Thoughts on ripple/noise level on 12V rail with PCI Sound Card?

[headfinoob]

I have an Intel D945GSEJT motherboard with an Onkyo SE-200PCI LTD sound card.
I have a Fluke 77 Series II DMM.  During playback I measure between 0.002 and 0.003Vrms AC on the 12V rail.  When idle I measure 0.002Vrms on the 12V rail.  The 5V rail measures 0.000Vrms at all times.  When the computer is powered on with the hard disk drive (SSD going to be replacing the HDD soon) disconnected I measure 0.001Vrms on the 12V rail.  The power supply rail itself measures 0.000Vrms at all times.
The Onkyo card to my knowledge uses an OSCON cap to filter noise from the PCI bus.
Does this AC measurement seem to be cause for alarm?  Honestly, I can't seem to hear any noise from the system.  Obviously, peak to peak ripple measurements may be much higher.
That brings me to my next point.  I am using an ATX SMPS, the Corsair VX450W (with fan removed), with measures 50mV peak to peak ripple.  The overall noise is rather low, but the 50mV comes from some rapid spikes and as I recall measured at full load.  In any case, with the fan removed, and at a fraction of load, I suspect the power supply to be pretty clean.
Would a sinusoidal waveform be easier to attenuate than rapid spikes?  I might switch to a power supply with smaller peak to peak ripple, (but it may have larger AC ripple) in order to help reduce noise.  Also, would spikes of say 25mV - 50mV peak to peak magnitude cause premature failure of the passive filtering caps or are they more robust than that? 

 

[headfinoob]

bump.  I am just tripping?  lol, there must be far noisier PCs out there...

 

[obobskivich]

I don't think you have a precise enough instrument to measure ripple. The ATX standard specifies it in mV, and you will never see absolute zero. I wouldn't worry about any sorts of early failures or anything like that, at least due to such low amounts of ripple. Regarding the power supply; ideally you want to load an ATX PSU at between 40 and 60%; anything higher or lower and the PSU will operate less efficiently and (potentially) create more ripple. In other words, if your computer demands around 200W, your power supply is properly sized. If your computer demands around 400W, your power supply could stand to be upgraded.
 
As far as "can this be heard" - it's very unlikely. It may be measurable with an Audio Precision device, but any "noise" coming from this is probably so far buried that you can't hear it. Here's a decent "overall" on the topic of noise in playback:
http://www.ethanwiner.com/audibility.html 
 
I do realize it's unrelated directly, but the concept is sound; if the noise created exists, but only at a very low level (say, -80 dB), you're unlikely to hear it. If you hear audible buzzing/humming/etc then you've got a problem. I don't think you can get "better" than what you have now (in other words, you have a very good sound card, and a decent power supply). 
 
 

 

[jasonlitka]

Two things:
 

• A few mV of ripple is nothing.  Try adding a high-end video card and firing up an intensive game.
• PCI slots are powered by 3.3V or 5V.  The +/- 12V rails are only used for cards with a JTAG, and even then only at a couple mA.

 

[headfinoob]

Quote:

Two things:
 

• A few mV of ripple is nothing.  Try adding a high-end video card and firing up an intensive game.
• PCI slots are powered by 3.3V or 5V.  The +/- 12V rails are only used for cards with a JTAG, and even then only at a couple mA.

 
Interesting, I didn't even realize PCI cards didn't use the 12V rail.  From what I've seen, many computer PSUs have most of the ripple on their 12V rail.  If I measure the 5V pin on the molex connector (there is a power connector soldered to the motherboard to provide power for molex devices) I get 0.000Vrms AC.  I assume there are a couple solder points were I can measure the 5V elsewhere.  Because the board is powered by a 4 pin ATX power connector (two 12V pins and two Ground pins), the 5V is coming from the DC-DC converters in the board.  In this thread, former head-fi.org member Thoppa measured took some measurements on my same board and concluded the DC-DC converters are adding some noise to the 12V rail.
http://www.head-fi.org/t/451608/onkyo-se200pci-ltd-or-asus-xonar-essence-st
 
I wonder if PCI-E sound cards use the 12V rail?
 
Quote:

I don't think you have a precise enough instrument to measure ripple. The ATX standard specifies it in mV, and you will never see absolute zero. I wouldn't worry about any sorts of early failures or anything like that, at least due to such low amounts of ripple. Regarding the power supply; ideally you want to load an ATX PSU at between 40 and 60%; anything higher or lower and the PSU will operate less efficiently and (potentially) create more ripple. In other words, if your computer demands around 200W, your power supply is properly sized. If your computer demands around 400W, your power supply could stand to be upgraded.
 
