[headfinoob]

Hello,
I am seeking some advice regarding filtering the power supply of my audio computer.  The audio computer is low-power, fan-less, and Intel Atom-based.  The computer is controlled via Remote Desktop, VNC, etc.
I am using the linear power connector on the back of the PC (5.5mm OD, 2.5mm ID).  The power supply is external (it is the black box with green writing as seen in the picture).  The power supply is the Corsair VX450W.  The power supply is "modded" by having removed all the wires with the exception of the 8-pin ATX power connection, to which I spliced one of the 12V and ground pins to the 5.5mm power plug to plug into the computer.  The fan has been removed from the power supply. 
Because I am using just a 12V and ground pin to power the computer, I do have some flexibility in how the computer is powered.  I would even consider a battery power solution if it were affordable, however, because lead-acid batteries are 12.7V or so, I think I will skip it.
 
Disclaimer:  I don't actually have any obviously perceivable noise problems with my setup.  As with every setup I have heard, I get small amounts of hiss out of the tweeters.  The hiss is rather low.  One thing I considered trying is floating the ground pin on the Corsair power supply.  In any case, I like the idea of providing very clean power to the computer anyway.
 
Note that the computer is using only an SSD hard drive for storage.  With my Fluke 77 Series II DMM, I measure 0.000Vac RMS on the 12V and 5V on the motherboard's Molex connector.  When using a conventional spinning HDD, I measured between 0.002V and 0.003V on 12V Molex pin.  The 5V pin measured 0.000V with the conventional hard drive.
 
Also note that the motherboard must use DC-DC converters to get 5V because the only voltage input to the motherboard is 12V.  One thing I am uncertain of is whether DC-DC converters are used for 12V since it is provided.  Another thing I am curious about is whether the sound card, being PCI, even uses 12V at all.  One head-fi member suggested that my PCI card only uses 5V power from the motherboard.
 
 
Here are some gut-shots of the computer:

 

 
Here is my setup:  the power supply is chillin' on the dresser to the left of the speaker.

 
Here is the back of the computer and receiver:  The power connector can be seen near the VGA connector.

 
 
 
Sound Card:
Onkyo SE-200PCI LTD  -  http://www.jp.onkyo.com/wavio/se_200pci_ltd/index.htm
Motherboard:
Intel D945GSEJT  -  http://www.intel.com/content/www/us/en/motherboards/desktop-motherboards/desktop-board-d945gsejt.html
Power Supply:
Corsair VX450W  -  http://www.corsair.com/vx450w.html
Power Plug (in case anybody is interested):
http://shop.willyselectronics.com/browse.cfm/dc-power-cord-2.5-x-5.5mm-r-a-plug-to-bare-leads-6ft-18awg-/4,8031.html
 
 
Here are a couple of places with measurements of ripple on the Corsair VX450W. 
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story2&reid=64
http://translate.google.com/translate?sl=auto&tl=en&js=n&prev=_t&hl=en&ie=UTF-8&layout=2&eotf=1&u=http%3A%2F%2Fwww.clubedohardware.com.br%2Fartigos%2FTeste-da-Fonte-de-Alimentacao-Corsair-VX450W%2F1478%2F7
It appears that the Corsair generally has about 29mV peak to peak of ripple under heavy load on the 12V rail.  Counting the number of peaks and time-steps, I would expect noise in the 3600Hz region.  Interestingly, the German site's measurements would indicate noise in the 1075Hz region.  Using RTA, the only noticeable anomaly is a spike at 5030Hz when holding the mic up to the tweeter with the amplifier volume control turned up.
 
 
Here is some discussion regarding some capacitors that could be used to build a filter this motherboard.  Thoppa, former head-fi member, graciously provided his recommendation for caps to use:
http://www.head-fi.org/t/451608/onkyo-se200pci-ltd-or-asus-xonar-essence-st/90#post_6401113
To be honest, I'm not sure how he expects each capacitor to be hooked up, but I would assume all the capacitors would be in parallel, with the positive legs of the capacitors wired to +12V and ground legs wired to the GRD wire of the power supply.  I would assume that the resulting power filter would be wired as close to the motherboards connector as possible to reduce wire length between the filter and the motherboard.  Presumably, I would solder each of the caps to a PCB and solder the power plug to the PCB as well.
 
