Mosfet Tesla coil driver

The only thing that determines how hot a MOSFET gets is the average power dissipated in watts. If you run 1 amp through the MOSFET with a drain to source voltage of 5V then you’ll dissipate 5W of power in the MOSFET. Now, I’ve never actually touched a MOSFET or really done any circuitry in real life, but I have a feeling that that much power would melt a typical MOSFET. Now, suppose you distribute that current between 5 MOSFETs in parallel. What will happen (if you bias them to maintain the 5V) is that each MOSFET will have 200mA through it for 1W, which is definitely less than five watts. Alternatively, if you only have one MOSFET on at a time but rapidly switch between them, then it spends 4 units of time at 0W and 1 unit of time at 5W for an average wattage of 1W, or the same as parallel configuration. Actually, the parallel configuration will likely dissipate less power by having a lower minimum resistance, and less resistance means less power dissipation. So, assuming one amp and an on resistance of one ohm, a single MOSFET will naturally dissipate one watt, and if you rapidly switch between multiple MOSFETs that watt will be evenly distributed between the MOSFETs. But if you put five in parallel the voltage drops as the new resistance is 200mΩ so the total power dissipated is 200mW and each MOSFET only dissipates 40mW, 25 times less than the full 1W. As such, I think parallel MOSFETs will work better than having the MOSFETs take turns, but perhaps constant current isn’t the best model for the conditions the MOSFETs are subject to in a Tesla coil. If I remember correctly, the main killer is the voltage spikes. Perhaps a small spark gap to limit the voltage to a couple hundred volts (or whatever you deem to be appropriate) could work, although I don’t know if it’ll activate quick enough, but I’ve heard that spark gaps are very fast acting.

@jason9, I really appreciate the suggestion, but I have already tried that we’ll known method, which Already have well understood for probably a few years now; but the issue is my mosfets still get to be more than 150°F hot after trying such method. However I have done some testing, and with a low enough frequency it actually helps. I have already well known about putting transistors in parallel to “ideally” divide the current equally and reduce power lost/heat dissipated... but I have found in several different types of circuits when dealing with a really intense over heating issue; that even if you add more transistors in parallel what usually happens is they actually build up heat very unevenly and one of the transistors suffers more instead of equally, which could be do to even small differences in the trans-conductive resistance of each mosfet and would only take a little bit of heat to worsen the in-equilibrium and further damage one transistors a time.... so I find that the way you drive them needs further care and improvement rather than just adding more transistors.... this is just one idea, but another might be instead of just using multiple transistors directly tied together, you might be able to effectively divide the current if you were to control each transistor with its own constant current driver and then tie the drains/collectors together, but I would have to find a simple way to implement several current sources before I would actually wanna try it.

Oh man, I just wrote a bunch only for EC to crash when I tried to send it. Ok, I’ll try to restate everything I said but with fewer words. Basically, if a transistor drops from say 100mΩ to 20mΩ with increasing temperature then you could add a 100mΩ resistor to each transistor so it only drops from 200mΩ to 120mΩ, thereby decreasing the positive feedback to something hopefully more manageable. You could also (or instead) mount them all on a common heatsink to keep their temperatures the same, or even hook them all up to a heat pipe if that doesn’t work. Another option is to have a second transistor that when heated up takes voltage away from the gate of the first transistor thereby reducing the temperature feedback, but this will only work if the two transistors are thermally coupled (you could even solder together the casings of the two transistors, although that might melt the plastic). Yet another option is to have a microcontroller monitor each transistor and adjust the biasing as necessary.

Ah dang yeah that sucks, it’s happened to me too probably at least 10 times before.

I get what your saying but honestly all of those ideas (besides the microcontroller one which would just be something I don’t have any experience with), really all just sounds like tactics that would be analogous to building an amplifier that relies completely on the Hfe of a transistor which your not supposed to do because it’s not always the same value, it fluctuates a lot, and so a good design would be able to function within a wide range of those hfe values for example instead of just something specific. It just doesn’t seem like the methods you’ve suggested would be good enough, and measuring resistance at the mOhm level is not something I have the equipment to do nor would I likely be able to find that low of a value anyway... it’s just overall not a reliable enough of an approach in my opinion, it should probably be something more clever rather than just a quick solution like that

Well, except for the resistor solution I just came up with all those solutions in the moment. I picked up the resistor solution from @thebugger because he always has 100mΩ resistors in front of his power transistors in all his amplifiers and when I asked him about it he said it was to prevent thermal runaway and if I remember correctly he said that quite small values would work and that 100mΩ is probably overkill. So, if you can find anything around or even a little below 100mΩ it should (hopefully) work. Also, regarding my designs relying on the Hfe which is bad because it fluctuates, well the whole point is that it changes with temperature to create a sort of temperature sensor to create compensation.

