High Voltage Forum
High voltage => Transformer (Iron Core) => Topic started by: rustedone on October 16, 2019, 04:20:15 PM
-
hello everyone .
i want to make an induction forge.
there is the schematic that i find witch is rather nice. i wanted to ask if it needs few tweaks or not ... cause as i mentioned before ... for start i want it to be at 'usable' level. have to go step by step.
as you can see it's pretty much is a full H bridge inverter ... with some diodes ,a feedback and isolation transformer .
it says the main voltage is 220vdc but it's clearly made to work up to 400vdc .
-
Haha... I wonder if that would actually work?
Note that:
- Inverter is half bridge (IGBTs are in parallel pairs; load current is returned to supply through a capacitor divider).
- STTH200- is superfluous (IGBTs are co-pack with internal diode).
- 10R+2n2's may be superfluous (may help to dampen oscillation from stray inductance on each transistor, but they must be made very short, so as not to have more stray inductance themselves than the transistor -- ca. 10nH).
- 4R7+4n7 is superfluous (inverter is switching full wave; unless the opto acts to shut down the oscillator?).
- The 16V zeners aren't really going to save anything if it explodes. You're replacing all four transistors and the driver when that happens.
- Has to be a series resistor to the opto LED, otherwise it probably explodes the first half-cycle it gets turned on. A small say 100 ohms should be enough. Also a good point to add filtering (10n across the LED?) to deal with switching noise, which will definitely be present.
- No idea what the orange shunt thing is.
- +12V must be an isolated supply as it is mains-referenced.
- 2u2 450V (or larger or more numerous, if needed) need to handle full inverter output current, rate accordingly; EMI X2 types will not suffice.
- L1 needs to be sizable, and adjustable -- this is an LLC resonant network and L1 acts as an impedance matching element between the inverter and the tank. (Note that transformer leakage inductance adds to it; this sets the minimum Q and maximum load impedance (maximum work coil inductance at minimum tank capacitance) the system can deliver full power into.)
- You may want to put "trafo" in front of L1, not after it: as shown, the transformer needs to handle L1 + tank VA, which can be, say, 20-300% higher than inverter VA alone. (L1 needs to handle lots of VARs at high Q; this probably won't be a problem for forging, as steel keeps the Q low, unless your coils are poorly fitted.)
- Obviously the tank cap needs to be a big bastard. A C500T should be adequate for forging work.
- Needless to say, the transformer is nontrivial to construct. Not sure what I'd suggest; if you can source a bunch of fat-ass Litz cable, a toroid made with a big ferrite core and several layers of windings (use multiple layers in parallel for each winding) should be okay.
Tim
-
oh hello tim ;D.
it's good to see that you answer it.
wow i wonder how the guy who designed this made it work.
this is the link by the way : https://www.homemade-circuits.com/induction-heater-circuit-using-igbt/
between all stuff. l1 here is air core cable let say ...
but ... does this works then? from what you said it's need some shuffling .
there is this that confuses me . the effect of heating depends on power . but is high amperage more effective or high voltage ... ? ???
-
Counterpoint to a few previous comments:
The IGBT parts used in this design do not have internal diodes, so the external ones are critical. If updating this design to use IGBT parts with anti-parallel diodes, then the external ones could be removed.
The TIL111 opto is used for over-current shut-down per the original article. As long as the shutdown functions correctly, it will shut down before the opto LED current gets too high. (I'd still add a small series resistor just for margin.)
Good point about snubber R/C pairs needing low inductance. I'd hesitate to remove them, however, unless the rest of the layout is very clean. The high-frequency ringing can couple to low-voltage logic even if it doesn't fry the high-power devices. Such R/C snubbers are common in commercial switching power supplies to limit spikes and reduce EMI.
Yes, the 16V zeners won't survive a catastrophic failure. However, they can help with ESD hits during construction and testing, and with over/undershoot if the gate wiring is long or sloppy. I typically add such zeners to FETs and IGBTs at the start of construction just for insurance.
And a few other random comments:
Orange shunt is current sense for over-current shutdown.
The transformer is air-core per the original design. Not something I'd recommend, but workable for demonstrating basic induction. It's just a coil of two-conductor wire. Same with L1 - air core in the original article, just a coil of wire.
