Electronics > Beginners
Induction heaters - the theory
<< < (4/5) > >>
T3sl4co1l:

--- Quote from: coppercone2 on January 15, 2019, 05:14:45 pm ---You need pump control, flow measurement (or at least indication) and resistance measurement to make sure its not getting soiled
--- End quote ---

In the industrial design, we used a flow meter, I believe the free turbine type with Hall effect sensor.  No fouling problem (not beyond what anything else will get, anyway), very reliable, fail safe (no pulses = must be no flow, shut it down).

A pressure switch is often used, too, since a pressure drop is expected across the plumbing.  We didn't do that (I think the pressure drop of the internal piping wasn't very high -- it had pretty good flow).  Obviously that's not telling you if pipes get blocked, so it can only be used with a flow sensor of some sort.

In a product, I'd be happy with a flow switch or meter.

In my project, I didn't get that far, but the trickling sound of water falling into the milk jug reservoir is pretty obvious, and failing that, the rapid bumping of water boiling in the work coil is a further reminder.  No, the semiconductors don't heat up nearly as fast as the coil -- this is fine for testing purposes.



--- Quote ---You need thermostats on the heat exchangers in the very least
--- End quote ---

You can.  A thermistor measuring outlet temp is not a bad idea (which we did on the industrial design).  Thermostat, no.  If the flow is >= enough, you don't care.


--- Quote ---Won't a SMPSU conduct more HF noise then a transformer given the types of magnetics that it uses? My plan was a regular transformer between a Rcore and a torroid and maybe some filtering after.

--- End quote ---

And the inverter isn't completely fucking beating the shit out of the SMPSU..?

We used RECOM Medical type (8kV reinforced, low capacitance) DC-DC's, which handled 650V, 50ns edges just fine.  They made no detectable noise.

What did create detectable, in fact annoying amounts of noise, was the 100V 10A Chinese shitbox PS we got for testing.  Its output had 10s of volts of common mode noise (spikes).  The 100V inverter waveform was noticeably fuzzy.  I had installed extra filtering inside the bastard to help.  Later, it cooked itself off; seems that, when operated at full current into a <10V load, it's a class A amplifier, rather than class D... its IGBTs didn't take kindly to that.  Which were overrated and under-driven anyway, in typical Chinese fashion.


--- Quote ---I thought to use a optical sensor in a feedback loop rather then calculating power and to use some kind of PID control with a low maximum ramp

--- End quote ---

You can do that, especially now with thermal sensors/arrays/cameras being common (they were still very boutique and expensive circa 2010 when we looked into this).

Thermal is slow, though, and you're likely to overshoot the setpoint.  Typical industrial application is a square pulse of power.  Open loop, just a programmed sequence.  Lock parts in jig, hit the go button, apply e.g. 8.2kW for 1.23s, off, let cool.

The tech designing the profile, adjusts P and t until the part has the desired heat treat pattern, or filler is distributed, or whatever.  Dwell or simmering is rarely needed.  The characteristics of steel mean that it's not much point to adjust power during a case hardening step (the heating zone follows the region below Curie temp, which is coincidentally very close to the austenizing temperature; the region above Curie doesn't overheat very much, because of the huge shift in properties).  And that covers 95% of applications, so there's very little need for more control.


--- Quote ---I thought to use bus bar with a bore hole in it for water cooling
--- End quote ---

PPECO does this, apparently.

Gun drilling is a huge waste of machining time.  I guess it's cheap enough in Taiwan.

More than adequate is brazing a pipe onto the bar.  Until you get to very high frequencies and powers (100s kW, >200kHz?), you don't have to worry very much about the combined thermal and electrical conductivity of copper.

Stacked and soldered/brazed construction is also excellent.  Or an embossed shape against a plate.  Good way to make big pockets connected to many fine channels.  Unless you find one off-the-shelf, it's a lot of custom design and building though.


--- Quote ---I did think though that the control board should have isolation on all the feedback signals. Would you just let the secondary sensors into the processor PCB directly?
--- End quote ---

Maybe.  The coil can be grounded, one side or the other, or CT grounded or whatever.  It's usually kept isolated, so you don't get screwdriver-melting fault currents when something accidentally touches it.  But that doesn't preclude a modest impedance (say a couple 100k resistors in a voltage divider and differential sense amp?) from being useful.

