Good explanations, Russell. It is very difficult to directly derive anything above 48V to something around 1V. The efficiency suffered a lot here due to low duty ratio. This is the same in the VRM world. 5V (1st generation Pentium I/II) VRMs are easier, 12V VRMs are much more difficult (Pentium III/IV) but can still be achieved without a transformer. In 48V VRMs ( server/workstation/dual core??), you need a transformer here if not resorting back to (48V -> 12V --> 1V). Efficiency is the key here. Sorry about the VRM stuff if you think they are not relevant here but I think they are really related. Xiaofan >Date: Tue, 01 Feb 2005 16:32:27 +1300 >From: Russell McMahon > >>> ... The method simply consists of PWMing an input (usually mains) >>> 50/60 Hz AC waveform so that its amplitude is effectively reduced. >>> The reduction is NOT achieved by converting to DC and then using >>> high frequency Ac to reconstitute the waveform. There is no >>> intentional flyback or energy storage at high frequency in the >>> following transformer. > >> Well, if we can do this then can we also discard the transformer, >> pass the reduced voltage AC through a full-wave rectifier and get >> controlled DC? I guess the downside is that there is no isolation. >> Is adding isolation with a 1:1 transformers, and are they cheap? > >Yes, you could definitely (try to) do this transformerless, but there >are large advantages in doing it with a transformer when peak output >voltage is much lower than mains voltage. If you want a 25 amp output >at 2 volts then if you switch direct off mains you are trying to draw >25 amp peaks for short periods off the mains. This is effectively >supplied by the filter, but it's not nice. If you supply the 2.5V >demand from say a 3 or 4v supply then the current drawn at peak load >matches more what you'd see when the switch was run at 100% duty >cycle. High power short term current peaks are much reduced. -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist