Question about mains powered SMPS

[Digibin]

I want a SMPS to supply 0-30V 5A from a mains supply (230V). Is it feasible to use a single offline flyback converter to achieve this range of output from rectified mains? It doesn't need to hit exactly ground, a few volts above ground is fine. I'm looking at using the UCC28600.

Otherwise would a better approach be to use a fixed offline flyback converter to produce 30V and follow that with an adjustable buck regulator?

In either case it's for a pre-regulator to be followed by a linear stage so accuracy and ripple etc. are not crucial. I've not worked with switch mode power supplies at this power level before so could use some advice. What's the best approach here?

[nctnico]

Also look at the service manuals for the HP6024A and HP6012A. Both are high power (200W and 1200W) variable voltage power supplies using a flyback topology. In these designs HP used 4 MOSFETs instead of one so the energy which otherwise would need to go into a snubber now get diverted back into the buffer capacitors.

[Miyuki]

two switch flyback is nice topology for moderate power levels

Just problem with output filtering, ripple peaks like to pass even linear regulator. I personally prefer forward converters when I need silent output.

[Digibin]

Thanks for the responses. So do you think a single flyback using the UCC28600 would be able to produce a variable voltage from 30V down to near ground?

Or is it better to use a fixed flyback followed by an adjustable buck converter?

[T3sl4co1l]

Single range, single converter, no -- there's nothing to keep the controller running.  There's usually an aux winding to power it.  At best, you'd be looking at maybe 8V minimum to 30V maximum, or any ratio of that range.

The actual output range will be narrower than this (maybe 10-25V, or 12-20V), depending on load current, because of leakage between secondary and aux.

With a separate aux converter (maybe just tossing in a cheap Meanwell or CUI PS?), you can keep the controller running, and probably go all the way to zero.  Be careful that it doesn't go into crummy operating modes: for example, a peak current mode controller like UC3842 will go into chaos (resulting in high output ripple and noise) at low output voltages and high currents.

An average current mode forward converter is probably a cleaner bet, as it won't mind varying the output voltage, even at full load current, and prefers to operate in CCM where ripple is lower.

Resonant controllers can be used, as long as the basic operation is clean over that voltage range.

Beware of easy traps: at low output voltage, high current is difficult because the freewheeling diodes carry basically full load current, and the on-time is a very short pulse.  Make sure the switching transistors are able to do their job in this time.  This is noteworthy because I've seen it done badly:

If you look inside any one of those no-name Chinese bench SMPS, you'll see a big heatsink with a couple of preposterously oversized IGBTs under it.  The one I saw used IRG4PC50UD or something like that -- 600V 50A, way too old and big for the purpose (only 10-20A was needed), and severely, gravely underdriven (at best, 200mA peak gate drive -- for >200nC gates!).  When loaded to rated amperage, at below 10% rated output voltage (namely, 10A at less than 10V, for a 100V supply), the transistors cooked off and exploded.  Further investigation showed transition times of about 1 microsecond, out of a 50kHz switching frequency.  At low duty cycles (under 5%), it's literally a linear amplifier, never even achieving Vce(sat) during each pulse.  Shitty design, exactly what you should expect from a $300 piece of junk!

Tim