Guru
Joined: 17/10/2017
Location: BelgiumPosts: 469
| Posted: 09:17am 04 May 2021 |
I'm so glad you guys think the same about those "minor" mistakes.
Ok, first of all, excuse me if my English is not coming thru the way I actually want/think and so could misinterpreted, but I don't really want to use it as an excuse, this is the only place to practice my English. With 1000 pages I actually meant it's something interesting to discuss, the longer topics are mostly the interesting ones.
What I meant with higher energy is not so the energy needed from the power supply, but the energy needed to charge the gate to make the mosfet turn on or off. When you use a negative bias to hold the mosfet off, you will need to supply more energy to the gate to get the mosfet in on state again. Because you need to go from -5V to +/-4V to have the mosfet turned on, so it will take longer compared to going from 0V to 4V. Unless you use a different gate resistor. The same goes for 15V, if your gate is at 15V, it will take longer (compared to 12V) to have the mosfet in off state again. Of course that's only true if you keep using the same resistors. It's just something to keep in mind.
In the last years there were a lot of designs created and a lot of new components and pcb layouts were introduced and used. We are at the stage where it gets very challenging for me to improve things without just copying things from others with consequences. So more time spend on application notes, but also a bit of analyzing other pcb designs, wondering why they did it that way. Different components, for example igbt's and mosfets with their own character/catches, low and higher frequency stuff. Things are not so interchangeable anymore, if this weren't the case already.
I had a lot to read up, so I will try to give some interesting input.
If you have time to read this, it's an interesting discussion about the importance of a low inductance stage and as Wiseguy suggested, paralleling (with series inductor) different power stages to run 1 load, transformer, motor,.. is something discussed as well.
Link
I love this forum, but I sometimes go to the suggested forum, you can find a lot of discussion about (motor) inverter/pcb designs.
Another approach is having multiple synchronized power stages each with its transformer, and then connecting the secondaries in series. A bit like those cheap HF inverters stepping up the input voltage to a 350VDC bus. But then you come close to the warpverter topology.
Wiseguy you are totally right, I think I'm sometimes a bit fanatic if I'm discovering new insights, and sometimes it's just not necessary for example keeping the gate driver as close as possible to the mosfets. The ozzinverter/madpower board proves they don't need it, time will tell. You make very good arguments about the importance about those last nanoseconds, it all doesn't matter when di/dt is slow enough.
The Mad power board in combination with the nanoverter is probably the best choice for people that just need a proven design that works when finished. With tht components and simple mechanical structure. That is ok.
For me it's important to evolve to perfection although it's not really necessary, for example I want my inverter to start up automatically and displaying the fault or startup sequence . So my wife could use it when needed. That involves some pre charge circuit and output relais, something I want to implement this year, probably (hopefully) used ones in a lifetime.
Probably the same story with dedicated turn on/off resistors, it works like it is, so why changing, so sorry if sometimes I'm to pushy about something.
Wiseguy did a very beautiful compact layout with the nano power inverter, roll your own style. But I have a small problem with the gatedrive routing, it's running parallel to the battery busbar wich carries a lot of current. This is probably not an issue for his design parameters but when used under verry high load, you can have inductive coupling on the gatewires and potentially cause problems. There is also no separate return path either which is not ideally. But the area between the gate and return path is low, important to make no antenna.
Also the resistive path of the power conductors is not symmetric, the inner fets will have a higher current to carry because they are placed closest to the battery and AC terminals. The mosfets on the outside will carry less current. But then again perfectly for his design parameters. This could be easily solved when the ac output terminals were replaced to the outside of the board, this would create an equally path for each mosfet. Sometimes it's just not possible and designing is also looking for a trade of between different things.
Wiseguy I hope I didn't offend you.
First were some designs I was playing with for a laminated busbar, caps with the right dimension could be placed under the busbar, if the mosfets were placed on an aluminium bar to allow space, like in a design further in this message. The copper could be glued with some solar eva foil, possible creepage issues when voltage to high. You can see a layer of EVA between the + and - plate. Not yet under the AC copper plate. Electrolytics could be placed on the input, compensating the wire between the battery and inverter, the pp should only be there to compensate the pcb inductance?
It's also possible with pcb material.
Then I wanted to make a pcb, but I made the mechanical part somewhat complex, if I just had swapped the AC side with the batteryside, it would've been so much easier.
I would like to develop a version with mosfets on the other side of the board. It's already some time ago but here are some examples I was (wanted) playing with. So in the meantime there are things to improve. I'm still searching, maybe we can find something togheter.
The same board but the fets on top:
This was the little driverboard: maybe that ground plane is not necessary, a separate driver board gives some space on the board itself, placed under 90degree with the power tracks should make it more immune against noise. And makes experimenting easy.
You can easily fit an totem pole drive on such a pcb.
Peter this is maybe how you could place the mosfetdriver, I could've used smaller resistors.
Edited 2021-05-05 01:10 by nickskethisniks