Non-PFC mains SMPS supplied from HV-DC. Theoretically possible, but efficient?

[debininja]

Newer SMPS have a PFC stage do this automatically before the main switch stage. The voltage stays more or less, constant, from what I understand, reducing the stresses on the bulk capacitors and decreasing high pulsed current draw from the wall outlet, in comparison to passive or no PFC SMPS. Apparently, 30% or so electricity generating capacity can be gained just by having people transition to PFC SMPS, and hence why a lot of EU regulations push for it.

Old ATX PSUs can be found with passive PFC (so the typical LC input filter, X and Y caps, etc etc), or even no PFC (for those horribly cheap SMPS).

Question is: if just the input voltage was changed from mains AC to a fixed 350V DC, would the old ATX PSU become more efficient? Has anyone here tried this themselves? My question is theoretical, and I'm not aiming to power an old ATX PSU off HV-DC any time soon.

Of course, the input capacitors would have to be swapped out for 400V capacitors (since it's no longer going to use a voltage doubler and evenly split and step up the voltage across the two bulk capacitors). And of course, the rectifier would no longer be necessary, since we're already inputting DC to it, buuuuut assuming we left it in place, it would have to be able to cope with double the power dissipation.

[Paul Price]

Conventional, non-PFC ATX PC power supplies already rectify(220V AC) or voltage-double(120V mains) the mains AC voltage to achieve +350V DC to feed the switcher transistors.

[debininja]

I'm aware of that. What I meant is, if the mains AC was switched to high voltage DC (350V), would the otherwise unmodified SMPS be as efficient, less efficient, or more efficient?

Say it normally, with the 120V AC input, has an efficiency of 75% with passive PFC. If you draw 100Watts at the output, you expect it to draw 133W at the wall. Now, if you swapped that 120V AC input for 350V DC input (and assuming the capacitors and rectifiers can handle the extra stress and voltage), what would happen efficiency wise?

[fourtytwo42]

I cannot see why you would think this would affect the efficiency as the rail voltage remains the same so the SMPS is unaware of the change.

[debininja]

Hmm, not exactly. The rail voltage sags quite a bit when it uses AC, so the input caps have to be able to withstand the high ripple currents after rectification to 120Hz pulsed DC. It may get 120V AC, but it's the RMS value, so the peak goes up to 170V. With a voltage doubler, it's 340V. Again, those are with low frequency 60Hz sine AC. Active PFCs step up the voltage to 350V DC using high frequency DC SMPS, so you can have smaller stresses, smaller ripple, and smaller caps.

It definitely will affect efficiency if 350V DC was used. I just don't know by how much--whether the outcome will be negative or positive.

[BrianHG]

I think what debininja is really asking is if there is an additional loss of efficiency in the voltage doubling stage from 120-350v.  There is obviously the main switcher losses in the step down, but, will removing that primary doubling stage remove it's associated power loss in the circuit?  What if we doubled the voltage in a different way, or, designed the main supply to step down from 120 to low voltage DC?

[fourtytwo42]

It definitely will affect efficiency if 350V DC was used. I just don't know by how much--whether the outcome will be negative or positive.

I am sorry but until you can explain the mechanism of this proposed efficiency change I don't think you will find anybody able to help you. Ripple voltage has no meaningful effect on efficiency.

[T3sl4co1l]

Assuming that the design is the standard design, and they aren't doing any funny business with AC mains voltage transformation (no AC transformers) or zero-crossing detection or anything like that:

No change.

The ones with a doubler may suffer from imbalanced voltage between the two capacitors.  These need to be switched to 240V mode (which, to be fair, runs them in series anyway, by design).

You will need to replace the fuse with one rated for DC.  These are usually more expensive, but are quite available for ratings like this (some amps, 350VDC+).

Precharge and current limiting are also important considerations.  The short-circuit current from a "master PFC" unit will be huge; it may be desirable to insert series inductance in the supply (>100uH) and/or resistance (if the power supplies have NTC inrush protection, that helps) to allow hot plugging.

Tim

[debininja]

Perfect. That answers my question. Same efficiency 
(Note: I'm not planning on running anything on 350V DC ...not yet at least...)