<krw@notreal.com> wrote in message
news:22o6qch4445a56qq36sb38hto39prbeqtf@4ax.com...
> Again, the fundamental is *NOT* the problem with switching power
> supplies. If it is, you're doing something *very* wrong.
Don't you just love generalizations?
I've got a pair of these wired up at the moment, as part of another project:
https://www.seventransistorlabs.com/Images/100WPowerSupplies.jpg
They just so happen to be running from a LISN, because I already know
they're noisy as hell, but I CBA to fix them right now. ;-)
So, what the hey, let's turn on the speccy and entertain that thought, huh?
https://www.seventransistorlabs.com/Images/100W_Flyback_3MHz_Conducted.jpg
(No, I don't have log sweep, so this is in segments.) Tons of crap at the
fundamental and low harmonics. Eww!
And yes, both are running pretty much spot on 200kHz. Kind of a silly
choice, 140kHz would be much better suited. But there you have it, eh? Oh,
but what about the rest of conducted --
https://www.seventransistorlabs.com/Images/100W_Flyback_30MHz_Conducted.jpg
Mid-band peak, about 84dBuV -- but still less than the 92dBuV fundamental
(note different ATTEN settings).
So, if one were to take your words literally, your argument is pretty well
fucked. If /NO/ supplies have issues at fundamental, then the above plot
doesn't exist, right?
But this is just one example, right? Maybe you don't literally mean "all",
you're just calling out a stereotype, and you've worded it badly. Over
several posts..
And what about radiated?
Well, I don't have antennas set up, of course, and you are welcome to
question the veracity of "300MHz conducted". But, whatever, it definitely
shows things. Here it is:
https://www.seventransistorlabs.com/Images/100W_Flyback_300MHz_Conducted.jpg
(The peak at 125MHz is related to the "another project". Actually, I'm not
sure why it's so strong, it should be well within a metal box. Another
thing to fix, it seems.)
Even if the mains wiring is particularly unlucky (-20dB or more?) in the VHF
band, and even if a lucky wiring accident made this radiate egregiously at
some frequency, it seems unlikely that those <= 60dBuV peaks could match the
fundamental, in a proper radiated test. (Also, there's less and less above
300MHz, and nearby activity at 360 and 540MHz makes that range even more
questionable. This isn't a shielded room, it's just a bench. Needless to
say, a proper 30MHz to 1GHz sweep ain't gonna be useful.)
And so, yeah, this is just a shitty test, on some shitty supplies, that
aren't even cleaned up for emissions -- but then... shouldn't they be all
kinds of horrible, if that's the case?
You're welcome to claim that /I've/ done something "very wrong" here. Well.
You've /seen/ the power supplies -- I'd be glad to share the schematic and
layout privately if you think you can spot what's "very wrong" with them.
But no. The thing is, I have seen it be the case, in many, real,
regulatory, tests: the PSUs don't kill radiated, the MCUs do.
Of course, I can't share these plots, you'll have to take me at my word on
this matter.
Most of those projects were powered by good, brand name, off the shelf
(e.g., Meanwell) PSUs, by the way. It seems they have the same general
downward slant to their spectra, without things being "very wrong"!
Tim
--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: http://seventransistorlabs.com