Turn-on and off times in a FET, weird data

Yep. One more knob to turn when trying to pass compliance.

Or a bead. Or a kicker inductor.

I usually do that as well but then in the end there remains a 0R "resistor" in that spot.

Sure, leave yourself all those options.

I've thought about the kicker inductor, to speed up gate rise time, but I've never done it. Seems like a mosfet, with its big gate charge plateau, doesn't really look like the capacitor in a classic RLC peaker situation; it looks more like a short circuit that gobbles up amps. And for the fast stuff with low Miller capacitance (phemts and such) an L would probably just make it slower.

I was thinking about using an EPC GaN fet to pump 100 amps into the gate of a mosfet. Might try that one day. Source inductance sees both the drain current and the gate current!

If switching big loads I don't see much sense in a peaker either. But in a Laser driver it has helped me.

Just don't blow out the gate bond wire :-)

That depends heavily on what "run of the mill" is. If it's something designed in the 80s that would be hard pressed to drive anything at 100 mA, plus a "healthy" 22R added in series (there are still places out there, where anything faster is considered "too new" and avoided with all the strength available), then "fast" is a rather relative term. Besides, the FET does not actually have to be slow. It just has to pass a review at design-in. Later, during the EMC testing the resistor will be increased until it's slow enough anyway :)

Well, my post was not meant all that literally, more as a pun on certain design methods in certain places. Obviously, engineers with common sense will know that the thing is much faster.

Besides, there really are still some switcher chips from the times of the dinosaurs out there, that seem to place their output driver into a tar pit - and customers who appear to like them that way. They will sometimes intentionally be clocked at slightly below 50 kHz (or even slightly below 30 kHz) in order to place the 3rd (and 5th) harmonic below 150 kHz, under the radar of EMC tests.

If all else fails, it's time for: begin repeat Rseries := Rseries + 10 Ohm; Csnubber := Csnubber + 1 nF; Screw(EFFICIENCY); until EMC.ReTest() = PASSED; end;

Actually, I just meant Germany, not any specific area. I'm in Berlin and have been living here for almost 20 years, since my family moved here after the Soviet Union's collapse. Basically, I've grown up here, since primary school age.

Greetings Dimitrij

Something in the 1-5 ohms Rdson range, or if bipolar something that will push 3-4 amps into the gate.

But who uses those anymore? Except with a follow-up driver.

Not out here because that would greatly increase switching losses and things would get toasty.

Full confession: I have run many switchers at 130-140kHz.

Ah, so you must be totally fluent in German then. Most of our relatives are from Berlin (Lankwitz/Marienfelde). That city is way too big for me though, I am a country boy.

And then there's ferrite beads -- high impedance just sitting there, so they damp well, but put more than 100mA or so through and they saturate. Who needs diodes? Miller overdrive in one tiny chip component. :)

Saturation manifests as a propagation delay (almost no current flows until saturation occurs), without affecting rise/fall time. In fact, a nice square material could be used to sharpen risetime even further (physicists have been known to do this, though avalanche pulsers and planar triodes often win out when it comes to Pockels cell drivers and whatnot). Though, good luck finding a cheap ferrite bead that's specified for squareness.

Let alone find one that's specified for flux at all. It's almost like they don't want you to know they saturate in the first place.

Tim

In at least one place here, that would be considered rocket fast ... ... and an unspeakable heresy!

You know, there are lots of strange things in this world.

Even a follow-up driver will probably not behave cleanly, with a microsecond ramp on its input :)

Well, that's not 49 (or even 29) kHz. In that range the 2nd subharmonic might still be audible now and then. These things hiss like a snake in distress at the slightest chaotic behavior of the regulator (read: "all the time"). Not something that I would willfully design - but yes, they do exist, even nowadays.

Yup, German is my second native tongue (and btw. Staaken is almost as good as countryside for some reasonably practical purposes).

Greetings Dimitrij

That must be a weird place :-)

Yeah, but ... Otto H.Schmitt has fixed that around 80 years ago.

Yeah, one Dell printer had one. I banned it from the office. Animals aren't too fond of this sub-par stuff either because many of them can hear 30kHz.

Whenever one of my converters went into song during Bode analysis our Shepherd left the lab and gave me the looks.

Yes, but Neu-Staaken with all the highrise flats gives me the goose bumps.

_This_ is what is country side in my eyes:

It's the bike path from here to Walmart. This is a road from one part of our village to the other that I travel at least once a week, by mountain bike:

Sometimes these:

Maybe they've seen all his constant complaining, and just ignore him as a huge PITA?