Ferrite transformer driven by induction heater

[Phoenix]

Hello everyone 

3 years ago I built a driver for a large DIY ferrite high voltage transformer. The results were impressive, but I always had the problem of failing IGBT modules. The reason was that the transformer primary was directly connected to the halfbridge which caused enormous peak currents everytime an arc ignited. This pushed the IGBTs into desaturation which causes massive losses and eventual failure.

I knew back then that the solution would be a LLC resonant converter which acts like a current source. But i did not further pursue this idea back then since a proper control aproach is rather complicated.

Recently I finished building a very powerful 60kW PDM induction heater. The induction heater controls the resonant current based on slow decay freewheeling. Here is a quick overview over the induction heater, you can also find further test videos of it on my channel:

/>
I suddenly had the idea that I could hook up my old ferrite transformer to the induction heater. I did some measurements with my LCR meter to make sure that the resonant frequency stays in an acceptable range during the whole operation spectrum (open secondary, shorted secondary, core fully saturated). I decided on 2 primary turns and a coupling factor of roughly 0,6 (airgap). This led to a theoretical frequency range of 31-55kHz. In practice the transformer seems to constantly operate at around 37kHz.

So far the whole setup is experimental, but I was able to generate up to 35kW arcs until now. I made a secondary with 250 turns, that should roughly lead to the maximum voltage that I am willing to put on the transformers isolation. The secondary has a center tap which is grounded, so only 1/2 the voltage is there in reference to the core.

The output voltage is around 30kV right now. Here you can see a picture of the transformer:


The arcs are absolutely insane, I have never seen something like that before live. Sunglasses for UV protetction are not enough, my eyes hurted a bit in the evening later. But right now I am using oxy-acetylene welding glasses with Shade 5 glass, those work really well. The whole basement fills up with nitrogen dioxide really fast, I have to open the windows even after short runs.

Here is the video:

I really have to try to stretch the arc vertically, so far the arcs are about 175cm long. I think with a standing arc I should be able to do 2m long arcs that continiously burn.



The core is pretty much on its limit right now, its fully saturating. With the airgap and primary turns configuration it saturates at around 530Arms but, I am pushing 900Arms into it. I am thinking about getting an even bigger ferrite core from a small german company, but its pretty expensive sadly. The induction heater itself could support a lot more power than 35kW, but I am not even sure if I want to go any further than that, the output is getting really scary.

Best
Phoenix

[PowerTech]

Very good result, insane arc 

[Uspring]

That is a very impressive design with equally impressive results you are showing here. You could possibly power a TC with this also.
I enjoyed your mad scientist looks 

[davekni]

Impressive project, both your induction heater and the HV arc accessory!

The core is pretty much on its limit right now, its fully saturating. With the airgap and primary turns configuration it saturates at around 530Arms but, I am pushing 900Arms into it. I am thinking about getting an even bigger ferrite core from a small german company, but its pretty expensive sadly. The induction heater itself could support a lot more power than 35kW, but I am not even sure if I want to go any further than that, the output is getting really scary.

I suspect you could get higher power from your existing core by placing primary coil around the center of secondary coil on the same ferrite leg.  This would reduce leakage inductance (increase coupling), increasing secondary current capability for a given magnetic flux in the core.  Don't know enough about your induction heater circuitry to guess how this lower-impedance load would affect it.

As you said, perhaps higher power isn't a wise goal.  Certainly keep up with good eye protection and ventilation.

[Phoenix]

Thank you for your replies, I appreciate it 

I suspect you could get higher power from your existing core by placing primary coil around the center of secondary coil on the same ferrite leg.

Yes, that would indeed work. The benefit of doing this is that it would increase the coupling without reducing the airgap. This is a good thing because reducing the airgap leads to earlier saturation. The issue is just that this is not as convenient as having the primary on the other side, I will see how much more coupling I get.

I have been thinking about this a while now on how to optimize the whole setup mathematically. I want to optimize for maximum real power transfer which means I want to maximize coupling and reactive power on the primary side. However, there are tradeoffs by doing both things and there should be a point where the transformer performs best.

If i want to make the coupling larger I can make the airgap smaller. This leads to more inductance which would mean I can only use a single primary turn in order to not have a too low resonant frequency which is bad for maximum power transfer. On the other hand reducing the airgap leads to earlier core saturation which is bad and limits my reactive power. Another thing that also comes into play is that my resonant capacitor is limited to 1000Arms, so going too high in frequency also limits my reactive power since the voltage becomes smaller.

I have almost all the equations to solve this, I am only missing one thing. I know that the coupling factor influences the voltage in such a way that the open loop voltage is (V/turns)×turns×k. But what about the influence of coupling on the short circuit current? I found nothing about that online but I played around in LT Spice and figured out that it also appears to be the current according to turns ratio multiplied by k. So can I assume that the maximum achievable power depends on k^2 since power is voltage times current?

Another thing that really surprises me are the open loop voltages I am achieving. They are way higher than what I could get from the V/turns. My turns ratio is 2:250.
My primary inductance is 2.3uH with open secondary. If i apply 200Apk with my induction heater I am getting 90V across the primary at a fres of 31kHz which leads to 45V/turn. Calculating the open loop voltage from that (45×250×k), I am expecting 6750Vpk. However, the transformer easily ignites an arc over 3cm distance at 200Apk. 

This is really weird since this translates to like >30kV with this electrode shape. The core is also still far away from saturation at 200Apk with this airgap, I can not explain the high open loop voltages. Secondary resonance is also impossible since I would need like 500pF to resonate at these frequencies.

Best
Phoenix

[Uspring]

So can I assume that the maximum achievable power depends on k^2 since power is voltage times current?

The max power of an ideally resistively loaded transformer for a given input voltage V is

Pmax = 1/2 * V^2 * k^2 / (1-k^2) / (ω*Lpri)

But possibly more relevant is that for a given input current:

Pmax = 1/2 * I^2 * k^2 * ω * Lpri

Current and voltage quantities are RMS. For the parameters you chose it looks like you can transfer all available power at max input current. The calculation presupposes an ideal arc load, though.

Another thing that really surprises me are the open loop voltages I am achieving.

In the case of low loads as at the time before the arc ignites, the PDM runs probably at a low duty cycle. If your current meter averages currents, you might underestimate the rarely occuring peak values.

[davekni]

   Another thing that really surprises me are the open loop voltages I am achieving.

In the case of low loads as at the time before the arc ignites, the PDM runs probably at a low duty cycle. If your current meter averages currents, you might underestimate the rarely occuring peak values.

Best way to figure out may be a single turn around center of secondary wired to a scope.  Should be a somewhat-reasonable match for scaled secondary voltage.
My guess would be either some sort of transient to higher primary current as Uspring mentioned, or high frequency components (harmonics) on the induction unit output.
One possible source for harmonics is core saturation.  That seems unlikely at 200A based on what you've said.
Perhaps more likely is H-bridge switching.  H-bridge output voltage is scaled by your induction supply's transformer turns ratio and added to resonant voltage at output terminals.  Even though scaled H-bridge output voltage should be relatively low, perhaps transitions hit secondary resonant frequency, causing high output voltage.  Higher-frequency output voltage will help with arc starting too, reducing required peak voltage.   Scoping secondary voltage through a 1-turn winding should be able to detect this.

Fascinating project!  My little 1kW version (inspired by your initial project) has +-5.5kV peak output voltage before arc starts.  Jump distance is very small, ~2mm.  Perhaps your data about 3cm arc initiation will inspire some simple addition to my setup to start longer arcs.