High frequencies don't feel the same as mains frequency.  Better to just insulate or guard the coil in the first place.  You can get electrocuted (cooked) by that kind of power, without feeling anything at all.  Again, not something a few 100kohms will aggravate, but a big risk if it were grounded, or if someone grabs both terminals at once.

In any case, everything is transformer isolated, so it's also no problem to transform it right back.  A hundred turns on a pot core makes a damn high impedance (effectively no load on the work coil circuit), and say a single turn secondary makes a quite acceptable signal level for the control circuit to work with (say 100:1 ratio, so 500V on the tank --> 5V at the circuit).

The direct-connected current limiter/clamp idea I mentioned, could be sent into a digital isolator, yes.  That would be a fine way to get phase information out (of course, the limiters destroy the amplitude information).


--- Quote ---I like the idea because then you can use digital isolators for all your signals. And then you can isolate all the subsections with their own transformers since its already a giant box that weighs 200kg commercially. Who gives a shit at that point.
--- End quote ---

Digital isolators are good for digital interfaces.  For example, you might have a core control board, that's mains referenced (to the inverters specifically, and maybe those use bootstrap gate drivers, hence the mains reference?), which does all the analog control itself.  It has an isolated digital input, and DACs onboard, to communicate enables, status, faults and setpoints.  The digital control board then is ground referenced, and has user input controls (pots or encoders on the front panel, say), or external inputs wired from a PLC, or comm ports, whatever's good.


--- Quote ---For how everything effects the scaling parameters for the frequency and power (phase shift ratio) it seems like a spider jumping on a trampoline. May be complicated to make all the summers/scalers. Then just make a temporary panel with a bunch of pots on it to adjust the circuit to find the values that kinda make it run and then replace those with digital signals from a MCU once you honed in on it.
--- End quote ---

Well, no, those are just design parameters.  Min/max V range, I range, F range, and control loop time constants (and zeroes if applicable).  Or PID parameters, as the case may be.  The loop parameters are the hardest to know from basics, but the others are just AC steady state solutions.


--- Quote ---Maybe turning it all into PWM would make it too drifty for case hardening turbines but I think it would be fine for a vacuum smelter.

--- End quote ---

One thing we played with was a direct drive inverter, 1MVA capacity at up to 50kHz, expected output ~150kW (depends on load, obviously).  The point was to direct drive so the square wave PWM% and frequency are adjustable, controlling the heating depth with the programmed cycle (which was linked to a CNC axis which moved the work through the coil).

The cross section from the process is pretty cool.  It looks like this,
https://www.industrialheating.com/ext/resources/Issues/Issues2/2018/Dec/ih1218-opportunities-lead-900.jpg
but imagine the depth can vary by section.  So, maybe those inside corners could be made deeper for better strength, or something.

Tim
coppercone2:
I don't understand what you mean by the hammering the shit out of the power supplies?

Maybe I am not being clear, I just meant that all the isolated control electronics would be powered by small signal transformers with extra filtering to power rectifiers to power linear regulators to power the electronics. I thought it would be more robust then using switching converters.

My first mockup of this actually used RECOM brand dc/dc converters (for the gate driver and MCU circuits), which were cheap on ebay as NOS, but I thought if one of them breaks it could be a hassle since they were not cheap on digikey. I thought if I built a serious shop tool it would be unobtanium to obtain the same footprint DCDC converter in 15 years. Then it would turn into a horrific bodge to fix it. I did not recognize them as a standard footprint. But I think the best recoms I found on ebay still needed a 48V stepdown transformer or stepdown converter to work. I just see it going south in the deep future if I make a nice fit chassis for it with all the ebay deals.

Of course I want to also power the power circuit through a big multiphase isolation transformer (but not used for isolation) but I also expect to run into design difficulties here, with inrush and humming. I don't know enough about those power levels to say anything ATM. But for the initial testing I would like to get series high power regulated lab supplies to run it on before letting it go on mains.

I thought they might do this given that it weighs 200kg.

