5
benj
6y

so, any EE here?
I'm a self-taught EE (shots fired! I was told to call myself an "inventor" so I don't offend my dear EE friends!)

Anyways... I just made a huge insight that wanted to share!

in the circuit that has been breaking my head for the past couple of days (switching DC-DC [boost, buck] converter the inductor takes care of holding the current stable while the capacitor the voltage. (apart from an low-pass filter...) The higher the frequency the smaller the capacity of the inductor needed, the less amount of wire, less resistance, more watts!!!!!

Comments
  • 2
    Anyone offended by you calling yourself a selftaught EE makes it sound like their degree is the best thing they could offer. Some areas in EE come closer to a craft than a science anyway.

    Onthe subject, though: Well, less losses in the inductor, but more in the semiconductors. If it were that easy, we'd all be using gigahertz switching and short pieces of wire as inductors.
  • 1
    @7400 username checks out 😁

    So, more strain in the mosfets or igbts because of switching speed? I've been so focused on trying to grasp what the inductor does (apart from filtering) than I haven't yet thought about the silicon.

    Working on getting a spice simulation going so I can start seeing and start playing with the parameters already
  • 0
    @benj the issue is that the transistor is used as digital switch. If it's closed, then no current runs, and with P=U*I, P becomes zero because I=0. If it is open, then it has very little resistance so that the voltage across it becomes nearly zero, and again P is zero.

    But in between, you have the zone where it is half open, and that's where the losses happen. Higher frequency means it is more often in this zone.
  • 1
    @Fast-Nop I so appreciate a plain english explanation (unlike quite a few text books!)

    From that I gather that the faster the switching capabilities of the transistor the better?
  • 0
    @benj yes, that is the flank steepness parameter, i.e. how fast you can get this thing from open to closed or vice versa. Should be somewhere in the datasheet for a chosen type.
  • 1
    @benj: On the first thought, faster switching is better, yes. Less time between fully switched on and fully off means less switching losses, if you make sure the timing of the different switches is correct.¹ Yet faster switching comes with its own problems: There are always parasitic inductances and capacitances in your circuit and these can resonate. A fast switching transition equals a lot of energy over a broad frequency spectrum exciting these. This leads to ringing, which meams additional stress for your semiconductors. Also, that energy has to go somewhere and it won't be fully dissipated within your circuit, ergo giving you EMI issues. In practical designs you'll have problems with semiconductors switching too fast as often as too slow.

    ____
    ¹ Since all diodes, like all transistors take their time to switch, the i teraction between semiconductors has to be considered as well.
  • 1
    @Fast-Nop @7400

    You guys have been so helpful! EEs both of you?
  • 1
    @benj: I'm an EE. I hope I'm not discouraging you or anything. All these non-ideal effects of real circuits can be a bit overwhelming at first but most of the time you just have to “keep them in mind”, meaning they often don't matter. I tend to be a bit on the cautionary side and sometimes it shows.
  • 2
    @benj Yes, I'm EE, though not an actual hardware specialist because I'm mostly into embedded software these days.

    Btw., here is an interesting link that measures decreasing efficiency with increasing frequency: https://eetimes.com/document.asp/...
  • 1
    @7400 you are not discouraging me at all! I'm aware of a few of the issues with real circuits that don't really show on calculations. I remember being annoyed because I needed a 262 resistor but they only sold 220! 😁

    TAoE said it right, a design that depends on exact component values is a bad design!
  • 1
    @Fast-Nop cool, thanks for the article!

    I'm a software dev looking to jump into firmware dev, but now I find myself needing something more specific than a development board and no hardware EE to help me! So SPICE it is and then KiCAD and Digikey! 😃
Add Comment