What does the efficiency bell curve of a good power supply look like?

[david90]

What does the efficiency bell curve of a good power supply look like? Should the curve peak out at typical load?

 

[berkeman]

What bell curve? What kind of power supply voltage regulator?

I'll guess that you mean the rolloff in efficiency at very low and high currents? It wouldn't usually follow a bell curve, though, just droop some. At very low currents, the power burned by the quiescent current of the regulator itself (the current it consumes to run the regulator) becomes a more significant fraction of the input power compared to the output power. At high currents, the power drop across the switching transistors and across the winding resistances in switching regulators causes a drop in overall efficiency. A linear regulator will not lose efficiency as the output current gets up near the current limit, however.

Actually, it would take a more careful analysis with the linear at high currents. There might be a small loss of efficiency, based on the size of the current limit sensing resistor...

 

[david90]

Is it realistic to expect a 85% efficiency from 10W to 1000W load from a switching ps?

 

[berkeman]

Is it realistic to expect a 85% efficiency from 10W to 1000W load from a switching ps?

That's a pretty big dynamic range. What are the input voltage and the impedance of the input voltage supply (like, is it the AC Mains, 110VAC, 20A breaker)? What is the output voltage and tolerance and allowed ripple? Are there limitations in terms of switching frequency? What are the cost limitations? How dynamic is the output loading? Like, does it swing from 10W to 1kW every second or so, or is it pretty steady once the load stabilizes? How quickly can it change?

My initial intuition is that yes, you can make 85% across that range, but it may take additional cost to be switching in different sub-regulator contributions, based on the output loading.

 

[david90]

the input voltage will be 3phase 115rms phase to phase. There will be 4 48V outputs. The combined power of the outputs will be 1000W and there is no cost limitation. I guess the output loading is stable and does not swing.

Thanks for you help.

 

[berkeman]

the input voltage will be 3phase 115rms phase to phase. There will be 4 48V outputs. The combined power of the outputs will be 1000W and there is no cost limitation. I guess the output loading is stable and does not swing.

Then yes, I would say that 85% is a reasonable efficiency goal. Where did the 10W number come from that you mentioned earlier?

 

[david90]

10W is the guaranteed minimum load.

At what reasonable load range can the PS achieve 85%?

 

[3gp]

check out http://www.80plus.org/manu/psu/manu_psu.htm

specifically, any of the pdfs: http://www.80plus.org/manu/psu/documents/Antec_EA-500_500W_80PLUS_Report.pdf