Novice-Built Induction Heater (In 14 days?) / General Science and Electronics / Forums

Forums

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

« Previous topic | Next topic »

Novice-Built Induction Heater (In 14 days?)

1 2 3 next

Move Thread

LAN_403

AwesomeMatt

Fri Aug 22 2014, 04:25AM

Registered Member #4454 Joined: Sun Feb 26 2012, 12:47AM
Location: Western Canada
Posts: 74

Hey everyone,

I'm going to try to build a 6kW induction heater in a couple weeks, without much knowledge. I could really use some help.

I've been wanting to build an induction heater for a couple years for my local hackerspace. It was just never worth the personal investment. Well, I got funding from a small arts grant (a couple hundred bucks). I'm building it and then I'm donating it for everyone to use.

Caveat - I should have it ready for our local mini Makerfaire which is in... 14 days. (I have lots of free time in these next 2 weeks).

I'm a complete novice but I've been doing a lot of reading the last couple weeks. I've somewhat stopped progressing. Most explanations answer my questions in ways I don't understand, I might need it dumbed down more.

Specs:

- Should be able to do molten aluminum levitation demos at events (oooh, ahhh).
- Will be used for casting/forging projects the rest of the year.
- Needs to run on 120v (for events)
- Should run on 240v (back at the 'space, err, actually 208v since it's 2 arms of 3-phase).
- I'm targeting 30A @ 208V, so... ~6,000W.

I have not settled on a design, much, except that it'll be series-resonant.

I understand some of the big pieces of the circuit, but not others. I have seen IAMSMOOTH's tutorial

, but it doesn't entirely make sense to me. I have read through some other induction heating projects on 4HV too.

I try my best when soliciting advice from the community to at least well-document my project with photos and video on the fly, and (eventually) tutorials later.

Resources/What I Understand:

- Ohm's law, Power law, Kirchoff's laws.
- I have access to lots of scopes & signal gens, but I haven't used them in years (rusty but capable).
- Huge assortment of general parts (resistors, caps, regulators).
- Fairly good access to scrap electronics (have salvaged/stripped a bunch over the years).
- A year's worth of school study for C++/Java/etc. I can understand and modify other's code, but probably not write anything from scratch.

What I Need Help With:

- Parts list!! - Before I even understand what I'm doing, I kinda need to go go go go go, as the fancy parts have to be shipped ASAP (I'm in Canada).
- Parts shopping - I've always just scavenged parts, so I'm not all that familiar with "Oh, you'd want to buy this from here" or actually searching for the right part or specs.
- Explain like I'm 5 years old the basic building blocks of the circuit. - Higher-level explanations are common, but they don't quite sink in enough for me to solve my own problems.
- Arduino stuff - I would like it to be auto-tuning, I've been meaning to tinker with micros for years but haven't.

Plan/Progress So Far:

- Settled on series-resonant
- 10x 1.25uF, 600vac 46A (each) tank caps (to be put in parallel)... if they'll ship them to Canada... in time.
- ~15khz resonance (??? ish ??? does that sound right?)
- I have 1/4" copper tubing
- I have a few square feet of double-sided circuitboard blank (I have both CNC mill & etching equipment).
- I have a couple small radiators and/or a couple really big buckets
- I have sheet metal/acrylic for enclosure
- I have a big beefy 50A rectifier
- I have plenty electrolytics to smooth the wall voltage to ~300VDC (or 170VDC on a 120v circuit).
- I have lots of heatsinks
- Tracking down sources for several large ferrite torroids (probably good there)
- I want a micro controller to auto-tune, Arduino or otherwise, I don't care, but I'm a complete novice.
- I need a "buck converter" (?) to drop the voltage down, as my power control method. I know what this is, I don't know where to start building one to these specs.
- I'll use an H-bridge to control the inverting (I only vaguely understand these)
- I need an H-bridge driver of some sort?
- IGBTs for something?

