PCI supply rail noise

[Nick]

Hi,

I'v experimented in the past on the power supplies and clocks of pci and pcie usb cards used to drive audio devices, oftern the results have been very posertive other times more "interesting". what is for sure is that pci and pcie power rail noise has a BIG effect on SQ and I think the periodicity (think buffer sizes) of the data being placed onto the PCIe bus also interacts with PCIe supply rial conditions to effect sound.

This card looked like a simple, fun way to experiment without having to mod my usb3 card. It's a passive bank of capacitors that bypasses the power 3.3 and 5v rails on the PCIe bus. The aim is to reduce rail noise for your sound critical PCIe cards. In my case i have a USB 3 card that connects to my NOS1 DAC, but it it should work for SPDIF cards or sound cards in other systems.

You just place the card into a vacant slot and that's it. The capacitor size choices look a bit odd but despite this IT WORKS..... and how !

Sound quality changes in my system comprise;

+ Wider deeper higher sound stage
+ Much more life like tone (voices strings etc)
+ Ability to track ver complex passages and not become confused (like a powerful amp or upgrading to a cartridge that track much better)
+ Bass goes down lower become more taught and very tuneful.

Quite a list I know, but the difference was heard in my system from the first note. This may spark conversations about "filters" but right now I think it moves the PC a big step forwards as a front end component. It might be that the cap values could be better tuned but card works.

I know Paul's card arrived recently and he may want to add more.

Available here and at cheap-fi money 

http://www.ebay.co.uk/itm/ApeXi-Power-Filter-Card-for-Computer-Audio-Enhancement-For-PCI-and-PCI-E-Slots-/251232340488?pt=AU_Components&hash=item3a7e9d4a08

Cheers,

Nick.

 

[PeterSt]

Hey Nick, nice ...

The capacitor size choices look a bit odd but despite this IT WORKS

I wouldn't mess with that, since it will be on purpose.

It's a passive bank of capacitors that bypasses the power 3.3 and 5v rails on the PCIe bus.

Bypasses ? Does it shunt the rails to ground or something ? (that looks like leading to a shortcut).
Wouldn't it be merely so that it's (the caps) just attached in parallel to the rails to smoothen the supply ?

Regards,
Peter

[Nick]

It's a passive bank of capacitors that bypasses the power 3.3 and 5v rails on the PCIe bus.

Bypasses ? Does it shunt the rails to ground or something ? (that looks like leading to a shortcut).
Wouldn't it be merely so that it's (the caps) just attached in parallel to the rails to smoothen the supply ?

Regards,
Peter

Peter hi,

Your right they are connected across the rails to smooth and reduce noise. Each 3.3 and 5 v rial (there are quite a few of them on the bus) has its own bank of caps.

Cheers,

Nick.

[Nick]

There was stock of four cards listed on the eBay link above when this post went up and now there are none left..... I wonder where they went so quickly .

'hope there might be some more feedback on their sound coming soon.

[Scroobius]

Nick called me a couple of weeks ago about this. Such a brilliant simple idea and cheap enough so why not give it a try. Mine arrived last night so I quickly installed it. I rebooted and straight away there was an improvement. Overnight and today it is even better so here goes this is what I emailed to Nick last night ................


Immediately the improvement is very clear in my system (to these ears anyway). You mentioned HF roll off - well there is no HF roll off at all that I can hear - none, zippo, zilch. So that is great news because I was concerned about that.

What is very clear though is a significant reduction in edge or grain that I really did not think was there to start with. I have never really been satisfied with the way any hi fi system reproduces massed strings and whilst my system has gradually improved in this respect with all the improvements made over the last couple of years I could hear that it was not quite where I would want it to be. Well that is the area where I hear a significant improvement. Recordings I have been listening to recently have just become sooo good to listen to.

But it gets better. Immediately after installing the PCi card the sound was more clearly etched cleaner and clearer. The image depth is better & there are more subtleties in the sound. The subtleties are so tiny I just have to wonder what tiny voltages in the signal can now be heard.

And then there is the bass - the very first impression is that it might be reduced - but it is not - now I can hear subtle bass notes and changes that I could not hear before.

This is a significant improvement well to my ears anyway. Others who do not know my system may not hear the difference I sure as hell can.

It sounds just like the card is doing what it says on the tin - reducing the noise floor even further. I do wonder however if a system has to be in tip top condition to hear this kind of improvement - probably a year ago I would not have heard this improvement at all.

Anyway I am a very happy bunny

PS - there is no graphics card on my mobo so I use a PCi graphics card maybe that contributes to why I hear such a difference.