Here’s an example circuit I made last night for BJTs since I don’t know what increased temperature does to MOSFETs in terms of how the parameters change: http://everycircuit.com/circuit/5811726002683904 I didn’t have much time so the temperature compensation isn’t 100% and the description is short and doesn’t say much. I think I’ll make a better version of that circuit later today if I have time.

Improved version: http://everycircuit.com/circuit/5273167436972032

Oh, and if you use resistors, put them after the emitter/source rather than before the collector/drain. This way, if the current increases, the voltage between the gate and source (or between the base and emitter) will decrease thereby creating a negative feedback.

If you wanted you could even do liquid cooling, insulating the wires with epoxy or hot glue and submerging the transistors in water. You could also use mineral oil or transformer oil instead of water to remove the risk of shorting anything so that you don’t need to take special care to insulate anything.

I appreciate all that you’ve gone through to try and help me with this, but I don’t have any of those resistor values and am really just trying to find a solution that I can already build at home without having to buy any specific components. And this whole project eventually has to be condensed enough to be able to fit on a pcb about half the size of a credit card so water cooling sounds pretty impractical as well. I hate to sound like nothing is a good enough solution but I seriously think there’s a better way than just using extra resistors

Well, given that your problem is overheating, I guess all I can think of that doesn’t add extra circuit complexity or use liquid cooling is to add a bigger heatsink or put more transistors in parallel as long as you make sure they’re all within a few degrees of each other, perhaps by mounting them all on the same heatsink. I suppose that given the restrictions, transistors taking turns is actually a reasonable idea. But you’ll still need to cool the same amount of heat, only difference being that said heat is already distributed across a larger area.

Oh I’m honestly fine with complexing things a little bit if you happen to have a good idea that would be practically fail-proof, because then I could just test it on the breadboard... And if it actually works well then I can get creative with saving space on a as pcb later. All I’m saying is if you can come up with such an idea than please let me know because I am open to whatever you have to say, I just won’t wanna try prototyping it unless it actually sounds reliable or innovative. And I pretty much have tons of all the basic components including some jk flip flop IC’s, BJTs (NPN+PNP)/mosfets (p and n channel, enhancement mode), resistors, caps, silicone + germanium diodes, relays, magnet wire and plenty more so feel free to let me know if you think of something good and I’ll test it and let you know.... this alternating pulsing transistors is just one thing I thought was possibly new and creative, but who knows maybe it’s not and someone else might have an even better idea, just let me know if you do think of something

I just looked it up and MOSFETs apparently have higher on resistance at elevated temperature, so I have no idea where the thermal runaway is coming from with parallel MOSFETs. Oh, I just had an idea, make a 100mΩ resistor from magnet wire. You can wind it in such a way as to minimize inductance. You can google how to do that.

You can find out the resistance of the magnet wire from the gauge. For example, a copper wire with an AWG (American Wire Gauge) of 30 has a resistance of 103.2Ω per 1000 feet at 20°C.

So, 1 foot of that wire would be about 100mΩ.

That’s good idea using magnet wire, but I’m looking for something that doesn’t have to do with adding resistance, you know what I mean 😕🤷🏽‍♂️

Also, you live in California right? I was wondering what part cause I’m currently stationed in CA for the Coast Guard, Just curious if I happen to be nearby

San Bernardino county. I live a pretty secluded life, so even if we are nearby I don’t think I could meet in person.

Ok, I’m pretty far away anyhow, up in Sonoma county. Was just curious.

CrazyEce91tblt suggested that the reason some MOSFETs were much hotter than others is due to manufacturing variations causing some to conduct most of the current. This could be fixed adding a potentiometer for each MOSFET and tuning as necessary until they all dissipate the same amount of heat.

Alright thanks Jason 👌👍