The closer to output the more I understand. Rectified and smoothed line voltage I understand. But the closer to the input the fuzzier and fuzzier I get as to what controls what and what's happening and why and what parts I would need.

...

So, ASAP, I'd like to order any obscure parts that I will obviously need. I think I need at least a couple 600V half-bridge drivers (?) and some (?) 500V MOSFETS? I don't know how to shop for these, I feel like I'm reading ingredient labels for a language I don't speak.

Anything else that I wouldn't have in a basics electronic kit or from salvaging CRT/PSU/VCR/Etc boards for a few years?

Any advice, explanations, narrowing-downs or criticism very much appreciated.

Thanks in advance everyone who helps me tackle this. Again, I'm immediately donating it to the local maker community.

Dago

Fri Aug 22 2014, 05:09AM

Registered Member #538 Joined: Sun Feb 18 2007, 08:33PM
Location: Finland
Posts: 181

I honestly doubt you will make it. Even if you'd know 100% what you're doing and have everything you need at hand you'd still have a hell of a hurry in two weeks. Even ordering something will take several days and you'd need to figure out the design before you order anything.

Sigurthr

Fri Aug 22 2014, 05:32AM

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

I echo the above. The planning and build time alone (never mind troubleshooting) is considerable.

Pit falls to watch out for: series resonant will result in the driver overloading when no load is present. Unlike in parallel resonant where you can get away with no OCD protection, it isn't an option for series resonant. You'll also need to keep the inverter at resonance or above resonance (inductive region) - inverters do NOT like being run in the capacitive region.

Think of it as a DRSSTC with the secondary removed, this will help you realize the control schemes and basic processes you're working with.

Most use a PLL and not a uC. I think it can be done on a uC, at least for the current monitoring part, but that's like using a pneumatic press for a hammer to hang a picture in drywall. Arduino's aren't good at precise timing above a few kHz. I've done several projects where I've pushed the boundaries a bit and gotten good results, but all of the ones what worked involved fixed frequency output at a frequency the arduino was happy with. Getting continuous frequency range and accuracy isn't something arduinos are good at.

Designing a buck converter capable of 6kW is a HUUUUUGE project alone. Think you may have bitten off too much here.

Dago

Fri Aug 22 2014, 06:19AM

Registered Member #538 Joined: Sun Feb 18 2007, 08:33PM
Location: Finland
Posts: 181

I'm currently building an induction heater myself and I've already spent quite a lot of time (months, not continuous time though) just getting the (digital) control loop to work as it should (

). And it's still lacking in features such as adjustable phase difference and detuning. Though I did kinda spend a lot of time making a VHDL version as well at the same time:

(haven't posted anything about it yet). I think it might be possible to use an arduino for that if you divide the feedback frequency down but not really that optimal. I used an STM32 ARM myself.

AwesomeMatt

Fri Aug 22 2014, 08:06AM

Registered Member #4454 Joined: Sun Feb 26 2012, 12:47AM
Location: Western Canada
Posts: 74

I know it's ambitious. There was hubris in me thinking "Well, I've seen a handful of people make these before. I'll just pick one from plans and build it."

Except, everyone's is different and everyone's is for slightly different things.

I do have local help, but everyone's busy with their own Makerfaire projects and can't help design until... it's too late for me to start planning.

My backup plan is the quick 'n dirty parallel resonant heater (a couple/few hundred watts) for the faire, and build the big one later. But I figure I can slap one of those together in a few hours.

I know it probably won't do levitation, even of tiny aluminum cutoffs, but, at least it'll show heating.

As to the Arduino, that's what IAMSMOOTH does on his big 10kw unit and suggests doing. I'm probably more comfortable tinkering with code than I am bumbling through analog electronics. I don't care if it's overkill, if it means I get it done sooner, can tweak it more easily, and it's more robust.

Build-time does not concern me. I can throw 15 hours a day at it. "I don't know where to start" time worries me. That, and the later "I don't know how to solve this" time.