[juanpmar]

If you guys, Paul and Nick, think that this is worthwhile to try this power filter card  I´m going to order one. 35€ is not too much money if the improvement in SQ is as you said, as soon as I get it I´ll let you know my opinion.

Thanks to both of you,

Juan

[GerardA]

I want one too!
But where to get it now there sold out?

[phantomax]

I ordered mine on march 5 and it took three weeks to arrive. Sadly when I placed it in the slot I broke a solder point from the DAC chip  just placed in the slot under. I didn't realize then so I went crazy looking for the cause of the rigth channel failure.
Happily I could fix it at last, but since then I haven't had much time for a careful listening session. That is why I don't want to be very conclusive but if there any improvement it isn't very apparent to me. Maybe the reason in my case is that, as I pointed in a earlier post from march 4, my soundcard has a capacitors section called by the manufacturer "PC noise killer " that seems to be intended to do the same function.

Maxi

[AlainGr]

Is the Sotm USB PCIe doing the same ?

[Nick]

Is the Sotm USB PCIe doing the same ?

Alain hi

There will be an overlap in what the sotm and x-sound cards do. To an extent the Sotm applies the same approach in that the supplies taken from the pcie bus are better built on the card from a noise point of view than on a standard usb card. These supplies are designed to supply the sotm card itself though and not to generally reduce noise on the pcie power rails.

The x-sound will have more of an effect on the pcie bus rails generally this may help reduce noise from video cards and other parts of the pc.

Nick

[PeterSt]

Can I stirr a little ?

The SOtM will try to provide a better supply by means of less noise, so next the USB transmission out of the same card will carry less noise. This is an explicit thing, like better regulators and/or better caps.

The card Nick/Paul ordered ADD noise to the whole PCI/PCI-e bus. I called that "smoothen" in my earlier post, but they add noise. And as you know from me, that isn't nesessarily a bad thing;
Now all attached to the bus will be fed with less profound (high) frequencies.


Let's say that the "proof" of how I think about the latter honestly, is that I ordered 2 ...


Peter

[Tore]

Hi!

Where can i buy this card?


Tore

[Scroobius]

After another night SQ has improved further. What strikes me probably above everything is just how tonally good instruments sound. Listening to an Arpeggione (I listened to it extensively before this mod so good for comparison) and the quality of the instrument shines. Now Bach Cello Suites the same thing and the tiny detail of the strings, better imaging smooth and clear.

Maybe there is noise but perhaps less spiky and more evenly distributed who knows - for sure it sounds good though.

P

[AlainGr]

Hi Nick, thanks for your very clear explanation I tried the link you provided, but it seems there are legal restrictions forbidding me to access this product...

Peter, my analogy may not be good, but would it be like "dithering" of some kind ? Bringing a better "diffusion" (distribution) of the noise spectrum ?

Alain

[PeterSt]

Maybe there is noise but perhaps less spiky and more evenly distributed who knows - for sure it sounds good though.

Paul, maybe you didn't get it;
Suppose your PCI bus draws such, that there's a (just making up something) 20MHz frequency, a 23.1MHz frequency and a 70MHz frequency hammering on everything. If you now add 20.1MHz, 20.2, 20.3, 20.4, etc. and fill all up up to 70MHz, there is no frequency anymore and it has become "flat noise".
The capacitors will sort of do this because of their (uncontrolled) frequency of loading.

No original peaks will drop because of this and the smoothening happens because the original frequencies can not be profound anymore.

But notice that this means of attacking noise is one big wild guess, because nobody tells how high the amplitudes of the original frequencies are and how high the added are. And you can bet that they are not evenly high by some sort of coincidence;
Now something else happens (has to happen to let it work out) and that is that the added noise must be higher in amplitude than the original. So, more noise really and it could be harmful theoretically. Higher noise but more random (more "real" noise).

To get the real picture (of what I mean and for what's *that* worth !), think about false harmonics (the THD figure); So, your NOS1 has a few and they are ~110dB down. Do we hear them ? possibly. How to solve it ? Add so much noise that the false harmonics vanish in the noise. And now they're really gone. Smart trick.

After another night SQ has improved further.

Yea ? I wouldn't count on that. Or better said : I wouldn't count on anything working really when that is the case. Thus, you are burning in capacitors and now the sound improves. Well, then it is only a matter of waiting until it degrades. And improves again, and degrades.
If you understand the theory (my theory) than you will agree that this is nothing to base your music reproduction on. You blame the washing machine on Monday morning, but actually a bank of caps is in a wild mood. And nicely resonate with eachother at an interval of a minute (slow judder).
But keep in mind, it is you who implied that described process is going on because sound improved. So which psychological way do you want to go ? haha

Mine won't burn in. I forbid them.

Peter


PS: Alain, -thus- Yes.