I am finishing up a homebrewed piece of crap furnace
I am using 20 2sc2625 transisters configured as a half bridge pushing a maximum of 100 amps @ 170 volts into the coil.
the 240vac is rectified into 300-340 volts dc across 400 volt capasitors (200volt capasitors in series) as to provide the connection for the coil.
modern heaters use a transformer and ac capasitors as to allow the use of SCR's to drive the circuit.
and most use currents of 600 amps or higher as to use a coil of 2-4 turns
I am using NO capasitor in series with the coil. so It is nesisary for the coil to have an inductance of at least 40 microhenerys to make a useable
duty cycle. because those tranistors have a turn on and turn off of 1 micro second, the max usable frequency is around 17khz to avoid high losses.
and the current cannot exeed 100 amps, so for my application,
a coil of 20 turns, 25 cm long and 15cm diameter is perfectly
useable.
I would be cooling the coil with a non electricly conductive liquid. so i could integrate a radiator, as the tubing ends at the tranistors.
contrasting a profesional heater where WATER circulates through the transformer/coil. that being a continuous tube.12AX7 - 13-7-2006 at 17:26
Nice to hear someone else is taking on the challenge. Pics/schematics/oscillographs?
FYI, modern heaters use IGBTs as much, if not more than, SCRs. Easier to handle, more efficient, a whole lot faster (wider bandwidth so especially in
the HF units). Them, and FETs, are much easier to use than BJTs or SCRs, too!
You're going to waste a *lot* of VAs skipping the capacitor. Figure a power factor of 0.2 or worse. That means, out of your 170 *
100 = 17000 = 17kVA capacity, you're reduced to, at most, 3.4kW. You could be getting more than 8kW out. Also, note that the horrible power factor
puts undue stress on your power supply capacitors, since the 90% inductive current that you fed in during one half cycle is dumped out in the next
half cycle (and restored back and forth).
The L-LC network I use seems to be pretty reasonable, and matches a high-amp coil to a reasonably dimensioned half bridge well. It's also completely
short-circuit-proof.
You'll probably be burning a lot of power (lesse, guess 2V saturation at 10A = 20W per transistor, that's um 400W... if that... count on noticably
less than the 98% efficiency I expect!). I suggest running the cooling water over the heatsink. With the dissipation requirement, you'll need those
transistors as cool as you can keep them.
Oh, and yes, water... don't give a second thought to it. Distilled water is more than resistive enough that you'll cause maybe 0.001% of loss to the
output power. It's AC, and fast for that matter, so electrolysis isn't a concern. Use a couple feet (<1 meter) of plastic hose where you need to
connect between hot or high and grounded pipes.
at 8-9 amps the Vce is 1 volt
and Vbe is about 1.7 volts at one amp drive
the reason for not using water also had to do with skipping the LC network-maybe not a good idea
I was expecting 200 watts lost in the transistors, with 100 watts lost in the emitor resistors- they are necessary because the Hfe is temperature
dependant. running the temperature high seems to be a good idea to me.
also if I could find a large enough ferrite core, I could use that to match the coil.
I have no shortage of 200 volt caps, 330, 470 and 680 uF
like 140 of them,
I also have around 40-70 of those x-rated .22 through .68 uF, 250 Vac mylar capacitors used for filtering out the EMI in computer power supplies;
btw a friend of mine said that due to the failure rate expected of them, they can be expected to withstand around twice their rated voltage, at an
amperage limited to what the 22 guage copper wire can stand
I don't know how effective they would be in this application though, due to the need to use about 60 of them in parallel.
[Edited on 15-7-2006 by tumadre]12AX7 - 15-7-2006 at 07:15
My tank contains 200 x 0.1uF 630VDC (250VAC) polypropylene "MKP" type capacitors. The little blue boxes. Ya, same sort of stuff you find across the
line (also mentioned in the data sheet).
I wouldn't trust the larger values, because although they may handle more current than a smaller one, you get strength in numbers. Try to find as
many 0.1 and 0.22uF caps as you can, rated for as high a voltage as you have.
If you're going to use a normally sized coil, your concern is current, not voltage. According to the data sheet, my capacitors can withstand 2.3
amperes RMS per capacitor, for a total of 460A.
Looking at the 2SC2625 datasheet, it looks like typical data is better than I remember. Suprisingly, hFE doesn't change much between 25-125C,
according to this datasheet. Remember to keep a few microseconds, maybe 5, inbetween switching events.
