Why does the computer matter?

[Part-Time Audiophile]

Seriously. Logically, there's only a couple of places where a computer really ought to have an impact on the end (sonic) result.

 

#1, better/different software.

#2, better/different hardware.

#3, better/different layout of hardware.

#4, more/less noise.

#5, more/less jitter.

 

The role #1 is tendentious as it probably has impact either on jitter or on some kind of alteration of the signal (think: volume or some kind of DSP-like effect). I'm going to assume that software also varies between platforms (Mac vs Windows vs Linux), but shouldn't vary within a platform.

 

#2 would matter more if the computer is doing much in the way of processing, but as a server that is sending content to a DAC, perhaps via some kind of off-board interface like an EVO, then it's unclear what better or lesser parts would have to do with anything. To some extent, the bits leave or they don't. But better connectors, cleaner traces, more reliable power, all could theoretically enhance the signal leaving the computer and making the DAC need to work much less hard.

 

#3 layout is the primary difference between a laptop and a tower case -- the parts are more or less the same, though it could be argued that the parts quality is higher on one, they're just laid out differently. This difference might account for #4 and/or #5.

 

I'm going to make #4 different from #5 in the following way -- noise is EMI/RFI that generates errant or corrupted data. It's logically feasible that a noisy environment might corrupt the outbound data stream in such a way that degrades it. Might flip some bits or degrade or attenuate the signal to create dropouts, hangups, pops, what have you. This might be ameliorated or exacerbated wrt #3 or even #2.

 

#5 is everyone's favorite digital bugaboo. I'm taking the liberty of simplifying jitter here to mean irregularly timed output.

 

Does this make sense? Am I missing something? I'm just trying work this through.

 

[John]

I guess, at least theoretically, you could add vibration (transmitted to other audio equipment through stands or connecting cables). This could be one of factors behind the reported differences between regular hd and ssd.

 

 

[barrows]

I would love to hear some answers to these questions that made some sense. Here are my musings on the subject:

Jitter, is not really relevant in the computer, or caused by the computer, unless one is using a non-async style interface. The async interfaces (async USB, Firewire, ethernet) all eliminate the computer as a source of jitter at the DAC (the only place it matters).

The only thing that makes sense to me as far as the computer being relevant to sound is RF interference generated in the computer getting into the analog signal path. This can happen through the connection from computer to DAC, from the connection from computer to AC power (if any), and from airborne RF broadcast from the computer and the various cables (acting as antennas) attached to it. RF problems makes sense, and might explain why a lot of the computer tweaks (which usually lower the amount of processes going on) could improve sonics: less processing equals lower power consumption and lower RF generation. Other than RF, I am at a loss as to how the computer can affect sonics (I am assuming here that we have a properly set up rig, outputting a bit perfect data stream).

 

[ItemAudio]

The two questions have much in common.

 

Assuming we've satisfied the 'bits are bits' brigade and that the same zeros and ones in the recording arrive at the DAC (not a foregone conclusion), we're left with jitter and noise.

 

Electrical noise and cabling modify jitter characteristics: both factors inter-relate. Even mechanical vibration (theoretically) matters - although its relevance to computer audio seems much less than when we we span discs.

 

It's now abundantly clear that system resource management impacts on first clock input: so the clearer the lanes of processor and bus traffic, the better the performance. Hence the 'different sound' of operating systems, playback software. Less error correction is good; avoiding on-the-fly decompression and unnecessary active processes can only be good, too. Almost everything a computer does affects the purity of its digital output in this regard.

 

The crucial takeaway is this: Wherever the master clock in your system (reclocker and/or DAC), it is sensitive to input.

 

With regard to conducted noise: the computer is just another device connected to an amplifier with really good conductors. Remember what an amplifier's job is...

 

So all power noise propagates at near-light speed throughout the system: oscillating from the mains to the speakers and back again - including the computer's power supply and all the switching processes and multi-frequency noise on the motherboard. Not only that, but you've connected computer peripherals!! The monitor has its own very nasty SMPS; the hard drive is spewing noise into the SATA and PATA ports: you may even have optical and external drives with their own switching power supplies - all beating together to create noise.

 

Then there's radiated interference: all those USB leads and power leads and SATA cables and internal wires are potentially aerials: picking up and conducting the noise broadcast by all of the above (especially the hard drive) back, ultimately, into your DAC and amplifier.