Either way, I'm proceeding.

Steve Conner

Fri Aug 22 2014, 08:25AM

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

A good starting point would be a series resonant heater with the Uzzors PLL driver circuit.

Check out the "OMG Induction Heater" thread on here for a good run down of what various people have built.

For the power levels you're planning, and especially for levitation you will need a water-cooled coil and some really good capacitors, also with water cooling. The capacitors are far and away the most important thing, the details of the driver hardly matter in comparison.

AwesomeMatt

Fri Aug 22 2014, 08:47AM

Registered Member #4454 Joined: Sun Feb 26 2012, 12:47AM
Location: Western Canada
Posts: 74

wrote ...
A good starting point would be a series resonant heater with the Uzzors PLL driver circuit.

I've already read through this a couple times, but, lacking context, was not sure if it was acceptable for my use.

Looks good, I'll read it again and probably just copy it outright if that's your suggestion.

wrote ...
For the power levels you're planning, and especially for levitation you will need a water-cooled coil and some really good capacitors, also with water cooling. The capacitors are far and away the most important thing, the details of the driver hardly matter in comparison.

Yes, the tank capacitors are what I've been spending most of my time on. I think I have that nailed down, *IF* they'll ship the ones I want to Canada.

1.25uF, 600VAC, 46amps continuous... and I'll take 10 of those in parallel.

I'm aware of water cooling requirements (and low-inductance pathing for the caps) and have that planned out.

It's more, the big gap between "Rectified & Smoothed 300VDC" and "Coupling transformer/Output Coil". All the H-bridge, half-bridge, PLL... I don't even understand what all those circuits are or how they work (in a macro sense).

I could really use a dumbed down ABC of what's going on. I have trouble contextualizing specifics when I don't have the big picture set. I keep reading specifics and I have no way to make sense of it so it doesn't add to my understanding.

I understand some part of it is pulsing on and off, and that gives me the frequency. I understand something about the H-bridge being an amplifier of that signal. The PLL using feedback from some part of the output to tune the frequency. There's something that drives the H-bridge because it doesn't do that by itself. I'm not sure exactly what happens to the 300VDC that it starts off at.

Lots of ??? in there.

Steve Conner

Fri Aug 22 2014, 09:43AM

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

I'll have a go then.

The "H-bridge" is an instance of the general circuit function that we call an inverter. (Not to be confused with an inverter in logic circuits.

)

An inverter is the opposite of a rectifier, it takes DC and turns it into AC using electronic switches that are pulsed on and off in some pattern.

To operate an induction heater we need high frequency AC in the region of about 100-500kHz. So, we use a rectifier to convert 60Hz AC into DC, followed by an inverter to convert it back to AC at the required frequency.

The "driver" is the function block that tells the inverter what to do. It does this by generating a small copy of the desired output waveform, which gets amplified by the transistors in the inverter. (Bearing in mind that in modern power electronics the transistors are always used as saturated switches, not linear amplifiers. The driver must supply a square wave drive signal big enough and fast enough that the devices are flipped quickly between off and fully on, hardly spending any time in between.)

For an induction heater the required signal is a square wave of 50% duty and variable frequency. The Uzzors circuit tries to find the resonant frequency of the work coil circuit and lock to it, but if the work coil voltage or current become excessive, it detunes away from the resonant frequency to limit it. This is a good general approach to the problem, though it is better suited to MOSFETs than IGBTs. With IGBTs you want to avoid detuning and keep them switching as close as possible to zero current.

This general topology of rectifier followed by inverter is found in solid-state Tesla coils and switchmode power supplies too.

AwesomeMatt

Fri Aug 22 2014, 11:02AM

Registered Member #4454 Joined: Sun Feb 26 2012, 12:47AM
Location: Western Canada
Posts: 74

wrote ...
An inverter is the opposite of a rectifier, it takes DC and turns it into AC using electronic switches that are pulsed on and off in some pattern.

