SMPS vs LINEAR PS: Let's build up a serious technical review-list

[Centvrion]

The idea of this thread is to compare linear and switching PS through some adequate parameters measures. TELL ME IF IT IS STUPID AND WHY!
I don't have the necessary means (and know-how!) to do this but:
can anyone get some measures at oscilloscope from a SMPS (e.g. the famous Meanwell's) and a linear PSU (of course the last one after rectifier and big good capacitors)?
It would be interesting to know the amount of remaining switching/source frequencies (25kHz/50-60Hz) and (probably more interesting) ability to supply instantaneous peak of power(how is it possible to do this?Signal alterning full power pink noise?).
Let's make a list of measured PS...
Can anyone begin, specifying PS model, features and price, possibly the best known?

[mastodon]

TELL ME IF IT IS STUPID AND WHY

Because there are literally thousands of different design solutions out there, and they all perform differently, and it's unrealistic to even attempt to make a comparison between even a small percentage of the designs out there.
Because testing procedures and equipment should be highly standardized in order for any comparison to be useful.
Because there are too many parameters to test for, and I don't see many people here spending hours and hours testing power supplies in order to fill a database.

I firmly believe that power supplies can and should always be engineered (or bought) starting from the (technical, commercial and organizational) requirements the project poses on them, and then working towards the optimal solution. In my opinion, it's not feasible to list all of the possibly relevant parameters of PSU's and comparing them willy-nilly.

[luka]

i would have to agree

[Centvrion]

You are right of course, and I didn't explain very well....I don't mean e real comparison, and probably we have to make a selection of parameters (only the most important one) and to point out a standard procedure (hope someone can help!). The object of the measures could be only famous kits (for example Sure tk2050 or Hifimediy or snubberized gainclone or other...IMVHO if we gather a lot of data we can understand better the interaction between amplifier and the supplied current in relation with sound quality). The main purpose is to see if differences in sound quality between PSs result from used technology or in project quality (ok, both, but which is the most important?). Let's reshape the purpose and procedure to make it realizable, if it can be useful as I tried to explain...

[DF96]

Before you do that perhaps you should test your ideas on a simpler problem: how long is a piece of string? A similar problem: how wide is a cat?

[tomchr]

I don't know much about the width of cats, but the length of a cat seems to be proportional to temperature. At least until the length of the cat reaches saturation. However, if the cat is operated within its saturation limits, the temperature may be approximated by, temp = tan^-1(length).

On a more serious note, I think the idea of comparing power supplies has some merit. However, aside from measurements of ripple, noise, line rejection, etc. I doubt any useful comparison can be made from spec sheets. One really has to evaluate the supply in its intended application under realistic load conditions to compare supply A vs supply B.

~Tom

[kevinkr]

It might help to make the comparison a lot more general and perhaps anecdotal with the required caveats. You can discuss trade offs very generically IMHO..

Cost, weight, thermal considerations, regulation, noise spectrum, topologies, etc can all be generalized. Obviously this would be of more use to the newbie and less so for a seasoned design engineer.

[kevinkr]

Originally Posted by tomchr

However, if the cat is operated within its saturation limits, the temperature may be approximated by, temp = tan^-1(length).

<snip>
~Tom

Operation of the cat in parametric modes outside of the recommended saturation limits should be avoided particularly after refueling.. This mode of operation is not recommended and may potentially void the warranty on surrounding objects.

[PedroDaGr8]

See I disagree that it is impossible to test the power supplies independent of the system. This was a commonly held belief amongst computer enthusiast until about 4-5 years ago. The only way they would test a PSU is load up a system. Then along came the website JonnyGuru and eventually HardOCP followed (jonnyguru was a mod there). They showed you can actually get better information by loading the system, checking ripple, checking the waveforms, varying the load amounts, testing instantaneous performance (by switching in a 10A load) and checking the waveform deviations. It turns out all of these things correlate very well with system stability and overclockability. As a part of this, they found out a lot of PSUs were heavily over rating their output (it could put out 500W but the ripple rose to around 300mV on the 12V rail).

[qusp]

as above, one power supply may perform admirably with a particular dynamic load and very poorly with another. without specifying a particular load requirement, measurements are pretty useless and if you do, you have limited the usefulness of such a database to the users of that particular design. i just dont think you have thought this through