 

It's mechanically easier for a powerful computer to unspool data from a little spinning hard drive than for a relatively dumb CD player to get it from a big spinning optical drive, but that doesn't mean that a computer is better than a transport: it presents a whole new set of challenges to the system in terms of noise and jitter. Ignoring these challenges - treating an amplifier like a printer - will result in that distinctive 'computer sound': hashed over with noise and jittery as heck.

 

[PeterSt]

Scot,

 

As long as you think the base it that bits stay not the same when SQ differs, the whole topic is moot. Ok, except for the jitter stuff. So :

 

Assuming we've satisfied the 'bits are bits' brigade and that the same zeros and ones in the recording arrive at the DAC (not a foregone conclusion), we're left with jitter and noise.

 

This *is* the good assumption (IMHetc.O).

Funnily enough all your #1-#5 can stay, but keep in mind that all goes (or at least can go) with major SQ differences, without the output being one bit different from eachother.

This last - but certainly not least - includes the (playback) software.

 

To (for example) tempt you a bit on #2 :

Nothing is needed to let the bits go out as should (they just always do), but *everything* matters. Whether this is jitter impeeded afterall or ...

 

leaving the computer and making the DAC need to work much less hard.

 

that one, is for some future geeks to find out. I can't be sure what you exactly mean by this, but it sure would be my subject. Haha.

 

Peter

 

 

 

 

[bikemig]

So how do you minimize the computer generated noises from your music system? I've read the threads on power supplies and SSDs and so forth; I was thinking more of the radiated interference you were writing about. Is there a way to minimize or defeat that sort of noise? Does this line of thought suggest that streaming music is likely to solve some of these problems (while perhaps introducing problems of its own?).

 

[PeterSt]

I was thinking more of the radiated interference you were writing about. Is there a way to minimize or defeat that sort of noise?

 

Could be mu-metal when applied correctly; Difficult for a "motherboard" environment.

 

Although many people shout about this, it is my opinion that this is more "nasty theory" than practice. I mean, it is not difficult to measure this radiation by itself, but I never saw it impact on electrical wires etc., nor did I perceive it at listening, nor will it be measured at measuring the end product (test signal). This doesn't proove all of course, and it will depend on how strong the radiation is.

 

What *will* impact all over the place is a good grounding scheme. But here too ... don't ask me what to do in the motherboard environment. Yeah, galvanically isolate completely.

 

2c

Peter

 

[PeterSt]

Does this line of thought suggest that streaming music is likely to solve some of these problems

 

Hey, you added that later ... :-)

 

Yes. Assumed that it's over Ethernet, yes. This is inherently galvanically isolated.

 

Peter

 

[bikemig]

Peter: thanks for the replies to my earlier and later questions, . I've never had any problem with computer generated noise when running a simple set-up (computer to powered speakers) but I do in my multi-media set up. The combination of the TV monitor, cable, computer, and a few other odds and ends has given me a few head-aches with computer generated noise. Miguel

 

[PeterSt]

Hey Miguel,

 

If I may ask ... what is it what you notice ?

(the obvious answer will be "noise", but I don't think noise as how we talk about it is audible as noise as such)

 

So, similarly, what do you perceive with the MM setup which is ok with the music-only setup ?

And what is connected additionally to the MM setup ?

(it may include a pre-amp, totally different main amps, special volume controls, etc.)

 

Peter

 

(I hope this is not too much offtopic ?)

 

[bikemig]

Peter, I have a really simple MM system that doesn't work, . I have a TV and a mac mini; I use Audioengine A5s with a sub for the sound. The mac works fine both to play/stream movies and for music (ripped in apple lossless); the problem is with the sound out from the TV to the A5s. This introduces hum which I realize may be grounding but I think it is computer related. The sound becomes worse, for example, when running certain programs such as a web browser. I also suspect that a sound system with 2 active line-ins (such as the Audio-engine A5s) may have problems with sound bleeding in from different parts of the system. Eventually I plan on getting a dac/preamp so I can do digital out from the TV and have a short analog run from the dac/preamp to the speakers. Sometimes I think I did too much reading on this issue and now I'm just hopelessly confused since I have no background in this area, . Miguel

 

[4est]

I have given up on thinking that 1s and 0s are solely "digital", and are not effected by the real world (read analog) in ways that we either do not understand or cannot measure w/o interfering. If you stop to think about it, a perfect square wave cannot exist, let alone thousands per second. Bits are not bits, because the waveform shape is not the same, and never will be.

 

Put another way, for me I think it is foolish to think that we are capable of fully understanding anything w/o some speculation and its subsequent errors. Especially in a system where there are millions of processes and billions of electrons in motion.

