Computer Audio Asylum Music servers and other computer based digital audio technologies.

---

[ Follow Ups ] [ Post Followup ] Thread: [ Display  All  Email ] [ Computer Audio Asylum ]

---

[ Alert Moderator ] [ Edit ] [ Delete ]

---

If the caps work, fit them . . .

88.109.106.57

Posted by Ryelands on December 27, 2008 at 11:06:42

Some recent posts have discussed improving PC-audio sound quality by modifying or replacing motherboard power supplies. Proposals have ranged from lead-acid batteries large enough to boost Exide stock (Rick McM) through high-end linear designs for "satellite" players (John S) to adding enough capacitance to dim the lights (Alfred).

As I recently had encouraging results from trying a slightly different approach on a PC system, I thought I’d report what I found.

I'm NOT claiming that what I've done is better than the above, just that it's cheap, quick, reversible and IMHO surprisingly effective.

Stage one: the P4 12v plug of an ATX PSU has two 12-volt lines joined together and two ground lines ditto. A 4-pin 12v P4 extension cable was cut in half to enable the fitting of a small paddleboard fitted with ten 1,000uF, 16vw capacitors wired in parallel accorss the 12-volt and Gnd lines. See:

http://www.maplin.co.uk/Module.aspx?ModuleNo=34109

It cost under £10 to make the board and under an hour to do the work. The effect on sound quality, however, was (is) dramatic and repeatable on both the Biostar and a Gigabyte G31 boards recommended for cMP2. The extender makes it easy to do “before-and-after” comparisons.

Incidentally, it also makes it easy (ish) directly to measure the CPU’s current draw, something I’ve never seen reported (though I’m sure it’s done often enough). So far, I’ve measured it only briefly and only on the 65nm Biostar board.

I had assumed it would vary wildly but it was surprisingly stable: an E1200 processor drew 1.4 amps on power-up and during POST but, as the OS loaded, it dropped to and stayed at between 0.7 and 0.8 amps and rose again by about 100 milliamps when running cPlay (without upsampling).

I assume that most of the drop is down to cMP2’s lowered clock speeds and other optimisations.

The total power draw of a Biostar motherboard and PSU measured at normal clock speeds with a power meter in the wall socket suggested that BIOS-disabling devices such as the LAN and on-board audio saved at most a couple of watts out of a total of about 50.

Presumably, it’s the effect on OS bloat that matters here as the changes are certainly audible. Whatever, a full "hair shirt" cMP2 configuration lowered consumption to well under 40 watts.

In the light of what seem to be rather modest CPU currents, I did wonder if much less than 10,000 added uF would have been equally effective.

I also measured voltage ripple before and after fitting the mod. Though I had expected it to fall, it was unchanged at about 200 mV with a Corsair HX520 unit and about 400 mV with a (much more costly) Nesteq NA4501. (BTW, I read somewhere recently that Corsair and Antec units have the same circuitry - can ayone confirm this?)

I’m sure that the SQ improvement would be apparent on any system, not just a cMP2, though I would not (even dare to) comment on the Mac. Could I persuade any Linux or Windoze user with at least modest soldering skills to repeat the test?

Stage two involved adding a 1,000uF cap to each power line of the ATX-24 socket, i.e. one each to the two 12-volt lines, the four 3.3-volt lines and the five 5-volt lines but leaving the signal and -12 volt lines untouched. See:

http://www.maplin.co.uk/Module.aspx?ModuleNo=98854

This mod is much more time consuming to do and, for obvious reasons, must be carefully checked. Again, however, the effect on sound quality was way beyond what I’d dared hope for. (That’s as much as I’ll say as I find audio-review-speak toe-curling.)

I made some notes on how to make up the board, which was trickier than I expected. If anyone would like a copy, please e-mail me.

For some reason, I cannot upload images to the list so if anyone wants a pic of the paddleboards, again just e-mail me.

+++++

The effort persuades me of the three things. First, those who stress that computer audio SQ depends critically on power supply quality are, without a shadow of doubt, right.

Second, it is possible to make some worthwhile improvements at negligible cost by paying attention to basic electrical parameters.

Third, proposals by John S and others to by-pass SMPS technology altogether are almost certainly worth pursuing.

If you’re still reading, thanks for your interest.


Dave

This post is made possible by the generous support of people like you and our sponsors:

Sonic Craft

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---