Timtumadre - 16-7-2006 at 22:51
I also discovered in my collection.
an h-bridge consisting of 40 rfp50no6 mosfets
using a 50 volt @ 400 amp power supply !good luck finding that
it could permit a 10 KW or even 15 KW portable unit consisting of 4 deep cycle lead acid batteries.
and the mosfets will power the coil at 100Khz no problem.
@400 amps running through 2 legs of the bridge, 704 watts lost to internal resistance
so at 400 amps peak current, maximum heating would be around 500-600 watts with additional energy lost in the forward voltage of the reverse current
diodes internal to the mosfets.
This would require a cap rated at 600 amps
and a lot of copper
The unit came out of a 1700 watt APC uninteruptable computer power supply.
I mounted the tramsformer in a 400 watt APC UPS case and it wil power 100 amps at 18 volts continuous.
no over-heating with a good fan. I would sell it but it does weigh 30-40 pounds12AX7 - 17-7-2006 at 14:06
Ooh, wonderful! :drool:
*Searches*
Nurrr that's RFP50N*zero*6, not "o"6.
But, honestly, good luck switching 400 amps. The thought occurred to me to research a series-resonant tank, that is, only the work coil and
capacitor, no Lmatch. But that means the work current flows through the bridge, which means constantly switching yea many amps a couple ten thousand
times a second.
I even posited the question to a forum where professional EE's reside and they basically laughed at it. Some chat, but the basic consensus is, it's
unfeasable.
It takes careful design to switch 500 amps in even 5 microseconds. Consider an inch of wire in free space contributes about 20nH (IIRC). Since V = L
* dI/dt, the voltage developed in switching 500A in 1us is 10V, about your gate "ON" voltage, or a third of your power supply. You can easily
overvoltage the transistor on switch-off, if the inductance to the nearby diode is too high.
And yes...you need lots of amps in your power supply capacitors.
Tim
P.S. I smell MIG welder, for that chunky transformer!
[Edited on 7-17-2006 by 12AX7]tumadre - 17-7-2006 at 21:41
didn't think about that.
back to the 2SC2625 design.today - 4-9-2006 at 21:30
One of my plans for a job is to build a large scale induction setup. I've done a test of 1mm nicro chrome wire and one battery can heat to 800C 60mm
of wire.
I would like to know can i add another battery in series or will it melt the wire, what about parerell.YT2095 - 5-9-2006 at 05:07
NiChrome wire across a DC source isn`t induction heating.
para you`ll just get the same effect but it`ll last longer, in series and you`ll likely burn the wire out.Quince - 1-10-2006 at 01:20
I'm interested in this, after I finish some audio projects. I've only really looked at the schematics on 12AX7's website, and I'm wondering if the
complexity of some of them is really necessary, or it's more like squeezing out the last 15% of performance.12AX7 - 1-10-2006 at 06:32
Compared to what? You're welcome to try a chunky bipolar free-running oscillator, and get good results, but class C oscillators are typically in the
60-80% range, and I didn't find them very reliable (at Vebo < 6V for most transistors, it's a fine line between adequate excitation and blown
transistors). It's a lot easier to switch 10kW with $40 of IGBTs or $60 of MOSFETs at 98% efficiency than who knows how much for BJTs (although 10 or
20 x MJE13009 seems to be a good deal at Digikey). Or you can pay upwards of $500 for a 60-70% efficient vacuum tube system (tube, socket, cooling,
transformers, etc.), which so long as you keep it cooled to spec, I can guarantee will let you beat it over the head with a brickbat, electrically
speaking.
TimQuince - 1-10-2006 at 17:05
Well, I'm aiming for no more than 3-4 kW.The_Davster - 1-10-2006 at 17:39
Hows about compared to a powerlabs type setup? With just a big cap, NST power supply and a sparkgap?
I have a 10000V 0.1uf capacitor which might be suited for one of these?12AX7 - 1-10-2006 at 19:09
I have some 4kJ worth of caps in a box somewhere, but that doesn't mean I can shrink dimes with them, they're electrolytic and would probably explode
if I tried it!
Now if it's a Maxwell labs or similar sort of construction, that's different, and very interesting indeed.
But as for spark excitation, it sucks. For one thing, at the ballpark efficiency Sam himself mentions, you'd be sucking down most of your house's (or
shop's) electrical feed just to get those precious few kilowatts. You still need to match load to source, which for high voltage means many turns (of
wire that can't be cooled with water) for the work coil, or a transformer (not very practical, and probably needs to be cooled too!).
Induction heating is most practical in the range of 2-10 turns of copper tubing, which is suited by roughly equal amounts of volts and amperes. Solid
state around 100-600V fits in pretty nicely.
