|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
61.239.75.90
In Reply to: RE: Too many purely theoretical points to respond to. Of course CPU is fed via regulator - .... posted by carcass93 on September 20, 2014 at 12:02:23
...but practice.
Though in my practice there aremaybe some crucial differences.
> As for the rest of the post, pertaining to PSU - I'm sorry, but
> all of this is largely non-experiential.
Actually, it is experimental/empirical. Except I never bothered with a linear ATX Supply. Pico Supply linear (big lab supply) or SMPS + Filter. The second combo wins. By a large margin. Of course the SMPS was modded. The current Seasonic is not though.
> - feed CPU via switching PSU versus linear regulated,
CPU always has switchers. Those switchers do have limited rejection of the main switching PSU's Ripple, which is why it is important to have a clean output from the main SMPS, but there is no need to go linear.
> EVERYTHING makes a difference, so combination of quality linear
> PSU with power filter sounds the best - much better than switcher
> with power filter, which in turn is better than switcher by itself.
Without having tried really serious filters (and they have to be non-audiophile types, as nothing I know in the audiophile world has a low enough corner frequency) you may not know.
Also, saying "SMPS" is near meaningless, as is "Linear PSU. "
The differences in noise kicked back into the mains between different SMPS is as much as 80dB (that is 1:10,000) depending on the filterig employed. Some "Linear supplies" kick more noise (from rectifier ringing) into the mains as a well designed SMPS.
Equally noise levels from SMPS and linear supply outputs vary widely. It is fairly easy to make a SMPS with all noise components at -100dBV (that is 10uV or equivalent to the output noise from a DAC with over 17 Bit ENOB, which even today many struggle at).
By comparison, many commercial grade SMPS are designed for around 100mV peak-peak ripple (if we scale this to compare to the 10uV above this is around -40dBV or 1,000 times worse than a good SMPS) at the switching frequency, which or course is hopeless.
Often if you open up the case of the SMPS you will see empty positions on the Circuit Board that could hold extra filter capacitors and chokes and mains input RFI Filtering but are empty. Filling them up and maxing out the filtering the Circuit Board can hold helps tremendously.
Incidentally, the Seasonic ATX SMPS I picked does not have empty places on the PCB... And it is pretty good for noise on both sides (out and in), if loaded lightly (which it is in my system). It is not as good as the best I can do, but very good.
> PS: The second part of your post, describing your "Music & Video PC"
> - this most likely explains why we have so different impressions of
> what makes a difference, and what doesn't. Don't be offended, but
> from a "computer optimized for audio playback" perspective, ALL
> the choices made while building that PC are totally wrong.
I had prior systems in the same case that where by far more Audio Optimised, including double versions of windows. I cannot say that in my system, using my gear I found much differences from any of these optimisations, not to say there were no changes in SQ, but most were near what I'd call the "noise floor" (in other words I constantly have to worry if I am hearing something or not, even if I test fully sighted).
Stopping the noise from the PC going into the system (that is galvanic isolation of the USB connection and seriously blocking the PC noise getting into the mains are the two ones that have huge impact. Without these almost anything seems to affect SQ dramatically. Once they are in place most stuff is dramatically reduced, to me to a point where I file it under "don't care enough to bother".
Ciao T
At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to untolerable pain. Really, what more could we need?
Follow Ups:
"Stopping the noise from the PC going into the system (that is galvanic isolation of the USB connection and seriously blocking the PC noise getting into the mains are the two ones that have huge impact. Without these almost anything seems to affect SQ dramatically. Once they are in place most stuff is dramatically reduced, to me to a point where I file it under "don't care enough to bother"."
Nice. Your explanation appears reasoned and the information seems very useful.
I have just junked one; the long and stiff cables make excellent emitters and it is impossible to make up a neat or tidy system.
Hi,
"long and stiff cables make excellent emitters"
1) Stiffness has no effect on the Aerial action of a cable.
2) Length must be seen in respect to the wavelength of the suspect signal. For most off mains SMPS 50 - 200KHz is the switching frequency so 6 - 1.5Km - next to which all cables attached to an ATX supply are VERY SHORT. Even if we allow the 50th harmonic of 200KHz we are still talking 30m wavelength. I doubt your ATX supply came with 30m cables.
3) In order to emit an electrical field it is recommended not to run any emitting aerial closely parallel with a ground/earth conductor. In most ATX power supplies ground and power cables are bundled together, if not twisted together (which would be better and less stiff BTW). I doubt the one you had was an exception.
4) In order to emit any signal, regardless of conductor length, presence of not of ground/earth conductors and (cough, cough) stiffness, there has to be a signal of appreciable magnitude to start with.
Ciao T
At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to untolerable pain. Really, what more could we need?
Stiffness and length mean that the loom is a mess in any computer enclosure.
Emitter is a colourful way of describing inter power rail pickup/modulation and is no more than your own colourful exaggerations as illustrated by the nonsense on the website describing the products that you said you designed.
If you have money to burn, try using a modular power supply that has lots of smaller conductors in a flexible ribbon cable format. I've got a bunch of Corsair RM1000's powering some equipment and these are have flexible cables that are very easy to work with. BTW, there are lots of ground wires interleaved between the +12 wires. Need lots of wires to handle 83 amps at 12 volts per power supply.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
I have - Superflower fanless. There is now another one with all detachable cables.
The problem with these makes is:
They all use stuff wire looms and there is no way to shorten cables except to cut them.
I am making one up based on a Corsair accessory ATX cable but it is quite a job as all wires are orange.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: