i am trying to desing a replacement switch mode psu for a computer. i am just trying to see if my idea is posible. how do you get - voltage from a positive source. EG in a computer -5 and -12 volts.
i'v taken apart many old psus and i noticed it looked like both -5&12 v are derived by a negitive voltage regulator from the +12 or +5 v.
is this right. if not then how is this done.

 

Sure, it's possible, using an inductive inverter. -5v and -12v are legacy voltages; really only around to support RS232, which is going away in favor of USB, Firewire and other more modern technologies.

Here's a great site for you to read up on SMPS's:
http://smps.us/

For a PC supply, you could start with an isolating converter to provide isolation from the mains - and then derive your other voltages from that point.

 

-12V is generated from the primary SMPS transformer. The transformer usually has four or five taps, +12V, +5V, COM, -12V. The +3.3V rail is often derived from a magamp regulator; if this is not the case then there is also a +3.3V tap. -5V from the -12V rail is easy using a 7905 regulator.

-12V is only for RS232 and for a few legacy devices. PCI cards take -12V as well, but PCI Express drops this.

 

A simple voltage doubler will also do this, though they have significant losses and tend to be noisy. With a decent source a voltage doubler will create almost any polarity and voltage.

 

i want to make it more eficiant than a switchmode psu. would this be more eficent using regulators than transformers. should i use a delay line to get the +5 v power good siginal.

 

Your average SMPS power supply is very efficient; on average 80% under full load. If you're messing with mains voltages be careful. A typical computer SMPS power supply has voltages of 350V peak in it, even in the US, because a voltage doubler is used when you select '115V'.

The +5V power good signal should be provided by a comparator. When the +5V line exceeds say +4.75V power good should be pulled high.

 

Look up charge pump -- good for doubling or inverting voltages. Usually for low currents in range of ma to 10's of ma.

 

Some simple theory can be read from the data sheet on this IC:
http://datasheets.maxim-ic.com/en/ds/ICL7662-Si7661.pdf