PCI supply rail noise

[PeterSt]

Since this apparently *does* have its high impacting influence, all we need (ahum) is getting it under control. Can it be done ? maybe ...

First base idea is that this card *creates* noise to mask the few profound noise in there. So : profound = sheer frequencies. Now, since I don't see much how to do this without actually doing the wrong thing (add noise ? we must be crazy), I'd like to think the other way around : try to eliminate the few noise which is profound now (okay, as my assumption is).  So :

What about turning it into a not-"switching" supply.

So what I am going to do now is making it a DC experience. Ok, attempt-to;
I will make a "pack" of 5 47000uF Mündorfs and put those instead of what's on there at this moment. Should fit in the PC.

Not knowing at all what the exact draw is per rail, they hopefully drain so slow that no resonance will be in order ever, and further the "spikes" of loading them will be at a so slow rate we won't notice. Next it should act as a sheer DC supply in parallel to the "switching" supply inherently in there. Lastly, I hope that this DC supply is faster than what's in there so it will be "profound" (overrule). The inherent noise should disappear in the caps themselves.

Such a setup should not be subject to resonance. Burn-in yes (I think), but this will stabilize; Once "the sound" is reached, it should stay the same (but could slowly change because of wear, but this time no different than any electronics with capacitors).


Before I am really going to do this, people may tell me whether there is ever more than 24V used in the PC. I don't think so. I ask this, because the card contains 80V caps for one one the rails and 35V for two of the others. This, while the Mündorfs I have are 25V only.
So people who know this by heart, please let me know !

Peter

[Nick]

Peter hi,
I like the conclusion in the last post that tuning of the passive supply filtering needs to be understood. I don’t know how much research the makers of the card already applied but it may be that our requirements are different to those tested. It could be a big undertaking though.
 
About resonance from a bypassed capacitor network, I certainly agree with your comments about the THD+N performance of 1 2 and 3 caps. I would add to this that when I think about using bypassed signal coupling caps in the past, more often than not the effect has been to fix one problem in the sound only to introduce another (usually smaller) problem elsewhere. You may remember I commented that the x-Sound cap values seem to be unusually chosen in an earlier post. The common start point for PSU bypassing values, is to select values between 1/100th  to 1/1000th of the larger cap as a sensible starting point to minimise resonance caused by combined caps. But it’s never perfect, and the X-Sound boards don’t seem to do this anyway for what ever reason. So I’m with you on the point that the X-Sound or any other bypass capacitor networks has the potential to introduce its own characteristic resonance noise.

I still think that the RC network of the card will lower the noise spectrum of the PCIe supply rail over the effective range determined by the card cap values. The response of the RC filter must be limited by the performance of caps (ESR and response frequency etc, which will also all change as the caps burn in) and card trace lengths. My assumption would be that over some frequency range the card will be effective in reducing noise already on the PCIe supply rails. However as you say the card may well present its own resonances back onto the rails. But if  the card is reasonably designed though, surely the resonance spectrum should be secondary to the filter effect ?

Thinking about the PCIe bus and what needs to happen to the supply rails, I think there is some evidence that transfer jitter as data moves over the bus (maybe as block transfer timing or even a bit timing level) has an impact on sound quality.

(The PCIe “jitter” idea might be evidenced by the effect that PCI latency BIOS setting has on SQ on older MoBos. I wonder also if one of the reasons that the Adnaco card was not as successful as hoped was that in the Adnaco system PCI data has to traverse something like 5 PCI to PCI bridges before it reaches the remote USB chipset of the Adnaco card)

My guess, and it is only a guess, is that there are there are probably two main requirements of the PCIe bus power rail filter. The first would be to reduce the noise on the supply rails across some [edit: frequency] range that is useful to audio replay (but key here is understanding the range ?). And the second might be to lower the effective impedance of the supply over a frequency range to improve transient response of the DC supply to support transient On / Off signalling of different devices on the bus.

Now I’m layering guess on guess here but also I think the DC noise filtering might need to be effective from a low frequency in the <10hz range as audio frequency power supply induced errors in digital data patens have a habit of cropping up in the sound via some digital domain coupling process within the signal after the DAC.

Don’t take this too seriously it’s just the ramblings of a non-engineer who probably knows "just enough to be dangerous"  .

At least we know one thing and that is that the x-Sound card shows that messing with the PCIe power rails has the potential for significant impact (?improvement?) on sound so it seems worth pursuing.

Best,

Nick.

PS might it be that whilst a test with 5 x 47000uf caps would result in very good low frequency rail smoothing, that the caps will probably have much too high an ESR at the higher PSU frequencies needed by buss components and this might result in the “Supper Bass” effect that we discussed when I tried large caps on the Adnaco card. Fingers crossed anyway.

