09-12-2011, 05:47 PM
I know a thing or two about switched-mode power supplies - one day I'll try to write a decent article, but am far too busy at the moment, sadly. But will try to answer questions as they pop up...
David's post was very good - the only thing that I'll add (simply by way of clarification) is that so-called "off-line switchers" - that is, anything that plugs into the mains, and produces an isolated DC output - do indeed contain a transformer, though not a "conventional" 50Hz one (as David indeed alluded to). Simplifying a bit, the circuit on the primary produces a square wave that is fed into a transformer that provides isolation and voltage scaling (via the non-unity turns ratios). The rectifiers on the secondary are fairly conventional - often only half-wave - but of course they are dealing with the higher frequencies, so need to use high-speed diodes (often Schotkey) and decent quality capacitors. Bear in mind the currents are often pretty high, but because the frequency is also high, the capacitor values don't need to be as big as normal. A commonly used rule of thumb for 50/100Hz supplies is 1000uF/A, but clearly you won't find 20,000uF in a 20 amp switcher...
Needless to say, the transformer and capacitors can be much smaller because the energy per cycle is much less - bearing in mind the period of 50Hz is 20ms, and the period of, say, 50kHz is 20us. They tend to use a powered ferrite core with carefully chosen properties. I've not (yet - never say never!) seen a torroid used for off-line switchers, although you'll see torroid cores used for DC-DC converters - computer motherboards being the best place to find these in the wild, as it were...
There is a massive safety issue with off-line switchers - after the Live and Neutral connections have been applied to a full-wave bridge rectifier, the negative output is now effectively live. If you do the thinking, you should see that it's negative w.r.t incoming Neutral by about 150V (ish!). Clearly, one shouldn't connect the earth clip of your 'scope to the so-called "Live earth". At the very least use an RCD, and preferably use an isolation transformer on the input to the power supply under test.
Back to Joe's question, that ripple looks unusually high to me. Typically, you'd expect to see low 10s of millivolts. But, it does depend on the load conditions, and some are a little odd when not (or very lightly) loaded. It might be worth adding a wire-wound resistor that takes 20% of the specified output load, just to see. Of course, if the output capacitors are shot, it will get worse under load.
What frequency is that waveform?
Hope this helps,
Mark
David's post was very good - the only thing that I'll add (simply by way of clarification) is that so-called "off-line switchers" - that is, anything that plugs into the mains, and produces an isolated DC output - do indeed contain a transformer, though not a "conventional" 50Hz one (as David indeed alluded to). Simplifying a bit, the circuit on the primary produces a square wave that is fed into a transformer that provides isolation and voltage scaling (via the non-unity turns ratios). The rectifiers on the secondary are fairly conventional - often only half-wave - but of course they are dealing with the higher frequencies, so need to use high-speed diodes (often Schotkey) and decent quality capacitors. Bear in mind the currents are often pretty high, but because the frequency is also high, the capacitor values don't need to be as big as normal. A commonly used rule of thumb for 50/100Hz supplies is 1000uF/A, but clearly you won't find 20,000uF in a 20 amp switcher...
Needless to say, the transformer and capacitors can be much smaller because the energy per cycle is much less - bearing in mind the period of 50Hz is 20ms, and the period of, say, 50kHz is 20us. They tend to use a powered ferrite core with carefully chosen properties. I've not (yet - never say never!) seen a torroid used for off-line switchers, although you'll see torroid cores used for DC-DC converters - computer motherboards being the best place to find these in the wild, as it were...
There is a massive safety issue with off-line switchers - after the Live and Neutral connections have been applied to a full-wave bridge rectifier, the negative output is now effectively live. If you do the thinking, you should see that it's negative w.r.t incoming Neutral by about 150V (ish!). Clearly, one shouldn't connect the earth clip of your 'scope to the so-called "Live earth". At the very least use an RCD, and preferably use an isolation transformer on the input to the power supply under test.
Back to Joe's question, that ripple looks unusually high to me. Typically, you'd expect to see low 10s of millivolts. But, it does depend on the load conditions, and some are a little odd when not (or very lightly) loaded. It might be worth adding a wire-wound resistor that takes 20% of the specified output load, just to see. Of course, if the output capacitors are shot, it will get worse under load.
What frequency is that waveform?
Hope this helps,
Mark