And I am pretty sure flow switches can get stuck open. If you don't have a second backup then stick to the flow paddle wheel or whatever, but keep in mind if it leaks you still get flow (particularly if you test it from a faucet or plant water), hence the desire for thermostats. If its only going to run with a closed loop tank then I guess it will drain fast enough before it explodes unless you decide to use some giant drum. Unless you put the flow sensor on the outlet then it will not be stopped by leaks. But the thermostat is going to break a circuit and not rely on some kind of interrogator that can malfunction. And of course you might hook it up backwards which will fool a single flow sensor in a leak condition if it does not have some kind of direction control. I also don't know what steam can do to a mechanical flow sensor. I suspect it would be likely to but not necessarily give a nonsense reading that would flag a fault.

I suspect it would be advantageous to the room to hook up multi kW waste heat to a radiator situated outside of the building and not include a pump inside of the unit. It would be unpleasant to work around the pump noise and the heat. Even an exhaust fan into a duct would be kinda loud and irritating, otherwise your air conditioner and AC bill would suffer. And no filters to clean for a interior vent with high air flow, no annoying duct work to run, no clogs, you just need to hose down some shit outside once in a while and give it a scrub with a brush.


A very interesting IH design I saw on youtube is that the guy totally sealed his chassis (to eliminate shop dust) and put a heat exchanger inside of the unit with a recirculating fan to maintain low chassis temperatures and cleanliness.
coppercone2:
I am kinda thinking about some kind of high capacity outdoors located pump/tank/radiator combo that would do some kW of cooling at a decent flow to cool various power electronics things for a power electronics lab (i.e. heat loads). It would have 2x flow meter and 2x pressure meter and differential temperature in addition to a pump controller and a tank. I wonder if you can mount it to a overhang or something. Something that does not look like a industrial site. Maybe built to look like a wall extension.

If you had that capability you can add cooling jackets to vacuum pumps and other things too. And put little heat exchangers behind equipment that generates lots of heat to keep the room cool. Like a rack of commercial equipment with a suction fan. Then you can run stock linear high power supplies.

It would beat the shit out of a extraction vent (omfg don't you hate that corrugated tubing shit?) if you need to move something around, especially if you put a grid of metal plates on the ceiling to which you can attach magnetic hooks if necessary to suspend heat extraction piping.

I guess you can cool it to with a comperssor if you want. 
coppercone2:
identified a use for large scale home iron casting (other then precision machine shop work and making your own fence):

capability to quickly make a custom weight for making grilled cheese sandwiches. Low tolerances and delicious result. Imagine stackable cast iron weights of various sizes that mate with each other so you can find the right amount to put the correct amount of pressure on bread. 

You no longer have to be at the mercy of either bread selection, using your hands to press odd sandwiches etc (imagine like a hoggie or large cuban on weird shaped bread).

Should do nicely with my planned experiments in bread manufacture.
drogus:
Thanks for the discussion here, it's a lot of interesting info.

That said, in the topic I was asking for thorough theory sources, ideally books, because I think that reading a good book on the topic is usually really valuable for me. I've been digging more in the topic and I found something that looks very promising: "Elements of Induction Heating. Design, Control and Application". It's on google books and first 37 pages are available to read and it looks very promising: https://books.google.com/books?id=zXmTLYwO3McC&pg=PA97&hl=en&source=gbs_selected_pages&redir_esc=y#v=onepage&q&f=false The book doesn't seem to cover a lot of the electronics involved, ie. it rather covers high level design considerations (parameters tuning, coil design etc), but it still seems like a very good knowledge source.

This also seems like an interesting book: History of Induction Heating and Melting. Again, it doesn't really go into electronics theory too much, but it seems that there's a bit of info about design.

Next book that I found is Handbook of Inducion Heating. Most of the book is about induction heating application, but there's also a chapter on power supplies for induction heating, which among the other things describes different topologies of various inverters.

A book that seems to go a bit more into electronics side of things is Resonant power converters. It has a chapter on inverters. It's very theoretical, though, I don't think that there is a lot of info about practical considerations (most of the book is not available for free reading, though, so it's hard to say).

All that said, I figured out that I need to get a thorough understanding of the basics because while I have some understanding of analog electronics, it definitely has holes and I'll start by going through "The Art of Electronics" (and maybe some other books about fundamentals) to have a really good understanding of all of the components involved. And then I'll try one of the books outlined above and then I'll get back to tutorials and articles online.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod