DIY Induction Heater Builds and References

[TommyDee]

That' it @Edgedamage . After deciding the -very center- of the coil of 31 turns, the center is on the ID. I have a hobby knife with a broken tip that I use to scrape the insulation from the wire. I make room to protect from the other wires by crowding the wires away from this center wire as you see in the pictures.

Tacking onto the capacitor's lead-wires just made it simple and robust connecting back to the board. I make the center-tap connection last with a lot of dexterity and VC breaks.

Another angle....



14 gauge solid wire J-hook holding the positive lead into the module. Wire is 16 gauge silicon hobby wire. High strand count and tough. All well supported by the center output pad.

 

[Edgedamage]

Thank you, wow now I can make smaller builds. And from what I understand the system works the same missing a cap and a inductor.

 

[TommyDee]

The system reduces output by about 5-8 watts at comparable voltages and therefore less current draw. Technically, you've reduced a 120W heater to 60 watts. I run it up to 80 because the FETs are rated for 120 watts and nothing gets remotely warm except the main VC coil. And at the right wire gauge and turns, it just barely gets warm.

The biggest difference comes from coil configurations. Count on a hotter IH at under 30 inches of 12 gauge wire, which is the stock coil as close as I can tell, to be hotter, require more current. Longer means less current. Sample - stock 65 watts at 10 turns; a coil was 60 watts with 11 turns and 70 watts with 9 turns. I am enjoying the 9 turn coil at this moment. They are the stock diameter. All else being equal so to speak.

...

And if you have this kind of board, I don't like the negative feed wire. I wire directly to this wiggly trace between the two source pins -



There is not much via feeding these leads. I don't like needless losses in traces. Some board with full ground planes don't suffer this issue. The copper grid on this board is not terminated that I could tell. Or else it would be terminated here.

...

And the switch bridging whatever trace connects that positive-out pad to the long trace. This enables LED's and the voltage to the gate pins on the FETs.

 

[Edgedamage]

Again thanks for the pics and explaining. I just want to confirm the center tap and positive input share the middle unused pad at the output stage of the board. And a switch connects the pad and the positive trace, to allow the led and the fets to fire.
Also my board is labeled MINI 028

 

[CaleidosCope]

Hmmm ... as a host of a Pacemaker (actually, a fancy double-duty one that also has a built-in defibrillator = ICD) this caught my attention. I never considered it. I don't want to discuss this with my cardiologist, but I'd like to know more about it. With whom should I discuss it, I wonder? Perhaps the ICD manufacturer (Medtronic) would be the best source. Dang ... why did I have to see this?????? But, thanks for making me think about it. Maybe I'll have to get on that's specially shielded somehow!

edit:
I just found this, regarding the use of Induction Stove tops.
"Induction cooktops produce electromagnetic fields (EMF) which may impact pacemakers and ICDs (Implanted Cardiac Devices). Notice in Induction Cooktop Use and Care Manual:"CAUTION: Persons with a pacemaker or similar medical device should exercise caution when using or standing near an induction unit, while it is in operation. The electromagnetic field may affect the working of the pacemaker or similar medical device. It is advisable to consult your doctor or the pacemaker manufacturer about your particular situation. We recommend that induction cooktop users with pacemakers should consult their doctors or pacemaker manufacturer regarding EMF levels and use of EMF producing equipment."

That's a pretty generic warning. Maybe at 12v DC, the little I.H.s for Dynavaps are not significant ... of course, that's just me standing in that river in Egypt ... da' Nile!!!!

Hey. Don't worry. Its just a warning. I don't expect nothing (I come from the medical side)... we are dealing with around 70 watts (correct me TommyDee) not with 1200w or 3000w with and induction stove. Keep cool. Watch your body and heart. Don't hold the IH near to your implants. Enjoy your live.
And think about reducing the power. We can use the range between 5 and 12V... if you are worried... wait a little longer with 7V or 9V

 

[TommyDee]

Today I replaced the coil on the FlatPack HalfPint. I went through 3 coils getting that right! A coil too short makes setting the optimum position much harder. I now have 2 data-points that tells me the optimum coil height is 9 turns of 12 gauge magnet wire wound on a 16-17mm mandrel. This makes the coils's height approximately 20mm. Unfortunately, that isn't enough wire so I also wound a few turns back to try and reach that overall original 30" wire length. The total turns is 11-1/2T with stubby legs. Both legs reach up from the underside of the board so they are out of the user's site this time.

I had reason to wander out in the wilderness and happened upon a wooden dowel of unknown material except that it is Vietnamese hardwood. Turned a new 16.2mm, 3/4" deep cup with a thin bottom. I tapered the inlet somewhat so it won't scorch quite so bad. I also used a 1/2" Forstner bit so the VC has plenty of room. These do get hot and I've been bit once or twice. Not like a VC but it'll jolt you if you park on it.

I had this coil wound already but I needed to shape the legs for the FlatPack. It needed to straddle the board by about a 1/4". Love my Leatherman! Very precise bends with no damage to the coil save a few insulation chips. Being already work-hardened, it was like cutting a diamond. A quick blue tape protection method worked in more sensitive areas, areas that could short out a winding in a static scenario. Very satisfied with the results. Only problem, its hardwired in the FlatPack so I can't get a current measurement.

Performance - I have some aesthetics work to do as the coil moved a bit north since it is a millimeter bigger. I've been want to move toward a two-tone PETG, clear and black. The pack is 1" thick throughout. The new wooden cup is 24mm long, and the coil is centered between the top and bottom. I have one millimeter adjustment left and it clicks -perfectly- with with the right amount of heat.

Longer coils work very well, of course. My aim with HalfPint and short coils was to see if I could keep things -very- thin. Well, I can with a coil performance hit. Not worth it in my book. Consider using a full 20mm coil length on the ID and add additional coils to the outside on the return trek. This all in an effort to keep a mean IH but make sure that the IH itself remains cool. That is so that we remain compatible with an open frame touchable unit which in turn will allow for 3D printed housings without melting them.

 

[TommyDee]

Interesting turn of world event have yet again changed the IR-sensor landscape. I guess the junk digital non-contact thermometers of the past arn't good enough for the health issues at hand today so they are coming back to normal in price. Just ordered a GM550 for under $10 so the IR sensing is back on the DIY table.

Covid thermometers don't have the same range and are more sensitive. That frees up the world stock of 550C sensors.

And Fuck-It! BOX!


Wood and glue is just fucking messy. 2:46 minutes and there's a box. Waiting on some switches so I can make a lit.
Made the cup out of the Vietnamese dowel. A full 24mm deep with a marker at 4mm. This thing is tight in the coils at 18.6mm. I thought I had it but at the last second the last coil tweaked on me. With the wooden core it is fine but damn that pissed me off. This thing ain't going nowhere. I can do micro-adjustments by squirming the coil. This I call adjustable to your liking. Nice clearance to the VC makes for less wood-burning. No more 'hot' on the far-end of the wooden cup. It is about 6mm thick.

64x50x39[+4] for the lid and corded. Cute for a full generic module.

 

[TommyDee]

How's this for lazy... Didn't even bother to move the switch;



Learned a new term - Electrical Reference Depth.

 

[Edgedamage]

Well removing one inductor and one cap works great. I did 5 back to back tests with this configuration no problems. Saves space and saves buying a MOSFET switch.

 

[TommyDee]

Very nice! Yep, that's how it's done.

 

[TommyDee]

Today I learned how long the 18 gauge wire is on the coils to make 100uh chokes. Turns out to be 33 inches. Since I have a spool of 18 gauge magnet wire, I thought I'd try to wind a HalfPint inductor with the wire running parallel. I might try to bring it back to 100uh by adding winds.

Back... I now have twice as much wire on the coil. Not twice the turns, more 150% from 31 turns to 45 turns for both lengths. This should put the inductor at around 60-70uh. Either way, it has a lower frequency of the HalfPint circuit and it still shares the core.

I used the 11-turn coil to test this circuit. Bread-boarded, I got 70 watts at 12V. This is a nice tame IH.

The other accomplishment I was looking for was another level of compactness. This required another strategy that I've avoided so far, drilling holes in the board. I put the capacitor at the very edge of the board and stood it up. This provided access to 4 holes for the inductor, and in turn, it leaves plenty of room on the board for a robust coil.



Backside is treated as before. This is the board with the common ground plane. This made termination and the switch location very simple.

We are now within the confines of the printer circuit board including the coil. And it also accounts for a full 20mm coil length. That makes the envelope 56x38x38 including the wires and the switch.



I don't thin I'll keep the coil orientation this way. I still have enough 12 gauge to wind a coil that stands up off the board. At 11 turns, it draws close to 80 watts at 12 volts on an '18 SS tip. That is perfect. Earlier I said it was tamer but that was the Ti tip. I should keep these straight. the '18 tip and cap is the best coupling pair for these tests.

Also remember that when you drill the board, there is no plated via. Note the wire between the capacitor lead and the coil lead. On this board, both sides have the coil trace. Be sure you account for the beefiness of the current path. I am not sure I can do this cleanly with my other boards. I might have to flip the board for better trace access.

There is still one curiosity to this circuit; It still draws a lot of extra current when 'idle'. I will have to see if coil orientation has anything to do with it. I also don't know if inductors interfere with capacitors. Or, in this case, it could easily be due to the extra windings on the inductor.

One last observation; I have the lower voltage FETs on this board from the MOSFET switch. They do get warm in this configuration. I think it was more efficient with both capacitors. I will add some beefier copper traces when I mount another coil. Right now, it's 7 seconds to the click on an '18 SS tip. And it's set for HOT.

One thing I am confident of, if you have 18 gauge magnet wire, you could rewind a HalfPint coil with maybe 36 inches of wire. Run 15 turns, leave a tap-length and continue with another 15 turns. Now you have a center-tap length without scraping. Likewise, I could have stopped with the parallel winding when it rejoined at 31 turns or so [extra turns wont hurt]. The take-away is to have equal lengths on both legs. A parallel winding will make sure the core saturation peek is optimal. I will be testing this one to the hilt to see if I'm stressing anything. This is definitely limited to around 80 watts with these lower voltage FETs.

 

[RedEyeFlightControl]

Today I learned how long the 18 gauge wire is on the coils to make 100uh chokes. Turns out to be 33 inches. Since I have a spool of 18 gauge magnet wire, I thought I'd try to wind a HalfPint inductor with the wire running parallel. I might try to bring it back to 100uh by adding winds.

Back... I now have twice as much wire on the coil. Not twice the turns, more 150% from 31 turns to 45 turns for both lengths. This should put the inductor at around 60-70uh. Either way, it has a lower frequency of the HalfPint circuit and it still shares the core.

I used the 11-turn coil to test this circuit. Bread-boarded, I got 70 watts at 12V. This is a nice tame IH.

The other accomplishment I was looking for was another level of compactness. This required another strategy that I've avoided so far, drilling holes in the board. I put the capacitor at the very edge of the board and stood it up. This provided access to 4 holes for the inductor, and in turn, it leaves plenty of room on the board for a robust coil.

View attachment 1839

Backside is treated as before. This is the board with the common ground plane. This made termination and the switch location very simple.

We are now within the confines of the printer circuit board including the coil. And it also accounts for a full 20mm coil length. That makes the envelope 56x38x38 including the wires and the switch.

View attachment 1840

I don't thin I'll keep the coil orientation this way. I still have enough 12 gauge to wind a coil that stands up off the board. At 11 turns, it draws close to 80 watts at 12 volts on an '18 SS tip. That is perfect. Earlier I said it was tamer but that was the Ti tip. I should keep these straight. the '18 tip and cap is the best coupling pair for these tests.

Also remember that when you drill the board, there is no plated via. Note the wire between the capacitor lead and the coil lead. On this board, both sides have the coil trace. Be sure you account for the beefiness of the current path. I am not sure I can do this cleanly with my other boards. I might have to flip the board for better trace access.

There is still one curiosity to this circuit; It still draws a lot of extra current when 'idle'. I will have to see if coil orientation has anything to do with it. I also don't know if inductors interfere with capacitors. Or, in this case, it could easily be due to the extra windings on the inductor.

One last observation; I have the lower voltage FETs on this board from the MOSFET switch. They do get warm in this configuration. I think it was more efficient with both capacitors. I will add some beefier copper traces when I mount another coil. Right now, it's 7 seconds to the click on an '18 SS tip. And it's set for HOT.

One thing I am confident of, if you have 18 gauge magnet wire, you could rewind a HalfPint coil with maybe 36 inches of wire. Run 15 turns, leave a tap-length and continue with another 15 turns. Now you have a center-tap length without scraping. Likewise, I could have stopped with the parallel winding when it rejoined at 31 turns or so [extra turns wont hurt]. The take-away is to have equal lengths on both legs. A parallel winding will make sure the core saturation peek is optimal. I will be testing this one to the hilt to see if I'm stressing anything. This is definitely limited to around 80 watts with these lower voltage FETs.

These mini builds are inspiring, this is what I was going for, long term Nice work so far! I'm looking to do something similar and have been working on a direct-to-piece IH, just with the stock ZVS driver for now. Spicy!

 

[TommyDee]

Welcome to my Covid Insanity Project This has pretty much occupied my time during this lock-down.

@RedEyeFlightControl -You said you worked with IH for decades... what discipline?

I have hardly anything to fiinish an IH at an artisan level. Yesterday I realized I don't even own a proper switch or LEDs. But I am well stocked on magnet wire now and 3D printing is making sure tooling isn't an issue. Now it is optimizing and testing. So much fucking testing!!!

Although I know my HalfPint half-ZVS circuit is different in evrery way possible as to how it interacts with the tip, my two favorite IH's are on both ends of the spectrum. Either a full implementation with a 9-turn coil 12 gauge, and the FlatPack HalfPint that uses one cap and a split inductor along with an equivalent coil. I don't have an ultimate favorite yet. I do know which one steps outside with me though. Self contained is a huge plus.

And to go back to the modified unit from today - No go. If I add another capacitor, maybe. As it is the hobbled FETs get hotter than hell. I feel them through the board. Tomorrow I will rebuild it with the coil and inductor of choice.

BTW, don't take up inductor winding to relieve stress. It won't



I wound this to be the same as the original HalfPint coil. I know this works without a sweat with the original ZVS FETs. I hope it works with these MOSFET switch FETs. The difference is that the circuit calls for 60V FETs and I now have 48V FETs in place. I'm thinking that the extra wire length on the inductor is allowing higher voltage swing.

If these FETs don't work in a HalfPint configuration, no big deal. They stay perfectly cool in a full ZVS configuration even at 80 watts. This is a mere imbalance of sorts.

This coil is also smaller. By moving the coil and capacitor to the other side of the board, the coil will not be taller than the capacitor. Right now the coil is on top of the zener diode.

 

[RedEyeFlightControl]

You said you worked with IH for decades... what discipline?

Not IH in particular, but advanced amateur electronics, data systems, digital and analog tinkering, design, repair, and troubleshooting across PC/Network, storage, Automotive, mains/Low Voltage, comms, vintage hifi, and other related disciplines. If it has electricity, I've probably messed with it. I'm no expert in electronics in particular by any means but well rounded enough to integrate and learn as needed, and consider myself skilled enough to know when not to do something stupid. I'm 20+ year veteran IT Systems and Network Professional by trade, but dabble+ in anything that has wires. I am generally the type to laugh at danger placards while peeling off warranty stickers. "pft"

I fell in love with my Dynavap and have been throwing some cycles at some design ideas - a few of which have stuck to the wall pretty well. One was pretty much exactly what you're up to with the half pint, but with a slightly different formfactor (because let's face it, there's only so many ways we can stuff a toroid next to a coil next to a cap next to a board ). I think the creativity will lie in how it's tuned. Great info!

I've spent the past few months doing some deep dives on coil design in particular and have a few designs on my board currently awaiting some IRL testing. I've had a smattering of parts for just that showing up this week with more on the way. Though, thank you COVID, my first few batches of stock ZVS modules are *somewhere* over the rainbow and have been for some weeks (months?) now. Additionally, I'm looking more to bolster the DIY market first before getting into the nitty gritty of custom PCBs and heater circuitry. There is absolutely a huge and growing market for boutique and artisan offerings, which is amazing. I can't wait to see what comes out! The new Flix rig that's out looks pretty neat, too. Ultimately it's going to become more difficult to differentiate as the designs shrink in size, but hell, these are awesome directions everyone is headed in. Very inspiring.

> I have hardly anything to fiinish an IH at an artisan level. Yesterday I realized I don't even own a proper switch or LEDs. But I am well stocked on magnet wire now and 3D printing is making sure tooling isn't an issue. Now it is optimizing and testing. So much fucking testing!!!

Dude what, you're cracking me up. You need me to mail you some LED's and a few 1ks?

 

[TommyDee]

We're of the same stone in many ways. You might have a few more tools on the EE side than I do. I have one trusted RadShack meter and lots of junk reference materials. My trusted iron is RadShack dual range that is beyond classic, yet it can built anything. But in the end, I am flying blind and using some 3 years of IH knowledge from a skunkworx project. I was fairly confident in HalfPint functioning with expected results. Today's results are a simple interjected variance to see where the stress moved to. Found it!

Making small is a risk in thermal management. It is also the art of making the circuit efficient. EE's know exactly how to tune a circuit like this with straight catalog specs. Add to this the instability of the circuit when you throw temperature at the components and you have a huge moving nightmare.

And then you have that little hysteresis problem with the cap clicky. That needs to become obsolete. Give me the cap temp. This is phase II IH. Been done in Germany. Easy enough to implement. My sensor order got cancelled though. Plenty of time to get to that.

Good to meet you @RedEyeFlightControl . I have some spare magnet wire if you want me to try something. 15 and 14 gauge are available. Also got 18 gauge to play with inductors, obviously. I'm still hording the 12 and 10 gauge until I finish winding. The 10 is reserved for 2S operation.

As to power quality, I am not having issues with simple 200W DC-DC converters. They have output caps that seem to take care of the dips. I suspect an analog output is required to maintain a simple gate-interface which is already available if you are going Arduino. This changes the FET parameters significantly as they will now begin to heat up. How much variance will the circuit allow until heat-sinking or a larger case device is needed? Here's a thought for a circuit - 9-12.6V input but constant IH power. Can we drop 3.6V across the FETs to have constant 9 volt power levels?

My favorite coil parameters for the work piece today - 20mm long, 16mm ID, 12 gauge. Additional windings along the OD to reach an appropriate length. 20mm is 9 turns of 12 gauge magnet wire. The Standard is 30 inches of 12 gauge magnet wire in 10 turns. Changing the length of the coil wire moved the voltage requirement to hit the same wattage. These parameters work well 3s packs where the click never happens when the cells are at 9 volts under load. A self-alarm to charge the cells. I need to do the same for 2S. I suspect 7 turns (20 inches) of 10 gauge will bring me into the right ballpark. I don't have a lot of 10 gauge to play with.

 

[RedEyeFlightControl]

We're of the same stone in many ways. You might have a few more tools on the EE side than I do. I have one trusted RadShack meter and lots of junk reference materials. My trusted iron is RadShack dual range that is beyond classic, yet it can built anything. But in the end, I am flying blind and using some 3 years of IH knowledge from a skunkworx project. I was fairly confident in HalfPint functioning with expected results. Today's results are a simple interjected variance to see where the stress moved to. Found it!

Making small is a risk in thermal management. It is also the art of making the circuit efficient. EE's know exactly how to tune a circuit like this with straight catalog specs. Add to this the instability of the circuit when you throw temperature at the components and you have a huge moving nightmare.

And then you have that little hysteresis problem with the cap clicky. That needs to become obsolete. Give me the cap temp. This is phase II IH. Been done in Germany. Easy enough to implement. My sensor order got cancelled though. Plenty of time to get to that.

Good to meet you @RedEyeFlightControl . I have some spare magnet wire if you want me to try something. 15 and 14 gauge are available. Also got 18 gauge to play with inductors, obviously. I'm still hording the 12 and 10 gauge until I finish winding. The 10 is reserved for 2S operation.

As to power quality, I am not having issues with simple 200W DC-DC converters. They have output caps that seem to take care of the dips. I suspect an analog output is required to maintain a simple gate-interface which is already available if you are going Arduino. This changes the FET parameters significantly as they will now begin to heat up. How much variance will the circuit allow until heat-sinking or a larger case device is needed? Here's a thought for a circuit - 9-12.6V input but constant IH power. Can we drop 3.6V across the FETs to have constant 9 volt power levels?

My favorite coil parameters for the work piece today - 20mm long, 16mm ID, 12 gauge. Additional windings along the OD to reach an appropriate length. 20mm is 9 turns of 12 gauge magnet wire. The Standard is 30 inches of 12 gauge magnet wire in 10 turns. Changing the length of the coil wire moved the voltage requirement to hit the same wattage. These parameters work well 3s packs where the click never happens when the cells are at 9 volts under load. A self-alarm to charge the cells. I need to do the same for 2S. I suspect 7 turns (20 inches) of 10 gauge will bring me into the right ballpark. I don't have a lot of 10 gauge to play with.

@tommyD Glad to meet you too! It's nice to be able to bang heads together with like minded individuals. I too have an ancient RS meter I bought when I was probably 12 or 13! It's still around here somewhere in my pile of bench equipment... Great little meter. Hard to beat an analog needle for some things a digital just can't do.

I have a few samples of magnet wire in stock. I've poured over coil design and prior to actually building it, I need to work up a custom jig to actually fabricate it the way I'd like it to turn out. So, that will be a bit yet. I'm deep in Fusion360 territory doing component layout for the beta(s) in the (what I hope is) coming weeks. My lab includes 3d modeling, a fairly decent electronics bench, 3d printing, wood/mech/telco tools, and most recently a 60w CO2 laser cutting table (it's new, but inop, crap hardware I acquired for the tube and chassis for a full CNCshield+2020roller RetroFit). Full mechanic's shop in the garage and a growing wood shop. COVID got me busy af tooling up for this. I learned CAD some 20+ years ago and got back into it when I fell down the 3d printer hole (fuck me, that was fun, where's my wallet?)

I am actually focusing most of my efforts on microcontroller firmware development and am taking the digital approach to instrumentation. Not sure if you saw the teasers on my insta. The firmware is pretty much fully baked at this point. I'm really pleased with the 1.0.0 build. Now that the basic hardware is selected and tested, and the firmware is people-ready to see if anyone can actually break it, all it needs is a few various case designs to start based on what I've liked so far, and what people want. I'm taking notes if anyone is giving them! Just got my first batch of controllers in for the public beta this afternoon! A full flashing and testing rig is already baked, so it's down to : make something kickass to put it in, then vap my ass off on it while I show it off to people. A good product sells itself! And if it doesn't, I'll probably still be smiling and happy with my contraption.

A few thoughts on your excess heat -

In terms of the halfpint, are you using a center-tapped work coil? Or are you using the same toroid, (what is your winding pattern?) to do the same job as the two individual toroids from the standard Mini12? To my knowledge you need at least one or the other, the Mini12 dumps the complexity of a center tap work coil for independent coils on each side of the work coil to achieve that balance, IIRC. To my understanding of ZVS, the resonance of the work tank is the affector of the ZVS frequency driver, no? As one leg ramps up it dumps the zener latch on the opposing fet and flops the current direction, which then repeats on the opposing leg. This then allows the work tank to properly shift phase a full 180 degrees using tank discharge inertia plus some additional voltage on like a gas pedal on an already moving car.

If you're only driving the work tank off of one side of this circuit, and the other side is simply resetting, would this not be akin to trying to push a kid on a swing all the way up to 90, letting gravity pull them back to 0, then immediately forcing them up in the same direction again? The hard stop would be the lack of the other secondary winding and subsequent phase shift/ tank charge-up that would be absent, I suspect. Correct me if I'm wrong but my understanding of the ZVS driver is that it is novel because it relies on the timing of the work tank to resonate. I'm neither a ZVS expert nor an induction expert but these are questions I had asked myself considering the same idea. Functionally, it should work, I'd suspect, just only making use of "half" of the pendulum swing in the work tank, so to speak. Could explain the excess heat of resonance you're getting, if that is the case. It's typically exhibited as 4x the work voltage of the tank, as far as I know. Those on-board fets (30N06?) top out close to peak (4x) 12v around 60 IIRC so you'd very certainly be seeing some fast heat buildup, if that's the case. I'm sure you're aware that it is far from advisable to run fets that far up in the voltage curve. Some heftier fets would probably mitigate this but I'm not entirely certain that would solve it, just potentially make it go away by overbuilding it.

And for the PSU -

For an external PSU, smoothing caps will help sustain a constant output voltage for a power converter for sure, but fets are understandably picky when being fed "signals" from multiple sources. If you're feeding that ZVS driver with PWM, it could also be injecting all sorts of resonant noise into the power loops in the ZVS driver which is VCC shared with the work system, which in turn make the gates on the fets go batshit (the reason for using debounce as a workaround to tac switch VHF interference if my understanding is correct). In a fixed voltage supply, I am pretty sure the output cap values are rather dependent on the desired output voltage. My worry was that a "generic value" set of output caps in a PWM or buckboost might not be enough to correct oddities from voltages off tolerance of the "generic value". This is purely speculation. Do you experience the same overheat scenario when using a fixed source like 12VDC batteries or a regulated bench supply, vs the adjustable one? The PSU may not be designed with feeding an appliance that uses the input voltage as its frequency control.

One good trick to diagnose and troubleshoot is to follow the problem one step and check at a time. When does it occur and when does it go away? Whatever scenarios reproduce it, there's probably a commonality to eyeball. Elimination troubleshooting is very effective.

 

[Edgedamage]

Did the half pint treatment to my Hammond box build. The extra space allows the lid to close without feeling forced. Got rid of the MOSFET switch, also jumpered the onboard led pinout. To allow me to light up the tube when the fire button is pressed

 

[Edgedamage]

Good video to understand induction heating. And looks like a simple circut to build:

 

[TommyDee]

Okay @RedEyeFlightControl - I am officially jealous of your covid acquisitions. 3D printing is a big hole indeed. Been wallowing in it for 3 years now. Love it still!

My circuit is conventional. Straight work coil with a split inductor. Nothing new, just reduced. Exactly what @Edgedamage did. It makes the conventional ZVS into 6-8 second heater instead of sub-5. I find 60-70 watts more than sufficient as input power t the ZVS. The cut-off is around 40 watts where one can draw faster than the cap heats up if you have good flow.

I have not seen your controller efforts yet. That would be fun for a non-functional beta drive.

I have knocked out my WiFi on several occasions playing with power inverters. Yes, they trip each other up all the time based on timing issues. Resonances, really. That doesn't work for balanced circuits. FETs -can- be controlled with PWM. Since the circuit is bouncing high energy levels back and forth, the interrupted state still has energy to dissipate.

My current thinking on controlling ZVS power is a direct change in input voltage using sufficient filtering on the input to -never- drop below the minimum gate voltage. Now you have full control over the power of the ZVS circuit. This also makes sure there is no potential across the FETs to generate heat during the on state.

However, I do prescribe to a super-slow PWM that ensure a minimum on or off time of 1/2 second. I am assuming this is plenty of time for the circuit to stabilize. This has been well tested over the months. I've had no failures from manual activation, and therefore, a good hard lock by a digital controller should work well also.

Are you working with the thermal sensor yet?

@Edgedamage - Love the new build. Interested in hearing your take on the performance of HalfPint. Just to clarify for my next ebay order, is that the 12mm switch or the 16mm switch?

 

[RedEyeFlightControl]

Okay @RedEyeFlightControl - I am officially jealous of your covid acquisitions. 3D printing is a big hole indeed. Been wallowing in it for 3 years now. Love it still!

My circuit is conventional. Straight work coil with a split inductor. Nothing new, just reduced. Exactly what @Edgedamage did. It makes the conventional ZVS into 6-8 second heater instead of sub-5. I find 60-70 watts more than sufficient as input power t the ZVS. The cut-off is around 40 watts where one can draw faster than the cap heats up if you have good flow.

I have not seen your controller efforts yet. That would be fun for a non-functional beta drive.

I have knocked out my WiFi on several occasions playing with power inverters. Yes, they trip each other up all the time based on timing issues. Resonances, really. That doesn't work for balanced circuits. FETs -can- be controlled with PWM. Since the circuit is bouncing high energy levels back and forth, the interrupted state still has energy to dissipate.

My current thinking on controlling ZVS power is a direct change in input voltage using sufficient filtering on the input to -never- drop below the minimum gate voltage. Now you have full control over the power of the ZVS circuit. This also makes sure there is no potential across the FETs to generate heat during the on state.

However, I do prescribe to a super-slow PWM that ensure a minimum on or off time of 1/2 second. I am assuming this is plenty of time for the circuit to stabilize. This has been well tested over the months. I've had no failures from manual activation, and therefore, a good hard lock by a digital controller should work well also.

Are you working with the thermal sensor yet?

@Edgedamage - Love the new build. Interested in hearing your take on the performance of HalfPint. Just to clarify for my next ebay order, is that the 12mm switch or the 16mm switch?

I wondered if the split inductor would be effective at this scale and layout - seems so. Agreed on the ~60w, I run a 6a brick on my tabletop now so it doesn't exceed that limit of ~72ish watts. How critical is having a sub-5 second heater to everyone here? Just wondering. I'm seeing good rise times, but they obvious vary with settings changes and materials heated.

Conventional PWM makes use of averages at the millisecond or even nanosecond level depending on application. These fets have ~15ns gate act threshold and ~100ish seconds worth of rise/fall time. Even 100 ns is 4 orders of magnitude smaller than a millisecond. So, ramp time for one half of the circuit fully energizing should be somewhere around 15-50ns tops, about 4-5 orders of magnitude under a millisecond.

https://cdn.sparkfun.com/datasheets/Components/General/FQP30N06L.pdf - page 2, switching characteristics

I do recall MrC's comments about running a 16hz PWM input with success and I agree with SDS (where the hell is *he*), that it's slow enough where there is balance and stability in each pluse. Duty cycle of course determines voltage. I would of course agree with the notion that 16hz is probably slow enough, if you're running a high enough duty cycle to keep from pissing off the fets. I would still consider that highly experimental territory and something I'd want to put on my scope to observe at various voltage inputs, just to see if there's wackiness in the tank.

Interestingly enough the Thruster Module will make use somewhat similar technology, but I don't want to spill the beans *just yet*. I'm planning on batching a few different sets of betas depending on what people are interested in building, so you'll have some options.

The beta sales will probably be uniqe, in some way, as well.

@TommyDee Now I want to carve one of these modules in half, just to dink around with it! Are you winding the toroid wraps in the same direction, or opposing? Same size toroid as the og build?

And FWIW I've killed SO many manual push-buttons pulsing my rig. Thruster fixes that permanently

 

[Edgedamage]

Okay @RedEyeFlightControl - I am officially jealous of your covid acquisitions. 3D printing is a big hole indeed. Been wallowing in it for 3 years now. Love it still!

My circuit is conventional. Straight work coil with a split inductor. Nothing new, just reduced. Exactly what @Edgedamage did. It makes the conventional ZVS into 6-8 second heater instead of sub-5. I find 60-70 watts more than sufficient as input power t the ZVS. The cut-off is around 40 watts where one can draw faster than the cap heats up if you have good flow.

I have not seen your controller efforts yet. That would be fun for a non-functional beta drive.

I have knocked out my WiFi on several occasions playing with power inverters. Yes, they trip each other up all the time based on timing issues. Resonances, really. That doesn't work for balanced circuits. FETs -can- be controlled with PWM. Since the circuit is bouncing high energy levels back and forth, the interrupted state still has energy to dissipate.

My current thinking on controlling ZVS power is a direct change in input voltage using sufficient filtering on the input to -never- drop below the minimum gate voltage. Now you have full control over the power of the ZVS circuit. This also makes sure there is no potential across the FETs to generate heat during the on state.

However, I do prescribe to a super-slow PWM that ensure a minimum on or off time of 1/2 second. I am assuming this is plenty of time for the circuit to stabilize. This has been well tested over the months. I've had no failures from manual activation, and therefore, a good hard lock by a digital controller should work well also.

Are you working with the thermal sensor yet?

@Edgedamage - Love the new build. Interested in hearing your take on the performance of HalfPint. Just to clarify for my next ebay order, is that the 12mm switch or the 16mm switch?

The switch is 16mm. As for the half pint yes I get first click @6 sec insted of 5, still very happy with it.

 

[RedEyeFlightControl]

Oh and @TommyDee, kindly point me in the right direction of the temp sensors you were considering. VERY interested! Please

 

[TommyDee]

@Edgedamage - cool! And thanks to confirm the switch size. I would have pegged it at "maybe 12" Even being a mechanical guy, I suck at "how many marbles in the jar" contests.

@RedEyeFlightControl - The speed of heating has been known to outpace the DV clicky. One reported yesterday on a Caldron clicking at 3-1/3 seconds. Thermal propagation from the cap to the clicker is far exceeded by the heating rate of the chamber and cap. Presumably, the IH has little effect on the clicker -if- it is at the edge of the coil. With my 16mm ID/20mm long coils, for a hot '18 VC, the cap is level with the last coil. With connector losses and such, I get a reliable 10.5V at 60 watts or there about with a 30" long work coil wire.

Your the only other person to ever refer to this as a Tank circuit. Indeed. That is the old world name for it. Why did the yuppies have to rename every fucking thing we know from our past? Drives me fucking crazy! Arduino is Basic on steroids. Why change every fucking command? Okay, my peeve.

I need a scope for sure. Found a couple of dual channel scopes that might fit my budget. I really don't now what it takes for the circuit to settle but I do know it trips up on itself. On new builds, I've had it stutter before it finally kicked in. I don't have a clue as to what can cause that but I am here to say it happened more than once. After initialization, it worked great and never again stalled. Component flaw? Maybe. Bottom line, I want to give the circuit all the room it wants to stabilize. 16hz seems reasonable. Conditioned half or quarter hertz would be more than sufficient in sorting out a 'session' with minimal circuit interrupts. I was there when Scott Dean broke ground on RC speed controls. I well know what FETs are capable of. That little advantage won me a lot of 10th scale offroad races. I'm more worried about the energy stored in the coils and the caps when polarities clash. As long as the FET are on with a minimal gate drive, the circuit is happy as a clam. Therefore I do want to take the page out of the PWM book, but bring it down to a human scale of 1/4-1/2 second intervals. This is easy math for session settings where you may want to dial the switching level to something akin to on-demand heating to the right wattage (average) one can keep up with on the draw. On demand reheating or power draws like a log. I am convinced that this is an easy way to manage power in an IH with a microprocessor. Again, remember I also considered just driving the FETs' gates with analog voltage levels from the MP or a voltage divider w/ trimpot, but I don't think the FETs will tolerate the heat that would induce. Right means but thermal management will raise its ugly head. I have no doubt that 16hz will work, but also consider you are entering an audible range. I hate PWM squeal!

There is a whole world to discuss on the world of small and thermal limits if a design is going into a case. Our standard components are huge compared to what is required. Jeff and the Lucid team work with allowable thermal rise a lot differently than the standard module. I feel I have more leeway because of that. I therefore use a few simple rules like 18 gauge minimal wire size, minimize coil thermal changes for stability, and balancing impedance on all traces on both sides of the circuit past the FETs.

The inductor with the common core is wound continuous. It does not change direction. I would 2-24" length and got ~16 turns each to fit. I divided the core into two halves. The proper way is to wind the wires parallel. Doing a parallel pair of wires puts all the terminations in one neat bundle. Coils occupying half the core will have wires on opposite sides. Because we have so much core, I don't think there is a difference other than preference of where the wires come out. I'll upload today's build after dinner. It recovered the hot FET problem from yesterday. I'm very pleased with its performance.

Sensors - MLX90614 which should be what's in the basic non-contact thermometer. It has Arduino libraries.

 

[RedEyeFlightControl]

@Edgedamage - cool! And thanks to confirm the switch size. I would have pegged it at "maybe 12" Even being a mechanical guy, I suck at "how many marbles in the jar" contests.

@RedEyeFlightControl - The speed of heating has been known to outpace the DV clicky. One reported yesterday on a Caldron clicking at 3-1/3 seconds. Thermal propagation from the cap to the clicker is far exceeded by the heating rate of the chamber and cap. Presumably, the IH has little effect on the clicker -if- it is at the edge of the coil. With my 16mm ID/20mm long coils, for a hot '18 VC, the cap is level with the last coil. With connector losses and such, I get a reliable 10.5V at 60 watts or there about with a 30" long work coil wire.

Your the only other person to ever refer to this as a Tank circuit. Indeed. That is the old world name for it. Why did the yuppies have to rename every fucking thing we know from our past? Drives me fucking crazy! Arduino is Basic on steroids. Why change every fucking command? Okay, my peeve.

