Induction heater research (questions) / General Science and Electronics / Forums

Forums

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

« Previous topic | Next topic »

Induction heater research (questions)

Move Thread

LAN_403

The Ideanator

Sun May 17 2015, 05:18PM

Registered Member #54599 Joined: Sat Mar 07 2015, 06:09AM
Location:
Posts: 8

Ok so my goal for the summer is to build a compact induction heater for general mayhem and perhaps brazing or hardening of tools as well. I've got tons of time, not much money and I have a patchwork knowledge of how this all works so I need some help filling in the blanks. My primary design inspiration and assumptions on this whole thing comes from here

1) I keep seeing emphasis on high voltage caps for the tank which doesn't make much sense as the work circuit (tank and work coil fed by a coupling transformer) ends up being fairly low with impressive levels of current(100+A). Are HV caps really necessary or would high current ones be better?

2) Since the goal here is to heat metals via a magnetic field and magnetic fields are dependent on current, the higher the better, why is optimization not a thing? Some back of the envelope math for an anticipated system seems to suggest that with the available wattage I could put around 1000A into the work circuit (assuming no losses and that it is in resonance). I'm not sure what this actually means. I suspect it means I can charge the tank really fast and have a lot of leftover current to directly drive the induction. I think I'm missing something, perhaps the rate at which the components of the work circuit are charging but I'm not sure how to solve that.
Optimization

3) What is a good way to build a good thermal joint that is nonconductive? Would something like this

work for hooking an IGBT to a water block?

4) What are some good methods of varying the power in the work circuit? I'd like to use a foot pedal to have some control over the temperature of a static workpiece.

Wolfram

Sun May 17 2015, 06:08PM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

1) I keep seeing emphasis on high voltage caps for the tank which doesn't make much sense as the work circuit (tank and work coil fed by a coupling transformer) ends up being fairly low with impressive levels of current(100+A). Are HV caps really necessary or would high current ones be better?

The work coil, like any inductor, will have a given reactance (AC resistance) which is given by the equation Zl = 2*pi*F*L. To get a given current through the coil, you need to apply a voltage according to ohms law, U = Z * I. Lowering the inductance of the coil without changing the coil or workpiece geometry will result in less heating of the workpiece for a given coil current. The same goes for lowering the frequency. So there's no way around needing significant voltage in the tank circuit to transfer significant power.

2) Since the goal here is to heat metals via a magnetic field and magnetic fields are dependent on current, the higher the better, why is optimization not a thing? Some back of the envelope math for an anticipated system seems to suggest that with the available wattage I could put around 1000A into the work circuit (assuming no losses and that it is in resonance). I'm not sure what this actually means. I suspect it means I can charge the tank really fast and have a lot of leftover current to directly drive the induction. I think I'm missing something, perhaps the rate at which the components of the work circuit are charging but I'm not sure how to solve that.

Optimization is always about tradeoffs in the real world, and there's no optimal design, it all depends on the application.

I find it easiest to start from the work coil. The electrical model of the work coil mainly consists of three components. The first and most obvious is the inductance, let's call it Lc. The purpose of the tank capacitor is to "cancel" this inductance. The second component is the resistance of the work coil, let's call it Rc. This one represents losses, and is more difficult to measure. It will be significantly higher than the DC resistance of the coil due to eddy currents (skin effect and proximity effect), and it rises with increasing frequency. The third component is the reflected load resistance, which we'll call Rl. This comes from the part you want to heat, and depends on workpiece geometry, resistivity and coil geometry. This one also rises with increasing frequency.

The amount of power delivered to the workpiece is simply I^2 * Rl, and the power lost in the coil is I^2 * Rc. The efficiency of the coil is thus Rl / (Rl + Rc), and this is determined only by the geometry of the workpiece and the coil. Adding turns to the work coil will increase both Rl and Rc by the same amount, and the same goes for increasing the frequency. Here's a nice chart to show the efficiency as a function of coil and workpiece geometry

. Notice that there is a frequency below which the load resistance will fall faster than the work coil resistance, due to eddy current cancellation in the workpiece. This is called the critical frequency, and efficiency usually starts to fall when the workpiece is thinner than four skin depths at the operating frequency.

The whole thing is a bit involved, and the main thing to note is that for a given work coil and workpiece, more power will be delivered to the workpiece for a given current when the frequency is increased, but due to the inductance of the work coil, you also need more voltage to drive a given current through the work coil when the frequency is increased.

For some more in-depth theory, check out this excellent book which is out of copyright:

3) What is a good way to build a good thermal joint that is nonconductive? Would something like this work for hooking an IGBT to a water block?

For low thermal resistance, you want something with good thermal conductivity. You want a large area, and a thin insulator. I like to use kapton foil, it doesn't have particularly low thermal resistance, but it's very tough and has a high breakdown voltage, so you can get away with using a very thin layer. The link doesn't work here.

4) What are some good methods of varying the power in the work circuit? I'd like to use a foot pedal to have some control over the temperature of a static workpiece.

I prefer to use pulse density modulation, it's a bit more complicated than other options, but losses are very low. See my thread in the Projects subforum for more info. If you run at a lower frequency or use MOSFETs, you can get away with detuning for power control, like in this circuit

The Ideanator

Wed May 20 2015, 08:59PM

Registered Member #54599 Joined: Sat Mar 07 2015, 06:09AM
Location:
Posts: 8

So there's no way around needing significant voltage in the tank circuit to transfer significant power.