Besides, a spark isn't elegant. It might've been in 1892, but not in 1950 when they had 833s and 4X1000As. Now that's elegant. Even more elegant,
but less visually appealing and more apparently complex, is solid state. The efficiency of SS is simply awesome, and what can I say, I'm an
efficiency whore when it comes to not buying heatsinks.
TimThe_Davster - 1-10-2006 at 20:41
Quote:
Originally posted by 12AX7
Now if it's a Maxwell labs or similar sort of construction, that's different, and very interesting indeed.
As in a big oil filled monstrosity that weighs as much as my 2.5L bottle of sulfuric?
I emailed the company when I got(or rather 'dived') it....no PCBs....yay.
EDIT: This reminds me...I gotta get rid of my couple dozen or so PCB capacitors.(600V, 3-5uf)
[Edited on 2-10-2006 by rogue chemist]Quince - 2-10-2006 at 00:49
Caps are cheap. Last year I bought four slightly used 3 kV film in oil, 40 uF each, for $40. That's 720 Joules for a decent price with virtually no
searching effort (i.e. went to eBay). With some effort around surplus places etc., probably could get 1000 Joules for $10.12AX7 - 2-10-2006 at 06:36
Ya, I got a board of electrolytics (totalling those 4kJ I mentioned) from eBay something surplus for a good deal, think it was like $20.
Timfuse123 - 27-10-2006 at 12:26
what the maximum temp?12AX7 - 27-10-2006 at 16:17
Unlimited, in theory.cbfull - 29-11-2006 at 17:53
I have a bit of experience with this type of circuit, mainly the self-tuning push-pull, half-bridge and full-bridge. Switching the large amounts of
power involved is not as scary as it sounds. The key is to get the transistors to switch when the sinusoidal current passes the zero point. This is
called zero current switching, or ZCS.
Another type of switching is zero-voltage, or ZVS. These conditions are key to building a circuit that does not turn transistors into firecrackers or
heaters. A self-tuning circuit will get pretty close to switching at zero all by itself, but I don't have an oscilloscope to see exactly what's going
on, so please excuse me if I am wrong.Quince - 29-11-2006 at 19:16
You can buy a decent scope for $100 on eBay.Contrabasso - 6-4-2008 at 03:41
Can I actually and legally buy a small (1kW max) induction furnace? ebay? mail order? UK supplier?microcosmicus - 6-4-2008 at 08:06
Legality isn't an issue unless you were planning to steal a furnace.
Rather the problem is in finding a furnace of that size ---- AFAIK,
the overwhelming majority of induction furnaces are much bigger
affairs used in places such as foundries or scrap metal recyclers.
You might have better luck finding yourself an induction heater
such as is used for heat treatment of small parts and similar small
jobs, then building a small furnace from copper pipe and refractory
material which could be connected to the power supply of the heater.
For instance, here is an advertisement for an induction heater in
the power range you are interested in:
If you look around, you might find one for a lower price or locate
a used unit for sale.
[Edited on 6-4-2008 by microcosmicus]Twospoons - 6-4-2008 at 14:09
Better still, get a portable bench-top induction cooker (~$100 used), gut it, and rebuild into a furnace. That should be good for 1-2 kW.evil_lurker - 6-4-2008 at 14:20
Quote:
Originally posted by Twospoons
Better still, get a portable bench-top induction cooker (~$100 used), gut it, and rebuild into a furnace. That should be good for 1-2 kW.
That is what I was thinking.
You could theoretically cast a small refractory furnace, put a steel or iron plate in the bottom, and then use it to melt aluminum or other low m.p.
metals.12AX7 - 6-4-2008 at 19:22
Sadly, induction cooktops aren't the right impedance or frequency to operate the kind of coils needed for a furnace or other general work.
Timmicrocosmicus - 6-4-2008 at 21:08
More specifically, since they work via the magnetization of iron,
they won't do diddlysquat above the Curie point of iron. Hence, if
you did as you proposed with the iron plate, you would find out
that your furnace would not go above 770C.
While such a furnace would not be good enough for tasks like
welting steel or brass, it should work for melting low melting
point metals, calcining, dehydrating, pyrolysis. annealing glass,
firing earthenware and other such applications around the lab.
For higher temperatures, you need to make use of the skin effect,
which means radio frequency.
Induction Heating
12AX7 - 24-6-2009 at 14:55
Induction heating has come up here on occasion. I thought some might be interested in the recent progress I've had.
This is the 1kW unit:
It works, oh yes it does: this is a tablet of graphite heated to about 1100°C. (It's actually glowing yellow, the camera saturates to white. The
glow seen around edges is more representative.)