-
Ah I see, the -W is obsolete now and I found the -WD on search. Recommended, in any case -- the diodes are physically closer to the transistors than you can possibly get an external diode in circuit.
The shown air core transformer will have almost as much leakage inductance as L1, keep that in mind as it limits the tuning range.
Tim
-
there is this that confuses me . the effect of heating depends on power . but is high amperage more effective or high voltage ... ? ???
Yes. (inclusive-or ;D )
Power is simply what power you get into the coil.
The inverter only delivers maximum power at its rated current. Too much current and it explodes, too little and you don't get much power.
Between the two, you have an impedance matching network.
Impedance is the ratio of voltage to current, Z = V/I. Power is the product, P = V I. If our components are low loss, we won't lose much power in them, and the inverter power will [nearly] equal the work coil power. Power is conserved. The impedance can be adjusted up or down, however.
We can approximate the value of L1 required, by noting the inverter impedance:
Zinv = (Vsupply/2) / Iout
If we were driving L1 alone (not a very interesting application), it would simply require Z_L1 = Zinv, or
L1 = Zinv / (2 pi F)
Actually a little less because of square and triangle waves and RMS current ratings, but this is the right ballpark.
If we have a resonant tank after L1, then we need L1 to act as an impedance matching element between the inverter and the tank.
Let L2 = work coil and C = resonant tank capacitor.
The tank impedance is not quite:
Ztank = sqrt(L2 / C)
it is actually higher, because C is partly used by both L1 and L2! This is why the complete calculation is hard, and these equations are approximate.
The work coil has some Q, so it has some parallel equivalent resistance:
Rp = Ztank Q
We need L1 to match between this and Zinv. It turns out:
L1 ~= sqrt(Q Ztank Zinv) / (2 pi F)
Hmm, something like that I think? It's been a while. Run some test numbers and check against the impedance measured in a simulator.
Note that the tank capacitance is divided between L1 and L2 -- we can split the network into a series resonant L1 + C_L1, attached to a parallel resonant L2 || C_L2, where C = C_L1 + C_L2. We calculate what tank capacitance is required:
Leff = L1 L2 / (L1 + L2)
C = 1 / ((2 pi F)^2 Leff)
If you have a fixed capacitor, go back and adjust Ztank and F, until C comes out about correct. (F = 1 / (2 pi sqrt(L2 C)) is a good starting point, and again, because C is effectively split up, the real F will be higher than this.)
Work coil L and Q are not hard to estimate. For solenoid coils, L can be calculated with whatever formula. Don't forget to add the loop inductance to the tank cap; lead length matters!
Q might be 10-20 for the nut pictured in the article, up to 30-50+ for looser coupling (more distance between coil and work) or high-conductivity work (aluminum, copper?), or down to ~5 (where the above approximation starts to break down) for tight fitting, low-conductivity and magnetic work (steel, stainless, titanium?).
Also note that I've absorbed the series leakage of the transformer into L1; don't forget that the leakage sets the minimum value of L1. We also need to absorb the transformer's parallel magnetizing inductance into L2. These parameters can be calculated from geometry; offhand, leakage should be around 2.5uH, and magnetizing around 30uH.
Tim
-
Just some thoughts from me.
It's directly mains powered, that's very dangerous and isolation should be handled with care.
The designer has a non water cooled working coil, what's an absolute no go.
It's a rather big physical build and I wouldn't prefer it.
Power adjust is only possible with the working coil geometry and the matching inductance.
I personally don't like these capacitor banks made of hundreds of caps, get some real ones, which are at least air cooled.
The power of that thing is around 1.5 kW so it's nothing special.
You can adjust your power with the magnetic field, since the frequency changes are rather small.
The higher the magnetic field at the workpiece, the higher the possible wattage.
The bigger the current through the coil, the bigger the magnetic field.
The current depends on the frequency and the supply voltage, a voltage between 40-100V is what I'm using in my heater.
Since this build has an isolation transformer other values may be better.
You can change the coil geometry to adjust the power.
With more windings and less distance between workpiece and coil you get more power.
Changing the coil will change the frequency, so I'm not sure how preferable that would be for this circuit.
My final heater is in nearly ready to be show able, so if you aren't in a rush and have a week time, you can take a look at mine. It's around 4-8 kW at 30-130V and is electrical more simple, it just needs some good coolers.
-
hello every one .