My intention with those equations is to find the lowest necessary voltage in the tank circuit in order to maximize the current. The reason for having a tank cap in parallel with the work coil as I understand it is so that at the resonance point they have the lowest resistance and therfor the maximum current for a given voltage, that voltage being just the DC resistance of the work circuit since the reactance of both the cap and coil cancel. see: Z = sqrt( R^2 + (X_L - X_C)^2 ), therefor Z = R when X_L = X_C.

there's no optimal design

In practice that is absolutely true, but currently I'm trying to nail down a theoretical system that I can approximate later on.

The second component is the resistance of the work coil, let's call it Rc. This one represents losses, and is more difficult to measure. It will be significantly higher than the DC resistance of the coil due to eddy currents (skin effect and proximity effect), and it rises with increasing frequency.

You're talking about the reactance of the inductor, right? The work circuit is an RLC one and should be analyzed as a whole thing instead of part by part, or am I missing something? Are you able to just ignore the other components in the system? I'm also not entirely clear on how proximity effect would affect the system outside of the coupling transformer. I read through your thread wolfram, have you figured out how to mitigate that yet?

Another couple of questions:
I've got some ~1uF microwave caps laying around doing nothing and I can't find good specs on them, do you guys know if they would be suitable for tank caps?
How would I go about using matlab to simulate stuff? I've got a student version and no idea how to use it effectively.

Ash Small

Wed May 20 2015, 10:21PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

The inductance of the coil plus workpiece changes as the workpiece heats up. Curie point is what you want to google here.

If you had enough MO capacitors in a parallel series array it might work, but they are not designed for high frequencies (at least, that's the simple answer)

The secret to success with these things seems to have a lot to do with using suitable capacitors, otherwise they have too much 'equivalent series resistance'.

GrantX

Sun May 31 2015, 06:37AM

Registered Member #4074 Joined: Mon Aug 29 2011, 06:58AM
Location: Australia
Posts: 335

I think Wolfram was referring to both the inductance of the work coil and its actual DC resistance (the resistivity of the length of copper pipe/cable). The resistance is only a few milli-ohms but this still causes losses when theres hundreds or even thousands of amps flowing.

Microwave oven capacitors are enticing due to the high voltage rating and the capacitance is convenient for induction heaters, but unfortunately they perform terribly. They're only designed for 50/60Hz applications and have huge losses at high frequencies. The lead wires between the terminals and foil plates are a bit feeble too, they can probably only withstand a few amps continuously.

Multi-kilowatt scale heaters usually use large water cooled capacitors with heavy duty terminals (I've got a pair of ~600A Inductotherm caps with 80mm x 80mm x 5mm copper plates for the electrical connections and water cooling). I remember seeing a few DIY heaters over one kilowatt that simply use a big pile of cheap polypropylene caps to handle the current. Smaller heaters around 500W can get away with a couple of cheap high current snubber caps, and I used a single IGBT snubber in my ~200W ZVS heater (can't remember what the package style is called, the ones with flat terminals designed to be directly bolted onto a IGBT brick).

Uspring

Sun May 31 2015, 03:24PM

Registered Member #3988 Joined: Thu Jul 07 2011, 03:25PM
Location:
Posts: 711

Ideanator wrote:

The reason for having a tank cap in parallel with the work coil as I understand it is so that at the resonance point they have the lowest resistance and therfor the maximum current for a given voltage, that voltage being just the DC resistance of the work circuit since the reactance of both the cap and coil cancel. see: Z = sqrt( R^2 + (X_L - X_C)^2 ), therefor Z = R when X_L = X_C.

That would be true for a serially driven tank. For a lossless parallel one, reactance is infinite at resonance. The advantage is, that there is little power draw once the metal you want to heat, has molten away.
Without a cap, there still would be a strong reactive current loading the driver, even if there is no metal present.

The Ideanator

Fri Jun 05 2015, 04:20PM

Registered Member #54599 Joined: Sat Mar 07 2015, 06:09AM
Location:
Posts: 8

In relation to my optimization equations to determine if they are even necessary, has anyone measured the actual current in the work circuit, with and without things to be heated? Another way to put it, how is the maximum current between the resonant elements determined? I know that in a series work circuit it would be dictated by ohms law which would be limited by the input wattage which brings me right back to optimization. What effect do actual components have here? I'm stuck in an infinite loop, send help.

have huge losses at high frequencies. The lead wires between the terminals and foil plates are a bit feeble too, they can probably only withstand a few amps continuously.

That is what I wanted to know.

That would be true for a serially driven tank.

Yep, and thatâ€™s what I'm constructing.

IamSmooth

Mon Aug 24 2015, 04:19AM

Registered Member #190 Joined: Fri Feb 17 2006, 12:00AM
Location:
Posts: 1567

The Ideanator wrote ...

how is the maximum current between the resonant elements determined?

If you put differential probes across the tank capacitor you will measure the voltage drop. If you know the frequency, f, and the capacitance you know the capacitive reactance. V/R equals the current. You will find it is quite large. With high enough currents you generate large magnetic fields which can do this:

The Ideanator

Tue Sept 01 2015, 11:23PM

Registered Member #54599 Joined: Sat Mar 07 2015, 06:09AM
Location:
Posts: 8

IamSmooth I was hoping you'd show up!

My initial goals, borne from the mindset of a physics/engineering student who must quantify and optimize everything, are probably not necessary for version 1. I don't think I need very much power (not that I have much available anyway) to melt silver solder so I'll save it for when I know what I'm doing. In the mean time I think I'll use your designs. I like the auto-locking arduino version but I can't find the code and I don't know enough about what's going on to write my own. Is there any chance I could see your code or pseudo code if you still have or remember any of it?

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