I need a scope for sure. Found a couple of dual channel scopes that might fit my budget. I really don't now what it takes for the circuit to settle but I do know it trips up on itself. On new builds, I've had it stutter before it finally kicked in. I don't have a clue as to what can cause that but I am here to say it happened more than once. After initialization, it worked great and never again stalled. Component flaw? Maybe. Bottom line, I want to give the circuit all the room it wants to stabilize. 16hz seems reasonable. Conditioned half or quarter hertz would be more than sufficient in sorting out a 'session' with minimal circuit interrupts. I was there when Scott Dean broke ground on RC speed controls. I well know what FETs are capable of. That little advantage won me a lot of 10th scale offroad races. I'm more worried about the energy stored in the coils and the caps when polarities clash. As long as the FET are on with a minimal gate drive, the circuit is happy as a clam. Therefore I do want to take the page out of the PWM book, but bring it down to a human scale of 1/4-1/2 second intervals. This is easy math for session settings where you may want to dial the switching level to something akin to on-demand heating to the right wattage (average) one can keep up with on the draw. On demand reheating or power draws like a log. I am convinced that this is an easy way to manage power in an IH with a microprocessor. Again, remember I also considered just driving the FETs' gates with analog voltage levels from the MP or a voltage divider w/ trimpot, but I don't think the FETs will tolerate the heat that would induce. Right means but thermal management will raise its ugly head. I have no doubt that 16hz will work, but also consider you are entering an audible range. I hate PWM squeal!

There is a whole world to discuss on the world of small and thermal limits if a design is going into a case. Our standard components are huge compared to what is required. Jeff and the Lucid team work with allowable thermal rise a lot differently than the standard module. I feel I have more leeway because of that. I therefore use a few simple rules like 18 gauge minimal wire size, minimize coil thermal changes for stability, and balancing impedance on all traces on both sides of the circuit past the FETs.

The inductor with the common core is wound continuous. It does not change direction. I would 2-24" length and got ~16 turns each to fit. I divided the core into two halves. The proper way is to wind the wires parallel. Doing a parallel pair of wires puts all the terminations in one neat bundle. Coils occupying half the core will have wires on opposite sides. Because we have so much core, I don't think there is a difference other than preference of where the wires come out. I'll upload today's build after dinner. It recovered the hot FET problem from yesterday. I'm very pleased with its performance.

>The speed of heating has been known to outpace the DV clicky.
Indeed, I noticed this on my first builds. Messing with the coil is currently the only way users can adapt their standard unit to their liking. You are 100% correct about the placement of the bimetal disc within the effect field. You can mitigate excessive coupling by tapering the coil at the thermometer end, OR ensure the clicker is past the end of the coil. Neither is ideal for our use case being such a small workpiece, but c'est la vie.

I think you will be *very* intrigued with the Thruster based on your comments regarding the complexity of pulsing and its effects on thermal inertia. That is the core concept behind which it was designed. Though it's not*true* PWM, and that is very purposefully so.

 

[TommyDee]

Looking forward to understanding your Thruster efforts.

One other element that only Mag Heaters has implemented that I feel is an absolute must in future offerings. We need to be able to change the depth of insertion. There would have been no need for a low temp cap if this was already built in. 1mm axial movement along the coil's axis is huge in clicky-delay. Having 3mm of adjustment range in this feature is necessary not so much for the user, but for the maker. Tuning to a single depth has got to be a nightmare for makers and a limitation for users. I'd happily give up a switch for dialing in temp by changing insertion depth. At the very least, make it adjustable as a setup routine.

okay, today's HalfPint story; one I am happier to stick to this time. Don't get me wrong, yesterday's build was very valuable for information and the build itself. I am very happy with today's updates.

First of all, yes, the higher frequency coil got rid of the overloading of the FETs. Now they barely get warm when heating the VC. There are still the under-rated FETs at 48V.

What I did wrong yesterday was not looking for opportunity. This 'scorpion' board is unique in my collection. The overall layout has a lot of useful copper that just needs to be exposed from under the solder mask. In this case, I can solder coils on either side. This is huge for this layout as you might notice from the images.

So again, I modified the board to separate the FET gate circuit from positive-input. I had to jumper the output center pin to positive input because this particular board has these isolated. Holes for the new capacitor location from yesterday. I added an 18-gauge coax wire since this is a wand layout and this makes it reasonably flexible. Also notice the 14 gauge rails I put on the component side to minimize the power circuit's impedance. All these mods fit within the height of the exiting components -



I wound the next 16mm coil very similarly to a previous coil. It is working perfectly in another wand at the moment. This is what I've zeroed in on for 3S operation. Delivered 60 watts right off the bat -



As stated earlier, I wound the inductor with a coil on each half. Dead-equivalent, I think I got 16 full turns. Original inductors have 31 turns and are rated for 100uh. The circuit is rated to operate from between 47uh to 100uh. That is the reason I fully believe we are still within the appropriate operating specifications. Notice that I needed to run a long lead for the positive input lead. Had I run the pair of wires parallel as I did yesterday, all the wires would have clustered in on location. I can still modify the other coil to do just that for a future build. The reason I mention this is to show what the difference means for clearance...



The other reason this works so nicely is that the height if the coil is now exactly 1" I cut a cup a little taller for a case.

So note, this will not be quite as simple on every board. If you have the board with a full ground plane, you might be in luck.



Performance - 10V/6A exactly what I get from a stock ZVS module with its complete coil. Idle current; 1.3 amps at 11 volts.