[PeterSt]

Before I forget it myself ...

Day before yesterday (day 4) all went back more towards the first two days. But, pleasantly enough, not all the way. So, fairly refined highs with enough mid in them. I psychologically forced myself not to like it too much, afraid of the (expected) due change later.
And no, I did not mangle with the card yet.

I was not 100% satisfied because I seemed to notice less explicit interest in the music I played. Not sure it was me or whether it was the general consensus about Windows 8 (remember, I am trying this on Windows 8 only). Fact is that the phenomenon "mechanical sound" slipped through my mind a dozen times during the session.

Then yesterday;

After an initial start with an album which screamed from the speakers in typically W8 fashion (way too bright) and which I let last for into the 2nd track only, I ended up in a HDCD collection. And actually by coincidence the real take off of the day was a HDCD left over of the day before and I could clearly recognize the sound;

Maybe not really blanketed (again like the first days), but a sound of which at least I like to tear out more freshness from the speakers. Well, almost and just not.
The both albums I played were from Deep Purple, Who do we thing we are, and Machine Head after that. The former I find famous for its special emphasized highs (always doubting how processed they are) and the latter just because I wanted to go back in time.

Both sounded almost too normal to me and at least equal for quality, which I don't recognize from earlier playings of them. The clear point : they wouldn't go wrong. LP like. Far more interesting though is that I was playing at -15dB/-12dB levels while normally this would be -21dB or so for these albums, and still I didn't perceive it as loud at all. I measured, and was totally surprised I was playing at 80dBSPL only. This should be well over 90dBSPL at the set levels. ?

Fact is also that yesterday now a one dozen times it slipped through my mind that this was very close to W7, up to "what to go to W8 for ?". Still not, because this sounded sheer analog and people hunting for that ... why not.

One thing is most profound to me : All the problems I always and ever had with W8 being too dynamical in the sense of the difference between the softest and loudest passages being to large (and therefore never being able to set the correct volume) are out of the way, no matter in what fashion the card influences the sound.
A general thing I notice also (but it is in the same realm I think) is that W8 needs a louder playback to draw the sound out of the speaker plane (not have it too laid back). So this remains. The better stereo image from W8 also remains, but that too is in this same realm (as I think).

Tonight, for a change, I will try to play without the card and see what comes from that and how long I can last.

Peter

[PeterSt]

Nick,

Thank you for your helpful thoughts.
And hey, maybe we 're at the beginning of something new to learn, now knowing how a PC can be influenced for SQ largely and without much hassle (none actually, if only it could be under our control more - ... as I think).

I don't think this can be about "direct jitter" because no jitter is in order at this stage (in the PC). So it can only result as current spikes towards the DAC, in there finally influencing. Same with the Adnaco btw.

But to be honest, I cannot understand how the influence can be that large, starting off with the enormous difference Windows 8 implies (differences from all directions in the mean time). So maybe there is more, but then someone has to find out what it is first. I can't do it at this time ...

Thanks !
Peter

[PeterSt]

Referring to the Windows 8 topic again ...
I am sorry, but I removed the card and it won't get back in.
Maybe I should also say "sorry" for being the ever and always one debunking tweaks I didn't like from the start, and the only thing I can say is "didn't I say it right form the start" (pricky me).

Sadly it seemed to solve the Windows 8 issues (at least for me that seemed so), which I think it actually will for those perceiving too high dynamics from W8. But no matter how (which I quite don't know myself, see W8 topic), I have proven now that it is the wrong tweak to hunt for. So like I said it earlier "adding noise ? we must be crazy".
And so I was.

All the card does (in my system) is flattening out all what should be preserved inherently. Is something wrong, causing the necessity to flatten out ? find the source of that.
Well, don't say I didn't try ( Windows 8 ) and the final result is that I solved it. For myself that is and again : not that I know exactly how at this time (and time involved is crazy to do such a thing on your own).

My personal conclusions and judgements :

- The card will (!) take out harshness indeed;
- The sound from it will get blanketed and operates especially on the midrange;
- Net result can be "nice analog" sound without the snap I am used to;
- I think even I could get used to it, but once you take it out again ...
- If the card is needed, something else is wrong first (still somewhat pretentious this is);
- YMMV because the SQ result is not the same each day (but maybe it can after long burn in).

I hope I got it right somewhat.
Peter

[juanpmar]

Hi Peter,

I wonder if you have tested the Filter card in W7 and if so if you got the same conclusions than with W8

Regards,
Juan

[PeterSt]

Juan - No, I didn't.

[juanpmar]

Juan - No, I didn't.

And are you going to test the Filter card in W7 or you have already discard it?

Juan

[AlainGr]

Peter, the most important thing is that you demonstrated that this tweak could be unstable depending of the resonance and the "blanket" sound that was happening.

Even if I don't trust my ears (and have said it frequently enough), there was something "soft" with the card. I was not sure if it was for the better.

But it worthed the attempt for the price. And sometimes it helps resolve a problem, would it be simply because it indicates another direction for investigation.

Alain

[boleary]

Well Peter seems you got it more than somewhat right but it may be a bit too early for final conclusions!  

After reading your initial post regarding the card, I gave it another try Thursday morning before heading out on a 2 day work trip. The sound, for the 3 hours I listened, was shockingly fantastic. Voices were no longer shouting at me, rather they were refined and articulate. They emerged from the music in a wholly natural way, being neither recessed nor too forward. On my trip I kept reflecting on what I heard and it just seemed improbable; made me feel a little crazy like I had imagined it. Maybe it now sounded good to me cause you liked it and I was just hearing it "differently?" Couldn't wait to get home.

 Well I listened for about four hours yesterday, which included pulling the card for a while, and again the sound with the card was just great.

The biggest change with the card in is in the sound stage. When I first listened on Thursday I was listening to a Beethoven String Quartet. Without the card the SQ was very good but I was a little disappointed with the sound stage. It was a bit congested, though congested might be too strong a word: there was separation between the instruments but I couldn't quite point to where they were in space. The recording was the last disc in a ten box set produced by Philips. I was expexting better and blamed the recording because the "sound staging" here has been quite good, especially since I replaced three power cords. The stock cords to my subs were replaced with pangea ac-9's and the one to my amp was replaced with an AC-9 SE. Anyway, after putting the card in I was shocked: the sound stage completely opened up, inner detail from each instrument was everywhere, and I could suddenly see, if not touch, where each player sat. Rather than a blanketed sound it was a rich, lush sound that I had not heard before from my system.

I then played Heart of the Sunrise, Yes, no lack of slam here, and the vocals: awesome.

Conclusions:

**Assuming the card has finally burned in, if the sound goes bad again, I'll remove it forever. I just can't have it fucking with my head like this! If it has stabilized with the sound I'm now getting it will be a permanent fixture in my system (We, here, are unapologetic about using whatever shortcuts, tricks, special potions, home made room treatments for first and second points of reflection, etc. that improve sound quality for under a hundred bucks!)

**Remember how clear the sound was with 8e before attempts were made to take the positive aspects of W8 and put them in XXhighend with W7? To me when this happened the focus of the entire "audio band" shifted downward, and though the mid range and base may have improved, the higher frequencies have sounded rolled off. The female vocals just lost their sense of presence. However, this shift to a W8 sounding XXHighend seemed to be caused by a harshness in 8e that you purist types wanted to eliminate (You should have tried some fiber glass sound panels at the reflection points. Ha!   ) Anyway, everyone who has this card should give it a listen on W7 with 8e. With the settings below, I'm currently in audio bliss.

Okay, now with the release of the "9" series all bets are off. Time to download and do some testing.....Are we having fun yet?  


 

[PeterSt]

I just can't have it fucking with my head like this!

Haha. Now you know why I don't like tubes either.

[PeterSt]

And are you going to test the Filter card in W7 or you have already discard it?

That would need me to revert to W7 first.
And at this moment I don't intend to do that.
Maybe if you guys all make me crazy (thank you Brian ).

Peter

[AlainGr]

I am starting to believe that all depends on the pc and components. I did not feel anything special, but with the card, at some music passages, the details were not as outstanding as they normally are...

Alain

[GerardA]

Maybe remove all capacitors from th PC, no resonances, no blankets, no sound?

[CoenP]

Hi,

Due to a fierce flu I've not been listening to my system for a week. Today I was recovered enough and in the mood for some healing tunes, and switched the lot on.

Well, the first tunes were somewhat of a disappointment. Where was before last week's airy soundstage and musical fluency? The sound was way to rough and dry for what I've been accustomed too. The word irritating comes to mind.

I played a little with the 09v9b settings, but could not find a satisfying combination. With my reference 09-8e the same nuisances displayed. Then I remembered boleary's cardexperience and removed it.

Well, the card IS the culprit for this issue, now it sounds familiar again and finally I can listen to music for healing purposes again!

I have to say that when the card "worked" it worked very well here and was a worthwile addition. Maybe I will find a way to reinstall it in "working" order again... I'll let you know.
OTOH 9z9b is supposed to sound good without it anyway, so let's focus there first.

regards, Coen