 

Frankly, I am amazed computers work in the first place considering all that can go wrong.

 

[PeterSt]

Hi Miguel - Yes, that is what I a kind of meant : not noise as such, but a grounding problem (thus hum -> can be noise as well, but merely symphonic noice (or how to call it, but it's sizzling and ticking a bit)).

 

You can go two ways here, which either not necessarily work :

 

Disconnect grounds everywhere (begin with the P.E. (Protective Earth to the mains), and attach additional wires in a star to all of the cabinets (including the PC's).

 

or

 

Connect as much to ground what you can think of, which may include connecting the cabinets to P.E., may they not do that themselves (which you often won't know, never mind the mains cable has an earth connection).

 

This doesn't necessarily solve it, but next steps normally are not for "consumers", so to say. What you could try though, is disconnect your DAC from everything (not only switch it off), and "proove" it is the DAC doing it. Most often this is the culprit, implied by "strange" or wrong grounding in the first place. If indeed your DAC is the device which introduces the hum, you may think about bringing it to a good engineer, who will decouple all the "active boards" from ground (think capacitors here), but couple it as soon as possible in the mean time (the engineer will understand).

 

Generally, without trying to teach some lessons, try to think that everything which uses power, needs a "ground reference". Also, all devices using power will have a different "potential", which in layman's terms comes down to the ground being not the same. That by itself causes a current to flow from one device to the other, but it travels through your interlinks. Here is your hum. So, be ahead of that, and shortcut it.

Any ground reference will seek the shortest path to some real ground. This is why a separate cable from the cabinet to P.E. may help; it goes through there, and not anymore over your interlink.

Connecting all cabinets is similar, but now you are creating a kind of virtual ground, hoping for all devices in there seeing the same ground, and thus no travel to unathorized paths (interlinks) is needed.

 

Sadly, within any device the same applies - and this is where you can't interfere. So, it is very easy for a device to pick up transformer hum if the internal grounding isn't arranged for properly.

Also, if e.g. an interlink is connected to the chassis right at entering the cabinet, the current flow stops there (doesn't reach the amplifier etc.), if only the cabinet is connected to P.E.

 

Please notice : I could blahblah somewhat more, but generally ground(loop) problems are tough to solve. For everybody. One thing must be clear though : this isn't going to be solved by shielding (etc.) (in) the PC. You *may* solve it my isolating the PC from the rest (like toslink would do), but that current flowing now, will seek its way to elsewhere. It may go in the good direction then, but it will be coincidence;

The problem is and remains : no proper grounding somewhere.

 

One more thing : don't underestimate what current flowing to/through P.E. itself may incur for. This is why the option of "not grounding at all to P.E." is there). I mean, once you ground a device to P.E., another may nicely pick of up from your own mains. This is also where separate "mains groups" come into play (all with their own earth pin), and you can connect the audio to a different group as the TV etc. is. Or the PC.

 

So far for now. :-)

Peter

 

[4est]

You may have already tried this, but Cable TV is notorious for causing hum. Try disconnecting the CATV cables. They make isolators for this.

 

[Al Jones]

No answers here, but just to complicate matters further...

 

Until recently, my music was on a ReadyNas Pro Pioneer Edition wired via 5 meter ethernet to a Linksys WRT350N router, then wireless to an Asus EB1012 mini-PC (for the rest, see signature). I was more than satisfied with the result.

 

A few days ago, I bought a Seagate Freeagent GoFlex Desk 2T external harddisk for data storage, but before setting it up, I decided to see what would happen if I transfered all my music (1T) to the Seagate and parked it next to the Asus conected via the no-name USB 3.0 cable that came with the Seagate.

 

I was shocked by what happened: improved clarity, detail, more extended, fuller bass, enhanced dimensionality. When the wife came home, she could hear the difference all the way out in the hall and asked how many $kilo I had spent this time. When I told her it was barely $200, she didn't believe me.

 

Obviously, too many variables have changed all at once to draw any usful conclusions about the source of the substantial improvement - Is it the new storage device, bypassing ethernet, bypassing the router,bypassing wireless, using USB 3.0, some combination of the above?

 

I don't really care, I'm just enjoying the cheapest audio upgrade I ever made!

 

I can't wait to hear what happens when I replace the GoFlex wall wart with a laboratory power supply.

 

Al Jones

 

 

[barrows]

Here is how I set up my system to try and minimize RF/noise from the analog signal path.