My search for an affordable, quality, low noise (audible hum), low-ripple, 12V linear power supply that can provide 3 amps of current was seemingly fruitless.  In reality, 3 amps is probably overkill for this computer, but I would like 3amps of headroom anyway.  A power filter would be nice, but sourcing capacitors and building a filter isn't something I really wanted to do as my soldering skill level is low.  If another member is interested in building something like this in exchange for compensation, PM me.
 
Any advice or information is appreciated, thanks.
 

 

[Skidood]

Putting gigantic capacitors on the 12V wires may give your power supply some grief..when initially turned on, the inrush current to the caps while they start charging up may trigger your PS to go into short-circuit protection.
 
If your motherboard uses a 12V>5V converter, this may be the source of the hiss.
 
Also, have you shorted the inputs on the receiver to see if the hiss disappears?
 
 

 

[joe_cool]

I'm confused, what is the amp for your speakers?

 

[headfinoob]

I am certain that my Corsair HX1000W (960W or more on 12V rails combined) can charge a discharged 1Farad capacitor without going into protection.  Therefore, I assume that my Corsair VX450W (396W on 12V rail) should be able to charge a capacitor with a value of 0.4Farads without going into protection.
Does anybody know if using a giant capacitor (0.5 to 1Farad or so) would be a problem when powering the computer?  I like the idea of the computer receiving the short bursts of current it requests and I'm wondering a giant capacitor would slow the PSU response to where the computer would not be receiving the short transients of power it demands.
 
I wanted to get an idea of the actual noise the computer was adding to the system.  Here are some interesting results from FFT using my Samsung Droid Charge with Audia Sound's FFT app.  The results are using FFT resolution of 16K with infinite averaging over a period of about 2 minutes each.  The phone's mic is placed 1cm from the tweeter in each measurement.  The receiver's volume knob was at full maximum or minimum for the measurements.
The conclusions I have come up with:
1.  The CD Input has the lowest noise of all the inputs (With RCAs from computer connected with the computer off-- this result may change if I disconnect the RCAs)
2.  The AUX Input is the worst input, with far more noise than the other inputs-- nasty.
3.  The computer adds some noise to the CD Input, but very little.  Some noise shows up between 10kHz and 20kHz as well.  One can see the difference between the CD Input with the Volume maxed with the computer on and off.
 
Looking at the noise levels on the CD Input with the computer on and off makes me wonder if it is really even worth the trouble at all to filter the computer's power supply.  Does anybody have an opinion of the noise added?  Remember the computer is providing 2.0Vrms to the receiver.  With any driver on Windows XP or 7, the sound card preout volume is always maxed out.  With the receiver volume at maximum, that would be some serious output.  Most of my listening occurs with barely 1/10 of a turn of the volume knob.
 
Room Sound:

 
CD Input - Minimum Volume:

 
CD Input - Maximum Volume - Computer OFF:

 
CD Input - Maximum Volume - Computer ON:

 
Tape Input - Maximum Volume:

 
Video Input - Maximum Volume:

 
USB Input - Maximum Volume:

 
AUX Input - Maximum Volume:

 
 
 

 

[headfinoob]

Quote:

I'm confused, what is the amp for your speakers?

The receiver is the Outlaw RR2150.  I didn't really make mention of it because I figured any decent receiver would be somewhat similar in that it would amplify the noise coming off the sound card just the same.

 

[Skidood]

Quote:

I am certain that my Corsair HX1000W (960W or more on 12V rails combined) can charge a discharged 1Farad capacitor without going into protection.  Therefore, I assume that my Corsair VX450W (396W on 12V rail) should be able to charge a capacitor with a value of 0.4Farads without going into protection.
Does anybody know if using a giant capacitor (0.5 to 1Farad or so) would be a problem when powering the computer?  I like the idea of the computer receiving the short bursts of current it requests and I'm wondering a giant capacitor would slow the PSU response to where the computer would not be receiving the short transients of power it demands.
 
 

You have it backwards...adding the cap will only improve the stability of the supply voltage during your hi-demand transients.  The cap would be connected across the 12V and ground leads, not in series, so it's charge is always on tap.
My question is, how much transient demand current do you think your sound card is going to need?  That fancy PSU has 1000 times the capacity your motherboard and sound card will ever need in terms of ability to handle transient spikes in demand. In terms of noise reduction on the 12V line, the cap should reduce the noise on the 12V line, but wont eliminate the noise 100%.
I feel that adding the caps will not reduce the hiss that much as I dont think the root cause is the noise on the 12V...the motherboard probably has some filtering caps on it already.....you can temporarily connect a large cap from your local car audio store to see what happens.