I'm not that smart. so i try to draw points and tell me if i miss something witch i'm sure i will...
1- use transformer for main supply.make different output for it for variation .
2-change igbts to ones that got diode.
3-(does using 2 ir2153 make things better ? not sure)
4- work on LRC parts value whether l1 or l2(isolation transformer)
. but i doubt i can add cores to them. specially l2 .(wont the cores get heated up too ?)
5-what can i add that can protect parts from exploding ?
6- by the way if I want to make it stronger . what can i do beside increasing main power ?
7-use water cooling. i wanted from start .
-
1- use transformer for main supply.make different output for it for variation .
That would be super for isolation.
2-change igbts to ones that got diode.
I only use ones with internal diode.
3-(does using 2 ir2153 make things better ? not sure)
It reduces complexity, one "good" is easier to handle than two bad ones in parallel.
5-what can i add that can protect parts from exploding ?
Nothing really.
It explodes normally if one of the things happen.
- Not oscillating at start up (shouldn't be the case for that design)
- Voltage spikes causing breakdown of the IGBTs (using lower input voltage/ higher voltage rating)
- wrong IGBTs or Diodes
- Shorting the Coil!
- Overheating of components!
So some cheap temperature switches and house fuses can help to prevent extreme damage, but some parts will Always be destructed.
6- by the way if I want to make it stronger . what can i do beside increasing main power ?
You will most likely be limited by the transformer and the non perfect tune of the LLC Circuit.
7-use water cooling. i wanted from start .
Building water cooler isn't that hard and will help with overheating. I started by soldering simple copper tube to copper plates and that worked fine.
-
one friend below advised to use magnetic field on work piece . but what kind of magnetic field ? with a permanent ones or electric ones that changes ? this one is new to me . not that i know much anyway...
-
I could probably enumerate a hundred things that can guarantee something between failure and explosion. This is NOT a circuit which can be lashed up and expected to work -- even the positioning of components and PCB traces is critical.
If you need something explosion-resistant, hell... I wonder if you could use an automotive alternator to generate AC. Still need matching and tuning (transformer and capacitors), but it won't explode when it's tuned wrong. The frequency is terribly low (100s Hz), but it'll get steel up to the Curie temperature I guess.
Vacuum tubes too, but they're rather expensive in the kW range, unless you just happen to stumble on a scrapped (but repairable) radio transmitter or actual induction heater. Also terribly lethal if you touch the plate circuit...
Tim
-
um... ok then just let it explode ... ::).
but that is concerning . what is chance of exploding because of layout ?
would it be better if I use cables for main and high power part of circuit ?
I have a very important question : what software should i use for simulation ?
-
ok. how about this one... honestly i don't understand it much.
but it does looks different enough to be mentioned .
does this works better or something ?
-
That one's better -- it has phase control which makes it easier to set power. It's lacking current limiting or power control, which means it's still prone to blowing up if you set it wrong. (It has all the hooks to do these things, the controls just need to be added.)
The extra diodes in the inverter aren't needed, no idea why he did that. Reducing reverse recovery I suppose, but you shouldn't need to operate in a condition where that's a problem.
The control is more complex, and it sounds like you're a green newbie at this, so that may not be very helpful to you.
The power circuitry still needs to be built correctly, and has the same hundred pitfalls.
The first and biggest problem you will find, is that wire has inductance, and when you switch inductance, it makes voltage spikes, and voltage spikes kill transistors.
Consider a coil: it has some inductance, and it uses some length of wire. Now unwind the coil, so you have a big loop of wire instead. Has the inductance gone away, just because it's not a coil anymore? No, certainly not, it's still something. It might be lower, it might be higher, it depends. Now cut the wire down so it's just a tiny link, has the inductance gone? No, certainly not, it's a tiny fraction of what it used to be but it's not zero.
You need inductances around the transistors to be less than, say, 100nH, 50nH, even less depending. This corresponds to lead lengths of a few inches. No cables lazily flapping around. That blows up transistors.
Wiring needs to be big enough to handle the current. Transistors need to be heatsinked otherwise they overheat and blow up. But also insulated from the heatsink.
Simulators -- LTspice is free. A bit quirky but what are you going to do.
Tim
-
thanks TIM.
yep. I'm green as levl1 goblin ... ;D .
so... the phase control is good but not helpful for me.
for control we need something to adjust duty cycle and frequency maybe .