That much I understood, it turns on and off and gives a square wave.

wrote ...
To operate an induction heater we need high frequency AC in the region of about 100-500kHz. So, we use a rectifier to convert 60Hz AC into DC, followed by an inverter to convert it back to AC at the required frequency.

Yep, I got that far. I don't quite understand why you need to set resonance (via selection of the output cap) to be high frequency, but, I don't need to. I know you want much higher frequency to jack up the skin effect for case hardening apps, but, I don't know that that'll be used too often on mine.

I don't for example, understand why 60hz (or whatever) isn't just fine.

wrote ...
The "driver" is the function block that tells the inverter what to do. It does this by generating a small copy of the desired output waveform, which gets amplified by the transistors in the inverter. (Bearing in mind that in modern power electronics the transistors are always used as saturated switches, not linear amplifiers. The driver must supply a square wave drive signal big enough and fast enough that the devices are flipped quickly between off and fully on, hardly spending any time in between.)

Ahhhh... thanks.

So I'd been picking gate drivers somewhat at random and asking various people "What about this one?", and get told it wasn't powerful enough and I didn't get it. Why would it need to be powerful, that's what what fets are for.

The part I was missing was that the MOSFETS have to be slammed full on or full off by the driver, so it has to be powerful enough for them not to hover in the ugly middleground.

wrote ...
For an induction heater the required signal is a square wave of 50% duty and variable frequency.

Right, because the output resonance is going to be a sine wave, and will vary with the load.

wrote ...
The Uzzors circuit tries to find the resonant frequency of the work coil circuit and lock to it, but if the work coil voltage or current become excessive, it detunes away from the resonant frequency to limit it.

Oh, even better. I didn't know it also did overcurrent protection. I'll have to look closer, I presume there's some trimpot or something in there to set maximum current.

wrote ...
This is a good general approach to the problem, though it is better suited to MOSFETs than IGBTs.

Perfect, I'm going with MOSFETS so, that works.

...

MOSFETS have to handle the full voltage that I'm feeding them, which is as high as 300V (i.e. 208v / 0.707 when smoothed). Plus, some extra for safety? 500V is a good target? And, only enough current for what it's drawing out of the wall (30A for a 30A breaker, 50A for a 50A breaker), plus some extra?

Do the MOSFETS amplify voltage or the current (both?) fed to them by the driver?

What are the constraints of my driver chip? It clearly can't be rated for the full current, because drivers are always physically small. Will it need to be full voltage (500+V) though? How much is "enough" current to slam the MOSFETS on and off on a machine my size?

Speed of turn on/off for the driver is also important I recall. I think my machine will oscillate around 20khz if I use the caps I was planning. Someone threw out 500nS as being plenty?

I look through Mouser or Digikey catalogs and I'm just throwing darts at a dartboard herpy derpy-like and not knowing how to interpret the results.

For example, I look up the UCC37322P driver

that Uzzors uses.. it says it's good for a max of 15V, and 9A. Except I know he's using 320VDC, and 9A seems kinda excessive for "just" a driver.

...

Hopefully I have fewer questions as I continue learning. Thank you very much for dumbing things down for me.

Steve Conner

Fri Aug 22 2014, 11:25AM

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

I don't actually know the exact reason why we use high frequencies and not 60Hz. But high frequencies work much better and are a must for heating non-ferrous metals.

The MOSFETs amplify both voltage and current.

The gate of a MOSFET takes 0V to turn off and about 12V to turn fully on. It looks like a small capacitor and draws no DC current at all.

However, the gate driver has to supply enough transient current to quickly charge and discharge the gate capacitances of all of the MOSFETs. This is how the peak current rating for it is calculated. 9A is not excessive when you want to quickly switch a bridge of 4 large MOSFETs.

All of the energy supplied to the gate capacitances ends up dissipated as heat in the gate resistors and gate driver chips. The UCCs can overheat when driving a large bridge continuously at >100kHz.

1 2 3 next

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