 

1. 4 shelf rack: top shelf is a Symposium Svelte Shelf (stainless steel two layers with firm foam between) on which sits the laptop and a Oyen Mimipro 1 TB drive. Drive is powered over Firewire (this eliminates a switching wall wart) and laptop is powered from internal batteries. the stainlees steel of the Svelte Shelf provides some sheilding of the computer and hard drive from the rest of the system below.

2. 2nd Shelf: DAC and Wavelink on a 1/2 thick acrylic shelf. DAC has a DIY power cable featuring "special", (secret), shielding and damping.

3. 3rd shelf: Preamp on 3/4" acrylic shelf. Preamp power cable is DIY with no shielding (sounds better); this cable is routed so that it is as far away from computer stuff as possible.

4. Bottom shelf: empty, mass loaded with a big piece of sandstone-decoupled (from rack frame) with sorbothane discs. All rack shelves also decoupled with sorbothane discs.

5. Floor: amplifier on a Finite Elemente stand, unshielded power cable, routed away from all other components.

6. Floor: P-300 AC regenerator, on custom 3/4" acrylic stand, Shielded PS Audio "Lab" cable, re-terminated with Furutech connectors.

The computer stuff in my system definitely sounds better when running from the computer's battery supply. By running the external drive off of the same supply I avoid cabling and a SMPS that could create more noise. All components are physically separated in space, and decoupled from each other in terms of vibration, by the rack arrangement.

I plan to add a turntable and phono stage in the future, and will probably use a wall mount, to try and keep my level of isolation (power and vibration) intact. I really think the Svelte Shelf is a great unit for the computer stuff, in that it provides some shielding, and also acts as a defacto heatsink for the laptop as the top stainless plate draws heat away. My computer is set to shut down the display. When I (finally) purchase a dedicated laptop (MacBook Pro, 13") I will install an SSD to further lower the potential noise and power consumption. I also place Shakti OnLines on the Firewire cable out of the external drive, and at the ports on the laptop-theoretically these should damp RF travelling on these cables as well (although they are of questionable audible impact).

 

[Vincent Borrelli]

Peter, thank you for your excellent description of how various grounding approaches can effect noise. I had a question for you (and Barrows): Should the computer's power line be on the same circuit as the stereo?

 

I currently have a dedicated 20a circuit for my system, with power fed through a Synergistic Research PowerCell. All of my connections are "actively shielded" Synergistic IC cables and power cords, which should reject any RFI/EMI emitted from the from the computer and in the environment. The computer source (stripped G5, Lynx AES16, SSD, no monitor, NAS over ethernet) is ALSO plugged into the same power source. This is what Synergistic recommends, but I question if noise from the G5 may be corrupting the power from inside the circuit? Would it be better to power the computer on a different (non-dedicated) circuit to separate it? Or, does this create other problems due to grounding paths for two circuits?

 

The speakers are on Sistrum SP-101 Platform stands (which are amazing!), and each component (except the G5) is on a Sistrum SP-1 Stand, which takes care of resonances very well. The PowerCell is on Finite Elemente Ceraballs and 1" Plexiglass.

 

Thank you for your advice!

Vincent

 

[barrows]

Plugging the computer into a different circuit may cause ground loops, as you suspect. Different circuits will usually present different impedances to ground, resulting in currents running on the grounds. When possible I always suggest to power all components of the audio system from a single, dedicated, circuit as this approach results in a single ground path. I am not familiar with the internal circuitry of the Synergistic Powercell, so it is hard to comment on what the impact of plugging the computer into it might be. Generally, I recommned powering the computer and peripherals from its own power conditioner, to provide as much isolation from the power supply of the other components as possible. This power conditioner need not be an expensive "audiophile" product; in fact, some of the cheaper home theatre conditioners can be great for isolating computers from the audio system, especially on their "video" circuits. APC, Moster, and PS Audio all make very powerful passive power conditioners which can be great for isolating computer power supplies.

 

[PeterSt]

Vincent - just outloud reasoning, and no law I guess :

 

If you put the PC in another "power source" (in your case (I think) meaning a normal wal outlet instead of the conditioner), there may be current flowing through the mains into the conditioner. It may not be bad (because it is a conditioner), but still, better to be avoided.

 

If you put both in the conditioners' outlets (and it is adviced to do so) we may assume some isolation in there, and both should be "conditioned" separately. So, yes.

 

but

 

still the potential of both the PC and the other equipment will be different, so still current will be flowing from one to the other through (e.g.) a fire wire cable (or interlink etc.). So, that is just another problem to attack, and that will be avoided by connecting the two (etc.) cases, *if* the connections (of firewire, interlink) are connected to the cases (directly or indirectly via e.g. a PSU). Try to make "your" connection shorther than the interlinks ...

 

If that "other source" would be that other mains group (separate earth pin) then things are different, and I guess you will be better off connecting the PC outside of the conditioner (so it will just *never* feed back into it, may it *not* be isolated internally).

 

YMMV

 

You can always grab a volt meter (multi meter) and measure the voltage between any ground point and another (like in between P.E. and a cabinet, or like in between the shield of an interlink and another cabinet). Do you see something ? not good. From that you will learn the obvious "already there connections" or the other way around ("hey, this case is not connected to P.E. ground").

 

A maybe not all that obvious advice for any die heard audiophile :

Get yourself a cheapish (could be refurbished) scope for $100-$200) and "see" the real noise or hum. It's priceless for these kind of things ...

 

Peter

 

[Vincent Borrelli]

Thanks so much! Extremely helpful!

The Powercell has 5 isolated circuits, so I'll put the G5 on 1 of the 5 (isolated from the rest of the system, but still all on one ground). Also, it does limit current so having everything running off the single 20a dedicated circuit, even the mono amps, is no problem). And, thanks for the tip of putting a conditioner between the G5 and the Powercell. I've got an APC H15 not being used and I'll try that... much appreciated!

Vincent

 

[Vincent Borrelli]

Correction: "...it does not limit current..."

 

[G7]

"I've never had any problem with computer generated noise when running a simple set-up (computer to powered speakers) but I do in my multi-media set up. The combination of the TV monitor, cable, computer, and a few other odds and ends has given me a few head-aches with computer generated noise. Miguel"

 

Miguel,

Your computer worked just fine through DAC, Pre, and Amp.

 

The problem/s started when you added components, so why blame the comp' that was OK?

Are all the cables properly shielded, or all the components properly grounded, for instance?

Are all the components connected to the same non-stabilized power line?

Later, you said that the sound from the TV was noisy.

I suggest you ease the unfair pressure on the poor computer, or it may start making problems :-)

 

barrows on Thu, 11/11/2010 - 16:51

"The only thing that makes sense to me as far as the computer being relevant to sound is RF interference generated in the computer getting into the analog signal path."

"Other than RF, I am at a loss as to how the computer can affect sonics (I am assuming here that we have a properly set up rig, outputting a bit perfect data stream)."

 

That's both what I know and what I've experienced so far.

 

Submitted by 4est on Mon, 11/15/2010 - 08:23

"I have given up on thinking that 1s and 0s are solely "digital"

 

If this was true ("1s and 0s are NOT solely digital"), no computer would have been working. Power grids would have failed. Navigation would have not brought you to the airport. The number of zeroes after the sum of your bank account would have decreased. (Or, increased...) and many more examples.

Systems, from infrastructure to home electronics, are working because the '1's and '0's are kept accurate.

At the moment this stops being true, modern life would be replaced with chaos.

 

 

 

 

[4est]

That is an overly simplistic model IMO that presumes: that because we can make something work, then we must understand everything about it. I disagree. The square waves all have slew rate, overshoot and ringing.How well the down stream components are able to deal with this is variable.

 

[barrows]

beware this claim:

 

"Correction: "...it does not limit current..."

 

this is patently false, the correct statement would be: "it does limit current very much", as even a plain large gauge wire will add to impedance vs not being there at all. Also beware statements concerning "isolated" outlets-these "isolated" outlets are not entirely isolated. Note that: "does not limit current" and "isolated" are also (relatively) mutually exclusive features for a power conditioner. I am not trying to critisize Synergistic, as all power conditioner marketing makes similar claims which are not entirely accurate.

To provide a strong degree of isolation you will need a separate power conditioner for the computer and peripherals. I would recommend plugging this separate conditioner into the wall, and not into the Synergistic PowerCell. Plugging both into the wall will provide another degree of "isolation".

 

[G7]

4est,

 

Yes, the 0s & 1s aren't represented by an ideal square wave.

Still, at the DAC's input, 0s & 1s are accurately identified as such.

 

"How well the down stream components are able to deal with this is variable."

 

Indeed, and you point to components after the comp'.

You don't blame a computer for the difference between two DACs, right?

 

Export a whole music library to an external HD. Did a single bit change? No.

Copy it back to an internal HD. Did a single bit change? No.

The DAC gets what the HDs get - bit-perfect data. What it makes with it, is another story.