I'd bet that many OEM chinese 300W and below supplies do not even meet the specs they claim to meet, and come nowhere close to meeting the ATX spec, which DOES include ripple specs.
The data sheets of a shit power supply may not, but the ATX spec certainly does, and a good power supply will also give a figure for noise as well as voltage regulation at full load.
No shit.
A supply maker should specify output voltage, power handling capacity, noise at full rated power, and voltage regulation through the load range, but particularly at full rated power.
It doesn't get any simpler, nor any more complicated than that.
And no, we do not need to get into power factor correction. That is not what this discussion is about.
On the other hand I feel these misspec'd PSU should be banned from the market entirely.
Maximum theoretical values yes, it is not what the user will expect to see when it's actually in a system. This is not so bad though, with any normal quality supply the resulting ripple is usually better than the spec.
It is not possible to rate noise at full rated power on any typical quality PSU with thermal fan feature. Ambient temp plays a large role in fan RPM so only within the explictly described ambient temp could the figure be valid. A graph would be nice, but a bit large to put onto a label and it would need be, as many psu don't have particularly detailed spec sheets, if the spec sheet can be found at all.
You also forgot recovery time again, it is among THE most important parameters on modern systems because they can easily have several amps current swing, constantly.
It would result in a higher temp, but the same issues of inaccuracy remain when taking temps of parts also encased in epoxy already as with all the ICs. The thermal imaging is also invalid for anything but the epoxy surface temp, an accurate sensor will have to be integrated INTO the part in the appropriate heat-dense spot if we want to make any effort towards a significantly more accurate reading. Otherwise, it's still a matter of thermal gradients, you'd only be measuring emitted heat at the point of measurement, not accurate die temp.
If you'd like to put epoxy everywhere though, be our guest... more data is always better than no data.
It IS also the spec in many other organizations, and on many other products besides PC power supplies.
Everything has to have a power source integrated into it somewhere along the way.
The stuff has to run in Japan, not just the rest of the world.
Our supplies would typically run full boat down to 85V in.
90 V is also the standard low line declaration in the ATX12V spec IIRC.
Here a tip on typical ripple requisites. We'll use the ATX12V specs as the example here.
Quote:
"The output ripple/noise requirements listed in Table 11. should be met throughout the load ranges specified in Section 3.2.3 and under all input voltage conditions as specified in Section 3.1.
Ripple and noise are defined as periodic or random signals over a frequency band of 10 Hz to 20 MHz. Measurements shall be made with an oscilloscope with 20 MHz bandwidth. Outputs should be bypassed at the connector with a 0.1 uF ceramic disk capacitor and a
10 uF electrolytic capacitor to simulate system loading."
ANY supply that actually meets the spec WILL supply the voltage AT the rated power, without varying in voltage out, OR having a ripple voltage included in the output that exceeds a 100% specified value.
If you supply does not, it is not an ATX supply. Your little obsession with recovery time and current swings are moot, because a conformal supply WILL function and a non conformal supply WILL NOT. If it was made well enough to conform, it will certainly take any swings while in normal operation, and do so at a VERY fast recovery moment.
If a switcher has all its outputs at rated load, and it meets the spec, ther will NEVER be ANY real world circumstance that takes it to that level again! INCLUDING your petty current swing prattle!
READ THE SPEC. It is TYPICAL of ANY supply made by ANY reputable supply maker for ANY purpose. It is part of the basic nomenclature for a supply!
Kick the dead current swing dog back into the corner.
Not at all. ANY educated user knows how to perform compensation for emissivity differences, and modern imagers do as well these days. Fully programmable as well as settable in real time. They are accurate to within 0.1 degree C in all cases, minimum. That INCLUDES epoxy (glossy or matte) and ANY other surface as well. Have you been in the stone age of IR technology?
The units I used to build were used by, and traceable to NIST. They also buy our black body calibration ovens. Beyond state of the art.
Using the same nonsense argument we could claim a good switcher has any and all positive attributes, but it still avoids the central issue that it is the more significant factor in voltage deviation than % of load, and that when it happens rapidly it is another way to describe ripple, though at a larger magnitude.
Sorry, I had to assume so because if you were talking about small magnitude ripple, it'd be irrelevant. It is not in any way necessary to reduce ripple beyond a certain point. The ripple induced by the load is more significant than that coming from any normal quality PSU at a steady load. You continually ignore this.
100% duty cycle is one way it could be spec'd, but unless expressly stated as such, is not necessarily so. On the other hand, any noise level would have to be put in the context of ambient temp as that is exactly what a regulated fan is combatting- the ambient temp as a tRise over external ambient temp.
Why are you continuing to supply basic, yet still non-applicable information?
The issue was one of your mistake- you seem to think full load ripple is the significant factor when it is not!
There is no such thing as a steady state full load from a PC. It is a dynamic load and that is were the largest magnitude ripple is seen. Whatever ripple you see at steady full load is a pointless measure beyond meeting the spec. That doesn't diminish it's usefulness in guarantee of at least a certain level of proper design (or supposed quality, but there are shortcuts to cheat one's way to lower ripple too which are counterproductive towards overall "quality").
The issue was your insistence a full load ripple was THE critera. It is only a beginning.
Nope, transient response and recovery time are extremely important, far moreso than your notion about ripple at full load. Do you not even realize that a few mV of ripple is buffered by every single device in a PC that has ceramic decouplers if not bulk 'lytics too?
Read up a bit, then a bit more. Here's a start, then proceed to the rest of that site and similar via Google search. Let us know when you're up to speed and can appreciate the significance of a highly dynamic load. Powering a static load is trivially easier, $1 parts change will fix that on most PSU.
formatting link
I can see you don't understand, and I'm ok with that.
It seems you still fail to recognize that it is the surface temp of an epoxy that is not meant to be a reflection of the die temp.
Until you DO recognize this, all your theories ignoring it will be faulty.
You want to argue going to more elaborate measures, which would be fine if we came reasonably close to an accurate die temp but we will not, the achieved reading would only be useful to determine whether the casing breakdown temp was nearby, not stability or die longevity itself unless presuming it to be the direct relation to prevention of casing breakdown.
Since it would be exceedingly hard to run these parts to a point casing breakdown induced die failure, rather than die failure first, then damaged case as a result, we can ignore the surface temp when willing to go to these more elaborate measures conceptually, and in practice.
PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.