I can try to use micro controller on it . the frequency of micro's pwm can't be changed in middle of program if i'm not wrong. but maybe it's possible to set it to external crystal and put several crystal with different frequencies that micro can switch to them with (relay) I think it was it's name.
so if i can add control to first schematic it would be better i believe . like using ir2102 and control it with micro ...
it can receive feed back and adjust .
-
this is a sketch that i draw with altium designer . didn't bother with parts values yet but more on overall picture.
I send two feedback to micro with 5v. one from opto coupler. another from voltage divider. and depending on changes from them. micro tries to control stuff. so what do you think?
-
You'll be better off with the IR2153, honestly. It oscillates so you aren't in danger of one or the other transistor getting stuck 'on' all the time (obviously a problem if the load goes short, or say if the MCU crashes?), and it has a disable so you can stop it, if nothing else. You can't do PWM with it, but you don't really want to, anyway (PWM doesn't really control power . You could do burst PWM where it's oscillating for some cycles, then disabled, and so forth.
Making it controlled frequency would be hard, especially if it's always going on and off...
Could also simply make an oscillator, like what's used in a CFL. Adding the control and protection is tricky. There are ways.
Tim
-
thanks TIM ;D. so i just stick with ir2153. what about other parts location like l1 . i putted it before transformer . is it alright in there ?
-
there is something i'm curies about. we basically need to send high power high frequency to work coil.
then can we use tesla coil to produce high vlotage-freaquency without switches so that nothing sensitive is in circuit. then send it to work coil .
I think with a little changes it can work. of course it wont be efficient but it wont get damaged easily either.
like it work with brute force. if there is good cooling system then it can handle a lot of power in form of high voltage .
-
Are you talking about using a spark gap style tesla coil circuit to drive the work coil? Thats going to require some expensive high voltage capacitors and heat a lot less effectively than a normal induction heater.
People often have two goals in making an induction heater. One is to have a functioning induction heater that they need for some other purpose. The second is to gain experience with power electronics (which is a great learning experience but often results in a more expensive and less functional induction heater).
If you primarily want a cheap and functional induction heater I would suggest you start with one of the induction heater kits available on ebay (here is a random example https://www.ebay.com/itm/1000W-ZVS-Low-Voltage-Induction-Heating-Board-Module-Flyback-Driver-Heater-DIY/122007805772 ) . They are cheaper than what it would cost to make an induction heater from scratch and have been extensively documented on the forum https://highvoltageforum.net/index.php?topic=530.0
Experimenting with the kit would also allow you to work up to designing and building your own custom induction heater. Blowing up power electronics stuff gets expensive fast, so starting with a kit like that is a good idea, regardless of what your end goal is.
-
you can make the caps your self. with aluminum sheets. the problem is. whatever induction heater circuit i find. people find tons of problem with it. so i thought . why not? and i think if you look for it you can find some solution for effectiveness
-
and of course maybe if you add one or two more isolation transformer in between to change the voltage and amperage to desired level . it can make things better.
-
I am assuming you found these two webpages?
http://www.loneoceans.com/lo_main/labs_01/inductionheating/index.htm
http://www.powerlabs.org/indheating.htm
Using a spark gap induction heater is going to be a lot more dangerous as you need to insulate the coil to a much higher voltage. Additionally, it's going to be a less efficient and have limited output power. High power high voltage transformers are relatively uncommon and high power spark gaps are a bit tricky to get right and can be expensive.
I would strongly suggest you start with the induction heater kit on ebay. They work well and people on the forum know a lot about them, there is already a 10 page thread about the kit. At $30 it is also going to be a lot cheaper than any other option. Also, unlike the spark gap approach, they use a lower voltage, so it's a significantly reduced electrocution hazard.
-
eehh?! there is page like that ?! I just thought it myself ... but well. you see man. i need strong one . and need to work with it. it may look dangerous. but if you pick the parts right. then it simply becomes more dangerous to you and less to device. then you can isolated whole thing and control it from safe distance .
but it's not like i just want this. if you have idea for high power induction forge. lets hear it brother . :)
-
so in the end i should use the first circuit with some improvements cause it's hard for tesla coil to lock on resonance frequency. then what about capacitor bank value . i doubt it's the bigger the better . or is it ?