Now, the interesting part is this: I'm getting ready to offer kits and finished units for sale. First, the 1kW model seen above. Models up to 10kW
later. Keep in mind this is an advanced project and requires more than a VOM to put together. If anyone is interested, drop me a line.
Fantastic Job, I was just about to link your story from hackaday too. Im really impressed and would love to build one (cash permitting). I would
consider adding digital controls and preset ability with a PIC or Ardunio..USB?..
12AX7 - 24-6-2009 at 23:09
I'd go with something Atmel (rather than use an Arduino, I'd cook my own hardware), PICs are stupid.
Schematics are new; unfortunately I may refrain from posting them. At this point I'd rather sell plans than give them away. With plans, of course,
you get much more than just schematics. The important part you can figure out from the picture, actually: it's series resonant.
TimUser - 24-6-2009 at 23:29
I'm really impressed
Can you tell me if in essence this system differs a lot from the one in a kitchen.12AX7 - 25-6-2009 at 02:45
How much essence are we talking?
The control circuit is different (e.g., quasi resonant), the inverter is different (e.g., SCR or single IGBT), and the output is different (high
enough impedance to drive directly, plus ferrite pole pieces). It's still the same principle of operation, of course.
The impedance is probably the most important difference. A low impedance can drive a work coil with just a few turns, necessary for the water cooled
tubing high-power heaters need to use. As such, you can heat up steel pipe or frying pans pretty well with a conventional unit, but a stovetop unit
will only work with one work coil driving frying pans. So if you were wondering why a stovetop unit isn't good enough, this is why.
The main question I had, then, was whether you are still rolling your own gate drive
circuitry or are using one of the gate drive chips like the IR2113 or TD350?hinz - 25-6-2009 at 08:56
Looks cool.
Is the big air gap between torroid and induction coil as seen in the picture really necessary?
I always thought that big air gaps decrease the conversion efficiency of transformers. Or does your IGBT burn if there's no more self induction?
Nice big torroid you've got. Could probably build also a badass SMPS for electrolysis with it.
Probably induction heating could be also useful for some high temperature reactions. No need for fancy heating wire, just a well isolated steel vessel
placed inside some coils.
Please upload some plans, especially of the 10kW unit, keep it open source dann2 - 25-6-2009 at 09:29
Hello,
WOW!
Wondering if it would melt Magnetite. Don't see why not, 1550°C.
Will have to build one of these. They would be a very useful tool in any workshop.
I used to work in a place where they put foil seals (under the cap) on plastic bottles of stuff. It was done with an induction heater. The foil, stuck
to a piece of paper like stuff, comes in the cap . When the cap is tightened the foil is squeezed against the plactic top of the container/bottle. It
then goes under the induction heating head where the foil heats up and melts the plastic of the container top and stick/seals to it.
There were rumours that some guy got a bad burn when he put his hand, with a wedding ring on it, too close the the coil.
Dann2 12AX7 - 25-6-2009 at 13:58
The air gap isn't -- since nearly all the flux is enclosed in the toroid, it just matters that the secondary loops through it a few times. That's why
I need a high permeability, ungapped, symmetrical core, exactly what I happen to have.
As a matter of fact, if you put a rectifier and filter on the output of this (instead of the tank), you'd get somewhere around 5.6VDC at up to 120A.
Hope you have big heatsinks on those diodes.
TimTwospoons - 25-6-2009 at 15:00
Very nice! Mind if I ask what frequency it operates on? What sort of caps did you use in the resonant tank - they don't look like Celems.12AX7 - 25-6-2009 at 20:03
Yes, the bank is 100 x 0.1uF 275VAC MKPs -- unusually small though. They get hot at 3A each! Unfortunately Celem doesn't appear to make a 50kVA
unit, which would be more appropriate here. Their 200kVA product is appropriate for the 10kW model.
Currently, the maximum I can operate at is 300A at 40kHz or so. That puts the work coil around 0.25 ohms, or 75V across it.
TimVegemeister - 4-7-2009 at 13:30
What about putting the capacitive part of the tank on the other side of the transformer? The capacitors wouldn't have to pass as much current, and
the impedance transformation would facilitate a lower operating frequency with the same amount of capacitance.
Edit: whoops, realized part of the idea was free lunch.
[Edited on 7-4-2009 by Vegemeister]
[Edited on 7-4-2009 by Vegemeister]12AX7 - 4-7-2009 at 15:35
Same VARs, just switching amps for volts. Plus now the transformer has to pass all the VARs, making it 10-20 times larger. No thanks.
)
that's RFP50N*zero*6, not "o"6.
