Is PFC required for high-efficiency SMPS?

[malow]

hi, im new here. i'm doing some research in SMPS and learning about them, and a came to a problem that i need help.

i got a few 200W SMPS from a few places over the web for personal projects, and one of them got 93% efficiency at 220v. great. other one got 70%. other one 75%.

as my initial goal was a high-efficiency to avoid the need for cooling, i was digging into the SMPS that i got to see exactly what make some of them so low efficient and others "so good". and "bingo", good one i have is the only that have active PFC.

so, my question: how is the "range" of efficiency on non-active PFC SMPS? cause i was looking at PC power supplies, and the cheap ones with no PFC got around 65~75% efficiency, and the good ones, all have active PFC (80~95% efficiency)

thanks in advance.

 

[KX36]

ATX power supplies have very strict criteria on efficiency and power factor. Modern power supplies have to have both high.

Is the 93% efficient SMPS an ATX power supply? is it a good named brand? Is that the peak efficiency or efficiency over a range? There are a handful of expensive SMPS that can probably manage that. If the anser is no to both, the claimed efficiency may be BS. You can get a rough idea of efficiency with a resistive load and a few (4) multimeters to measure voltage and current in and out.

The 2 don't necessarily go together. PFC is generally a pre-regulator which is cascaded with the main converter. Cascading anything generally means cascading losses and cascading efficiency figures for a lower total. High efficiency might sometimes mean a more advanced topology than strictly necessary, a controller IC designed for high efficiency with various features for low load and standby efficiency, a larger footprint, more EMI etc as a trade off. It depends on the design goals.

You might get high efficiency by using synchronous rectifiers, a resonant topology, a full bridge topology, choosing the best possible FETs etc. It makes design more challenging. The 2 switch forward converter with self driven synchronous rectifiers I'm working on is about 76% efficient at 20% load, 86% efficient at 40% load (excluding the PFC). I'm still debugging and tweaking at the moment so haven't measured all the way to 100% I may have damaged the secondary FETs with too high a Vgs spike at high loads which would degrade efficiency and the rectifier body diodes conduct current which is less efficient than having SR FETs in parallel with schottkys or having a more complex SR driver circuit. It would probably be higher efficiency if it was a higher power converter as the couple of watts of fixed losses would count for a lower percentage. This is coming after a PFC, but I don't see that making a huge difference to efficiency except that there is a narrower range of duty cycle so it can be better adjusted for optimum power transfer, but the PFC has is own losses too.

 

[malow]

Thanks for the answer, it really helped

the 93% PS is a led driver, 36V/6A. I measured the efficiency using 180W load. this is the one.

then the manufactures ceased production to that model , and changed to this one. this is the 70% efficiency one (measured at 180W too)

as im in brazil, is very hard to get goods here. so i pretend to modify some good quality PC PSU's (up the voltage a bit) to have a high efficiency power source. and in the way im disassembling tem to understand them better

as the sellers lie about efficiency (the 70% one was sold as "＞92% efficiency" ) i can't keep buying "in the hope" i got a high efficient one

 

[KX36]

Interesting.

I'm just speculating here but the higher efficiency one looks like it has more semiconductors and more magnetics, so it probably is a more complex design, perhaps a bridge or resonant topology. The less efficient one looks from here ver much like a simple single switch flyback based on the single magnetic element and single primary semiconductor as far as I can see, which could be made quasi-resonant for improved efficiency, but the driving force for most flybacks is minimal cost.

I'd have to see the schematic traced out to really know what they're doing.

If you get pictures before you buy, that may help you verify whether they are likely to meet their efficiency claims but it's very vague, you could still have a simple topology with a higher efficiency than a complex one. I'd be very suspicious of anything that claims >85-90%.