?
Tonskulus , In case of voltage-fed inverter with series connection to a parallel load you'll need extra power capacitor in series with Lmatch?

HernÃ¡n

GeordieBoy

Wed Aug 12 2009, 05:06PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

Hi Hernan and Tonskulus, all

Tonskulus: My comments were mostly directed towards Hernan after seeing his mechanically controlled variable-gap inductor.

> I Â´ve considered freq. detuning since is simple to implement but I want to keep the inverter locked at resonant freq because I noticed that Mosfets stay cool.

That is a fair comment, that switching losses due to commutating current will be lower if you switch close to the zero-current crossings. However, do not under-estimate the benefits of running a MOSFET inverter into a net inductive load, like that found by detuning the driver to a higher frequency. Unlike IGBTs, MOSFETs can switch large currents very quickly so do not typically experience problems due to loss of ZCS. However, high voltage power MOSFETS typically have large Cds and Cgd values which cause significant turn-on losses due to discharging of this capacitance into the channel at turn-on. This capacitive turn-on loss is one of the main contributors to switching loss in high voltage high frequency MOSFET inverters, particularly for large die devices.

Operating the bridge into a net inductive load with generous dead-time ensures that chage is swept out of the output capacitance before each device is commanded to turn on. Therefore ZVS is achieved and turn-on losses are eliminated.

Burst fire is also known as pulse position modulation or pulse density modulation. I know that Steve Conner had good results using a modified DRSSTC controller with over-current trip to control induction heating load current in a series resonant induction heater. The matching network of the LCLR topology essentially performs an imedance inversion on the parallel resonant tank circuit making it appear like a series resonant tank, so I don't see any reason why the same scheme won't work for LCLR topology also. Significant practical advantages may be acoustic noise, and conducted EMI (flicker noise) back through the mains wiring though!

Finally you might want to consider phase-shifting the drive to one bridge leg as another method to achieve power control. Most SMPSU manufacturers produce Phase shifted full bridge controller ICs now. A similar technique of "out-phasing" can also be done with two half-bridges when their outputs are combined in parallel through seperate L-matching inductors. If both half bridges are commanded to switch in phase their powers combine. If both are phase shifted to be in antiphase, current only sloshes back and forth between them, and no current is sourced to the parallel resonant IH tank circuit. Phase shift values in between 0 and 180 degrees give you varying power levels. Those familiar with old AM radio transmitter designs will recognise this as the "Ampliphase" technique used by RCA! It is very effective, particularly in HF induction heating because it maintains the inductive current presented to each bridge leg even as the overall load current is reduced right down to zero. This means that ZVS can be maintained over a large range of power settings.

I hope this info helps,

-Richie,

Hernan

Wed Aug 12 2009, 10:32PM

Registered Member #1614 Joined: Wed Jul 30 2008, 03:08PM
Location: Argentina
Posts: 52

Richie, thanks for your information it always helps!.the control techniques that you mention are only suitable with full bridge inverters? because I use Half Bridge with a 2:1 HF transformer .

I will try to use the automatic servo for curiosity if it works. I'm starting a simple program to see what happens...
I was thinking if LCLR network will work if I tranlate Lmatch to the primary side of the transformer to lower the current trough it .will work?

thanks for your time
HernÃ¡n

GeordieBoy

Wed Aug 12 2009, 10:49PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

The variable frequency drive, DC bus regulation, and burst-fire methods should work equally well with a half-bridge or with a full-bridge (H bridge).

The phase-shifted full bridge control requires two bridge legs so obviously needs a full-bridge with the load network connected between the two bridge legs. The technique I called out-phasing also requires two bridge legs but the two outputs are combined in parallel by being tied to the same end of the work coil through two matching inductors.

Yes, you can definitely move the L-match coil to the primary side of your impedance-matching / isolation transformer. In fact it is desirable to do that so that you don't have to water cool the winding. The leakage inductance of the isolation transformer just adds to the inductance of the L-match inductor anyway, so it's not critical at all.

You may even find that your isolation transformer runs cooler because it will see a more sinusoidal impressed voltage when the Lmatch coil goes between the inverter and the transformer. Less high frequency content reduces core losses, skin effect and proximity effect in the windings.

-Richie,

Hernan

Thu Aug 13 2009, 12:55PM

Registered Member #1614 Joined: Wed Jul 30 2008, 03:08PM
Location: Argentina
Posts: 52

Hi all!
Richie, so if I translate Lmatch to primary side the inverter will 'see' the same load without needing to calculate L match turns? and transformer turns must be half now to reach the same current?

Load will be: Lmatch in series with leakake inductance of transformer and then connected to parallel tank impedance multiplied turns ratio squared in parallel with magnetizing inductance of the transformer

thanks!
HernÃ¡n

GeordieBoy

Thu Aug 13 2009, 03:41PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

If you move the L-match inductor from one side of a transformer to the other, (ie from secondary to primary side) you need to recalculate the required value.

The effective inductance of the L-match network will be multiplied by the square of the turns ratio when you move it from one side of the transformer to the other.

So for example, if you moved your high-current, low-inductance water-cooled matching inductor from the secondary side of the isolating transformer to the primary side, you would now use a low-current, high inductance wire-wound one no longer needing water cooling.

-Richie,

previous 1 2 3 next

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