Linear power supply for pc

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Oh when cascading PSU's generaly you will have a on board SMPS for the main supply (say 3V3) then quite often small local LDO's are used for isolated derived voltages (such as a local 3V3 for a clock etc). AQ lot of new main SMPS's are now employing spread spectrum and multphase design topology. But as a technology the PC is not exactly new, I've been using one to do my job since 1985.
 
But as a technology the PC is not exactly new, I've been using one to do my job since 1985.


And so have i and that proves, what?

Otoh, 1985 was indeed a very different time and i believe audio systems at the time sounded in many ways much better than they sound today. Around that time British audio mag reviewers had an obsession with front panel LEDs. I think it was Jimmy Hughes who first started disabling such devices and claimed improved sound. What in '85 was a potential source of RF, today is among the audiophile's favourite voltage references, commended for it's low noise. The irony :)



It is, of course, speculation, but one explanation for the sonic preference for linear supplies is that switched mode regulators integrated into mother boards have negligible effect on RF pollution compared to off-board power supplies. Ground planes and the absence of long wires certainly do make a difference. The (anecdotal indeed) effect of sata filters points in the same direction.
 
What worries me about a class D is that it is transmitting noise into the envoironment, if it is a commercial piece of equipement it should be EMC compatable, the noise spectrum is already bad enought with all the wireless devices and mains borne noise as it is.
Either a linear or a SMPS designed correctly will do the job, but my belief having been involved with the design of lots of various systems over the years is that iif you are worried about PC noise then isolate the DAC stage galvonicly. Myslef I use squeezeboxes, and have a PC exclusivly for running squeezbox server, thats all it does, bare minimum running, players isolated either wirelessly or via an therenet connection.
My reference to PC's was that they are not new, the digital transmission side of things is way more stable than then, I run quite power hungry CAD systems and PC crashes are a rare event these days, whereas that has not always been the case. Many years of research and development have given us digital devices that are stable and just keep working.
I am all for reducing noise in any circuitry and have a big interest in said noise. One of the main concerns with commercial gear is the ever increasing demand for low cost, as consumers we have got use to reasonable prices, and the cost cutting has an effect of the designs, and compramises are made. I think with a PC with a realy well designed SMPS will be no different from one with a linear supply.
:)
 
Hi kp93300

Nothing particularly surprising in your findings. I know nothing about Atom boards but quite possibly you can take this a step further by disabling on-board switching regulators one by one and supplying your own regulated voltages. Are you able to identify the regulators and the corresponding circuits they supply?

Hi
I do not have the expertise to do this but i am planning to get a refurbished and cheap motherboard for this purpose !
may be i can learn a thing or two.
cheers
kp93300
 
That's why I'm using an 8086 CPU. And it sounds so much better at Normal mode (5Mhz) than at Turbo mode (8Mhz)! Switching is bad, M'Kay?!

I think it was Jimmy Hughes who first started disabling such devices and claimed improved sound. What in '85 was a potential source of RF, today is among the audiophile's favourite voltage references, commended for it's low noise.

Should give you a good indication of how nonsensical those claims were...
 
Last edited:
switching may be bad but it is basicly what digital is, so you design your boards and decoupling, power planes etc to cater for the high dI/dt switching circuitry requires. Provided the designs are with certain parameters using linear of switchmode does not affect the digital signals in anyway, though I have heard comments regarding digital such as passing through numerous servers can add low level distortion!!!! The poster refused to discuss this further.:D
 
...though I have heard comments regarding digital such as passing through numerous servers can add low level distortion!!!! The poster refused to discuss this further.:D

Yeah, we should design a routing protocol that favors audiophile servers running on linear supplies and SSDs. :p

The most .... absurd thing I've read (and "absurd" is me trying - really hard - to keep it civil) is that files play worse the deeper they are in the directory structure. So if, say, you reproduce a file directly from D: it'll play better compare to the same file in D:\folder1\folder2\music.

I know, I know... I feel dumb even reproducing this stuff and I'm afraid too many people (i.e. more than 0) reading this will go ahead and try it, but I'm sure many more will have a good laugh.

If any of you don't believe someone would seriously write that stuff in public, I can name and shame. :D
 
Last edited:
I think I already know who it is, even though i've not read it before.

I have named these anti-bits 'Dark Bits' the ones that can still corrupt the sound of bit-perfect data, just data, no timing information, just data, bit perfect, yet immeasurably corrupt data.... as per usual the audible effects are night and day, fatiguing usually....

these bits are there to stay too, it doesnt matter the check-sum is fine, once corrupted like this its all over, it gets worse the more times its copied too, the more servers it passes through. nevermind that on SSD especially its spread over multiple locations and moved around constantly as VM has its way.

it does seem to be reduced if less processes are running on the computer when playing it though and even more if a linear or even better, battery supply is used
 
The digital audio equivilent of the Higson particle...
I like the 'Dark Bits' name, it conjours up images of an anti audiophile demon, the Darth Vader of digital audio design, and to quote said gentleman...
"I find your lack of faith disturbing.":p
and as I am in a rather silly mood, another:
"I sense something, a presence I've not felt since.......I swapped my linear supply for a dark SMPS"
 
Last edited:
This thread is becoming entertaining I always like how that evil RF can find its way into the best audio systems and corrupt them. Funny how I can stand next to a waveguide with over a MW inside yet my mobile phone with uV sensitivity to similar frequencies works perfectly.
That evil RF must sense when audiophile systems are being operated before working it's evil.
 
;) yes indeed marce those were the sort of images I had in mind too... like the Higgs, perhaps what is needed to uncover this phenomenon in the light of day is some sort of massive underground electrical super-conductor-collider? constructed of Ex Russian Military Surplus of amorphous alloy, powered with batteries.... or the existing accelerator and satellite network they used re-tooled for tracking data/bits?
 
Last edited:
Yet another thread has started up regarding the evils of digital, boy am I glad that:
1> My system is unable to resolve detaill in the music to a great extent, so I cant here how bad my PC-Squeezbox-amp-speakers sounds.
2> I am probably 90% deaf (or daft) as I cant realy here the digital distortion (funny though I can here the imperfections when I play my records!)
3> I am personaly immune from all the RF noise in the world that ruins the music as I wear underpants hand crafted from aluminium foil, this has been my best and most effective mod yet to combat the attack of the 'Dark Bits'. i find the boxer short pattern with no flap superior to the y-front design, I believe the little slot lets some dark bits sneak in.
4> I tend to listen to the music and not concentrate on the noise, RF and the missis nagging.
 
alluminium? marce I had given you more credit than that, surely copper foil gauss would be more effective at these frequencies?

however as opposed to the alloy, copper foil pants need changing more often from the oxidisation
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.