jason9

modified 4 years ago

Class-D 115V 3kW Inverter

1171

14:18:52

The blue trace is the output voltage. The green trace is the output current. The orange trace is the target voltage. The red trace is the PWM voltage. The two MOSFET-BJT pairs are actually two IGBTs which I simulate here with MOSFET-BJT pairs since the internal construction of an IGBT is such that it is effectively a MOSFET and BJT fused together. The 3kW rating is just based on what lamp wattage was required to cause the PWM to max out (4kW maxed it out but not 3kW, so I’m rating this as a 3kW inverter). The kW rating can be increased by reducing the filter inductance (which may involve other changes too), reducing the IGBT internal resistance (currently ~380mΩ), and/or increasing the supply voltage. Also, at 3kW this circuit will supply an RMS current of about 26 amps so the IGBTs and diodes will need some a lot of cooling. Given a resistance of 380mΩ and a current of 26A that works out to be a voltage of about 10V which means that about 260W will be dissipated from the two IGBTs for about 130W per IGBT. Doing the same calculation for the diodes with a 100mΩ resistance and an additional 1V voltage drop to account for the turn-on voltage gives a power dissipation of about 50W per diode. These figures can of course be reduced by keeping the resistance to a minimum, so using lower resistance IGBTs and lower resistance diodes will help, as will putting multiple in parallel, but in the parallel case one must take care to ensure that they are strongly thermally coupled or else thermal runaway might destroy them.

Also this circuit skips some details that would be present in a complete version. For example, I doubt there’s an op-amp capable of swinging 220V in each direction, so this should really be replaced with a discrete circuit with high voltage transistors. Also it’ll require it’s own power supply for the +/- 220V since the main power supply is only +/- 200V. And when something is disconnected from the output during a portion of the cycle when the inductors are conducting many amps then all that current will go straight to the caps and fry them with many hundreds of volts (if not thousands) so there should be a pair of protective diodes for each capacitor. They’ll be wired up the same way as the existing two diodes but connected to the blue and green wires instead of the red one. The only important characteristics of these diodes I think should be a high max current and a high breakdown voltage (greater than 400V). A high power rating shouldn’t be necessary since they should only conduct in unusual situations, although I imagine something like a light dimmer might actually trigger them every cycle so maybe a high power rating will be necessary.

Please note that none of this has been tested in real life and that I give no guarantee that this won’t fail horribly and/or fry whatever you plug into it.

I also welcome any and all feedback, especially anything regarding design changes and/or concerns because I don’t usually do high voltage or high current things (and this is both) so I’d like to know what you think. Is there too much current and should I increase the voltage and use a step-down transformer? Or is the current not as much of a problem as I thought and it would be better to reduce the voltage and use a step-up transformer? Or does omitting the transformer entirely like I did (or equivalently using a 1:1 transformer) happen to strike just the right balance in this case? What about higher wattage versions (5kW, 10kW, even 20kW or above)? I’m hoping to be as practical as possible with this particular project, although as per usual for me I have no means to construct any of this IRL so it’ll be strictly in the sim.

Also what are some reasonable parasitic resistance/capacitance values for the inductors and are they enough to affect the operation of this circuit? Will the ultrasound (40kHz PWM) that still leaks through the filter be enough to damage hearing if something plugged in acts as an efficient speaker in this frequency range? Will animals that can hear ultrasound be unhappy with this inverter? Or am I attenuating it plenty and I have nothing to worry about?

I’m looking forward to your feedback. Please don’t feel like you need to answer all the questions or anything, I just want to hear what you have to share.

published 4 years ago

kiani

4 years ago

Untangling the mangling,, https://everycircuit.com/circuit/6485775396634624

Issacsutt

4 years ago

This circuit is interesting and looks like it performs very well, but I don't think that you should classify it as a true invertor because its using a large 220v supply... Isn't a true Power invertor's purpose to convert a low voltage DC source into a 120/220v Residential AC source, to suffice for when there is no Residential Mains Power available?

kiani

4 years ago

18, 24v truck batteries would do it.

jason9

4 years ago

As I understand the purpose of an inverter is to simply convert DC (any DC I guess) to 120/220V residential AC. As for acquiring that DC supply, that’s another matter and can be handled by boost converters and such if you need to start with a low voltage DC source, but of course the current draw on the low voltage source would be immense if you want 3kW. For 12V that works out to be 250 amps, plus a little more to account for inefficiencies.

jason9

4 years ago

Thanks kiani, your circuit is indeed easier to understand. I might change the layout of mine to look more like yours.

jason9

4 years ago

@issacsutt regarding the DC power source I had in mind something like the output of a home generator running on gasoline/petrol (or even propane or LNG) so it could generate “dirty AC” of any voltage and frequency which could be transformed to 200V and rectified and fed to this circuit.

kiani

4 years ago

It is definitely an Inverter... If it is not an inverter then what is it @issacsutt !? An inverter circuit does not know or care what the city supply is...

Issacsutt

4 years ago

Regardless of the input supply voltage, as long as it is DC only, and the output is AC and it provides enough power to drive a typical load without dropping much in voltage... than I would agree it is still an invertor; but not a very attainable one, because of the required high power high voltage supply / Boost converter. Kiani, As an example, there are power invertors readily available in stores that fit in your hand and plug into a 12v car socket and typically provide 110watts at minimum.

jason9

4 years ago

I looked it up on wikipedia and inverters can have DC supply voltages anywhere from 12V (car battery) to a few hundred volts (solar) to even hundreds of kilovolts (long distance DC power transmission). As I mentioned the use case I imagined for this particular inverter is that a gasoline/petrol generator would make dirty AC (voltage spikes, unregulated frequency, etc.) which would be transformed to 200V and rectified to AC and fed to this inverter to power a house or something,

kiani

4 years ago

Right I agree its not a practical circuit,,.. There must be easier ways to clean up ac supplies from generators,,.

jason9

4 years ago

I think the usual method is to just run the engine at whatever RPM produces a 60Hz output and just hope that whatever’s plugged in doesn’t mind the waveform the generator produces (which is probably roughly a sine wave but with a voltage spike every time the gas in the cylinder detonates) and that generally works but isn’t exactly ideal, especially if there’s something sensitive to the waveform. Also the generator has to run at a constant RPM and therefore have a mostly constant fuel usage regardless of how much power is actually being used. With my inverter thing here you get a nice clean sine wave and the generator can run at whatever speed is most efficient for the current power consumption, so in other words it gives a much cleaner output and boosts the fuel efficiency, especially for low loads.

kiani

4 years ago

Frequency regulator,, good idea..

Issacsutt

4 years ago

Sorry for the long gaps between responses, I've been out at sea for a few weeks.

jason9

4 years ago

Ooh, nice! Sounds fun!

Issacsutt

4 years ago

Yeah it’s pretty interesting for sure, gotta watch out for sea sickness in the big waves though

jason9

4 years ago

Oh I have motion sickness too, but pretty mild. I’ve never been on a boat before so I have no idea how bad or not it is out at sea. Also, any progress or anything to note on your radio project?

Issacsutt

4 years ago

Oh hey, sorry I missed your comment for a few weeks; I'm still at sea so far, but not for much longer... Sooo, I haven't made any physical progress towards the radio project yet, but I have had a few new ideas, such as one consisting of experimenting with just the electric field of RF, and to basically start by treating the tx and rx antenna as more of a really really tiny capacitor with a large air gap, and I beleive that will help me to be able to understand RF a lot more in the design aspect, espcially when thinking of how to receive and amplify the extremely small signal. With that being said, I think it will require, to start, design of at least a very simple low noise amplifer, to include a decent filter, and as much gain as necessary, and I expect or at least hope to acheive a range of a foot to hopefully even several feet at best using just a very simple straight wire antenna to start, and maybe progress towards using a metal strip and/or hollow metal tube for my second and third experimental antenna designs. Depending on how all that goes, i may move stright into experimenting with solely the magnetic field of RF, and/or skip to the electromagnetic designs, but knowing me, i'll probably have try both to feel like i truely understand everything from a thorough enough level. Let me know what you think and if you happen to have any ideas / theories to add!

jason9

4 years ago

This kind of thing is actually done in RFID tags but with inductance instead of capacitance (see en.m.wikipedia.org/wiki/Loop_antenna#RFID_coils_and_induction_heating). Basically there are two coils and they act as an extremely loosely coupled transformer. I feel that this would likely work a lot better than the capacitance because there would probably be a lot more sources of capacitive noise than magnetic noise (every wire everywhere would capacitively couple to your circuit, but only coils would magnetically couple). As for receiving the signal, I imagine that the signal will be largely drowned out by noise (although this depends entirely on the distance of transmission) and therefore need to be bandpass filtered to get the desired signal and reject the unwanted noise. Narrower bandwidths will reject more noise but will also reduce the maximum data transmission rate. Given that much of the noise will be produced by the circuitry itself (probably mostly thermal noise) you’ll want to amplify the signal as much as possible as soon as possible so that it’s much more immune to noise in further processing. Given that this process itself will inevitably induce some noise, I imagine that the best setup is to start with a filter (just put a capacitor in parallel with the receiving coil and maybe reduce the specificity if necessary by adding a series resistance to the capacitor) and then filter again after amplification. On the transmitting end all you’ll want to do is just supply your signal directly to the coil and that’s it. Of course, exactly how you do this depends on the inductance and frequency of operation, but if you have it resonate with a parallel capacitor then you can just treat it like a high-impedance load. For modulation and demodulation you can use any of the techniques already developed for normal radio communication.

jason9

4 years ago

Also, on that same wikipedia page you’ll want to look at the Small Loops section to see how best to construct your inductive transmitter/receiver antenna.

jason9

4 years ago

Basically the main thing to take away from it is that the “strength” of the antenna is proportional to the area squared multiplied by the number of turns (or maybe the area isn’t squared for near-field inductive coupling, I’m not sure since this equation is for far-field electromagnetic waves, but the basic idea is more area is better and more turns is better). Also putting a magnetically permeable material in the coil (e.g. ferrite rod) should greatly improve the performance by forcing the magnetic field lines that would normally pass by the coil to go through the coil. In this case I’m not sure whether or not the area matters as much, so try to go for more turns if you use a ferrite rod.

Issacsutt

4 years ago

Ok, so far I have managed to briefly transmit and receive an electric field, but it needs a much more carefully designed filter and amplifier on the other end. I was using the same oscillator circuit that I designed and posted as: “RL 1GHz Astable Multivibrator”, and was using a frequency between 35-50MHz at a pk-pk amplitude of between 100-700mV (I can’t remember exactly what it was cause I tried a variety of different amplitudes). As far as distance, I was able to receive at least an inch away, but I am positive it can reach much farther because I was only using a 2-2.5in antenna on both sides, and of course the received amplitude was greatly increased just by adding a plate on the end of the transmitter's antenna (yet no plate was necessary on the receiving side). Funny enough, it was so difficult to find any metal around my apartment that I could use and manipulate into an actual plate shape, so instead I used a simple 1 inch screwdriver tip 👍😅 …of course I wouldn't do this in a more finalized version of the project, but it's all I have at the moment unfortunately, I don't even have any aluminum foil 🥲

jason9

4 years ago

In the case of capacitive coupling I think you’ll probably need to connect the two circuit’s grounds together for any decent results. Otherwise it’ll just raise and drop the potential of the entire receiving circuit which is no good.

Issacsutt

4 years ago

I tried it and, I didn’t run into any of those problems for the moment it worked, but I was also using Super-Caps at the time. I just need a better way to manage the Q-point bias of the single stage amp.

add comment in editor