As far as "can this be heard" - it's very unlikely. It may be measurable with an Audio Precision device, but any "noise" coming from this is probably so far buried that you can't hear it. Here's a decent "overall" on the topic of noise in playback:
http://www.ethanwiner.com/audibility.html 
 
I do realize it's unrelated directly, but the concept is sound; if the noise created exists, but only at a very low level (say, -80 dB), you're unlikely to hear it. If you hear audible buzzing/humming/etc then you've got a problem. I don't think you can get "better" than what you have now (in other words, you have a very good sound card, and a decent power supply). 
 
 

 
Yeah, a scope should would be nice.  I've noticed there are some decent looking digital scopes that are becoming more affordable.  You are probably right that it is such a low amount of noise that it probably isn't worth worrying about.  I don't notice any buzzing or humming at all.  There is of course, a small amount of hiss that is generally only audible with my ear right up to the tweeter.  The amount of hiss seems to be less than that of any system I've owned previously, so that is a good sign.
 
 

 

[obobskivich]

The hiss is the amplifier<->speaker interaction, not the computer/source/magic. This is common across the board; even expensive equipment will do this - ideal equipment just does it less (less meaning you don't hear it over music or at your listening spot). 
 
As far as the power on the board - the board has a 20 or 24pin ATX primary that carries all three voltage rails into the set; the 12V aux is to provide additional current carrying capacity. 
 
 
 

 

[headfinoob]

 
Quote:

The hiss is the amplifier<->speaker interaction, not the computer/source/magic. This is common across the board; even expensive equipment will do this - ideal equipment just does it less (less meaning you don't hear it over music or at your listening spot). 
 
As far as the power on the board - the board has a 20 or 24pin ATX primary that carries all three voltage rails into the set; the 12V aux is to provide additional current carrying capacity. 
 
 
 

I agree regarding the hiss.  I am rather pleased that hiss in my current system is less than any other system I've had.
Note that my motherboard only uses either the 4-pin ATX connector (two 12V pins and two Ground pins) or a circular power connector for wall-warts and the like.  It is a low-power Atom-based board.  I suspect that the 5V is coming from DC-DC converters or maybe just a voltage divider?  In any case, my Fluke 77 Series II measures 0.000V AC on the 5V molex connector soldered to the board.
I am curious if there is a much of a difference between using the linear power connector vs the 4-pin ATX connector.
 

 

[obobskivich]

Your motherboard has DC-DC conversion onboard then - you are correct there. I'd personally go for a regulated SMPS over a cheap wall-wart, but that's me. 
 
Quote:

 

I agree regarding the hiss.  I am rather pleased that hiss in my current system is less than any other system I've had.
Note that my motherboard only uses either the 4-pin ATX connector (two 12V pins and two Ground pins) or a circular power connector for wall-warts and the like.  It is a low-power Atom-based board.  I suspect that the 5V is coming from DC-DC converters or maybe just a voltage divider?  In any case, my Fluke 77 Series II measures 0.000V AC on the 5V molex connector soldered to the board.
I am curious if there is a much of a difference between using the linear power connector vs the 4-pin ATX connector.
 

 

 

[headfinoob]

Quote:

Your motherboard has DC-DC conversion onboard then - you are correct there. I'd personally go for a regulated SMPS over a cheap wall-wart, but that's me. 
 


 

Currently I am using a Corsair VX450W with no fan.  I'm fairly pleased with it.  I'm just curious the difference between the linear power connector on the I/O panel of the board vs using the 4-pin ATX connector.  It would be interesting to see measurements of each when the power source was identical- I wonder if it uses different DC-DC converters or if circuit paths just go to the same one.
 
I've thought about doing what Thoppa did with his setup (see link above) and using a Sigma 11 with upgraded Toroid for increased current capability.  I've also thought about using some sort of lab power supply.  I've searched around for good 12.0V linear power supplies with low ripple and 3-amp current capability for low cost, but I am having some trouble.

 

[obobskivich]

A cheap plug-pack will be less consistent is my point. Should use the same converters, look at the board - do the traces go to the same place? From what I remember of xformer "stuff" toroids are less efficient/capable but potentially more linear. I do know that you don't need to worry too much about the type of transformer, as long as it can provide what you need within spec. I think the SMPS you have now is probably going to be as good as it gets, and you probably won't realize any sonic/performance benefits with a new PSU (you may have fun building something new, or make a more power efficient device though).