I don't mind splitting hairs, so I'll split this one again
. I stopped from pointing this out in our previous exchange on the topic, but there's a method to my madness.
I doubt much of this is new to you, but I detail it just in case other readers are curious, and so you can see why I calculate values as I do for power supplies.
Voltage and wattage meters are averaging over time. Only a scope can provide a time sensitive view of fast peaks, and even then only within it's frequency limits. The transformers and capacitors within the power supply are also going to smooth out load peaks at the wall, and within the machine the many layers of voltage regulation will do the same.
When determining margins for safety or durability, average voltage, current or power requirements should be combined, as a range, with peak demands, which can be very short but repetative in computer equipment. In electronic design, absolute voltage peak is paramount when selecting capacitors, which is why you won't find 12v capacitors serving on a 12v line. For power, absolute peak isn't quite as important if the durations are short.
However, if one measures 10 amps of current on a 12 v circuit with a standard meter, that could be an average from a high frequency range which dips as low as 5 amps and peaks as high as 15 amps. That would be presented like white noise (in the Mhz range at he PSU most likely), such that the 15 amp peak figure is happening throughout operation, not just on rare occasions. This is similar to how PCM works on voltage for output, the reverse is working to average out current demand.
The same, of course, applies to power. If I'm evaluating a demand which when adding up the stated requirement of the equipment comes to, say, 400 watts, but I measure only 300 watts of consumption, I can assume the equipment is more efficient than the manufacturers have stated. Yet, if I grab a scope and watch in the region of frequencies at which the equipment operates (Mhz range for most of the intermediate VRM's in these systems), I'll note significant fluctuations in demand during what appears to be constant high performance operation. I'd expect the peaks of this white noise like variation to be in the vicinity of the stated aggregate, but I would expect my observation with a meter to be much lower.
At the wall there would be a significant averaging, such that if I placed a scope there I wouldn't see much of this Mhz range white noise style demand fluctuation, because that wouldn't work it's way backwards through the coils and capacitors of the power supply (let alone the switching sytem). Yet, for power requirements it's going to be the output stages of the power supply which experience high speed demand variations.
Now, yes, we're really splitting hairs here
. Most good units are designed to handle peak demands well above their stated limits, so a 500w PSU in good shape should withstand an average 500w power load with noise peaks toward the 600 w vicinity. I wouldn't want to design a black box circuit with such an arrangement, but it would probably work ok.
Effeciency, overall, of the PSU would tend to work toward the average load than the peak load of a noisy demand like computers present, but heat generation and stress on the capacitors don't. They get worked hard in very short time scales. Power loads which oscillate current work capacitors heavily, and in that sense the peak load is a little more important for longevity than the average. The capacitors are, after all, the electronic shock absorber when used in that role.
This is why I don't work with average power when determining what limits could be problematic, and why I inquired.
If the PSU were 3 years old, and the power measured at the wall were 400 watts, the PSU was rated 500...the PSU could be a problem. Even though the demand on the PSU would be 360 watts, which should be fine, peaks could be hitting 450, and if the unit were aged enough that caps are near 50% of their ratings, PSU noise could destabilize the machine. You can see this on a scope, but not with a standard meter, and even with a scope not at the wall.
Of course, for the OP, there isn't a chance there's a problem. Average power demand is probably 285 watts of the devices, peaks are at a maximum of 375. The PSU would have to be old, and electronically noisy no matter what the load, to be the issue.
To hair splitters everywhere,
