Originally posted by: VirtualLarrySpeaking of which, I posted
this thread in GH recently, about PSUs, and got no responses. If you would care to take the time to comment, it would be much appreciated.
Done! You seem to have a pretty good feel for the generalities of electronics and PSU design. You miss an occasional detail but we all do that.
I'm curious what brand of generic you were in fact testing there. Was it by any chance an "FSP" (or AOpen/Sparkle) model?
By generic I meant no-name stuff and not the ones you suggested which to me are sort of "name brands". I buy mine on sales like this and use them in the many systems that my kids and grand kids and friends build at my place. It is my experience that if the cheapie generics spec individual outputs as well as combined power then they are OK. But of course a XXX Watt spec only means nothing.
The only question I have, with respect to the measured ratings you mentioned above, on the generic PSUs - wouldn't a steady DC resistive load, present a higher spec, due to the fact that it shouldn't affect the "efficiency" (PF) of the PSU in a negative manner, whereas the rather dynamic loads of a PC under normal/heavy operation might cause the real-world efficiency of the PSU to drop slightly? As well, if you
did any sort of "noise" measurements on the outputs of the PSU, a steady load like a light bulb wouldn't really cause any noise, but a PC system might, at least to my thinking. Or is that what your MOSFET testing was for?
Efficiency and PF (power factor) are two different unrelated specs.
Efficiency = Pout/Pin (true or real power)
PF involves a comparison of true/real power vs apparent/imaginary power. Low PF means extra input AC current draw (imaginary current) but no extra significant power consumption or power dissipation losses. In the US, homes are not charged for imaginary power but it is important for large power draws of industry.
The MOSFET was used to switch the output current draw (dynamic) while observing the voltage waveform for transients.
Dynamic loads should not appreciably affect the efficiency though.
I don't think that the biggest problem with "cheapy" PSU's is that they don't work, rather that they tend to output "noisy power" when stressed towards the maximum end of their capacity, combined with the fact that they are a bit over-spec'ed on the label, along with the fact that they tend to use cheaper, lighter-weight components (smaller heatsinks - they run hotter, and smaller buffering caps - if the AC voltage dips, they have less hold-up time), and most of all, they tend to have much shorter longevity because of the "cheapness factor", and when they go out, they also tend to take other components in your system with them. That's why they aren't well-loved around here, they are like ticking time-bombs to hardware, in some cases.
Ya - Some cheapies are crap but some are OK.
"Noisy" is not the right word because it is too general. "Noise" on the outputs can include:
Transients due to 115VAC input line voltage changes
Transients due to DC load changes
60Hz ripple
DC-CD converter ripple at maybe 1000 times the 60Hz frequency.
...Some MSI P4 boards are known to use two-phase rather than the recommended three-phase power arrangement, as a result, their VRMs are under more load (individually) and run hotter, and therefore are more prone to failure. Also, plenty of their mobos are missing caps, like on the power lines between PCIs slots, for exampleSo basically, cheap PSUs rely on higher-quality mobos doing more work for noise-filtering and regulation further "downstream", and cheap mobos rely on higher-quality PSUs doing more of the same "upstream", so that when you put them together (as often happens in a cheaply-built system), you are generally just asking for problems.
Mostly correct but this is more of an example of a bad design more than 2 vs 3 stage. The filter cap size can be reduced by 1/N though so this is an important advantage. Ya PSU and motherboard "cheap" that are really "cheap mfg cost" are not going to compare with a solid designes. Sorting out the difference is not always easy. I guess that is a good reason to go with "name brands" at a higher power spec if you want to be safe. I can't argue with that.
BTW, physical inspection of a PSU can provide some clues as to its quality. As you have said, the sizes of heat sinks and filter caps are important. But, one more important thing to look for is the orientation of the heat sink fins relative to the fans. The normal "old style" PSUs had heat sinks fins parallel to the airflow from front to back (air direction for a single exhaust fan). Some of the new 3 fan units involve *only* adding a fan to the side if the PSU without redesigning the heat sink. This airflow is 90 degrees to those fins so the cooling is NOT improved.