I understand EMI/RFI, but it's simply not the issue here that's being discussed, a building full of ethernet cabling doesn't have noise issues, the physical interface was designed with this in mind. It's more likely that there is noise coming from the internals of the DAC or the DC supply powering it or the computer rather than noise coming from the ethernet interface.
The real problem we have is the OP doesn't understand what he is doing, he got half an idea somewhere else off the internet. His first couple of posts were related to phase noise of the CPU xtal for the PCIe card which was supposedly causing jitter at the ethernet physical layer, which is doesn't.
Indiglo, you're right. Yet the intent of the OP doing this wasn't because it needed to be measured, reasoned, proven, disproven, falsified or whatever and of course he owes it to himself to have a new, unmodified nic, compare it and maybe even throw some measurements at it.
Oke, look at it like this. Suppose you'd want to make your own ethernet cable. Just to see if you can do it.
One comes and says you don't need to use teflon, because pvc would just be aok.
Another comes at you and says if you wanna do it right, you should use wires with a slightly different number of turns for the receiving pair than for the sending pair.
Another one comes and asks why you wanna do it in the first place, what's wrong with the stuff from the shop?
And another and so on.
The OP just wants to see if he can do it.
Even though all above advice is true, he doesn't need it right now, it's all distraction.
I know most (if not all) of you are right. Yet you gotta respect he's looking at it from a different angle, very hands on approach and just has to get through this. Make someone's day. He could've taken a dac board and try different power supplies or clocks on it to compare. Very likely nobody would've argued, true.
Worst that can happen is breaking a 10 dollar NIC and some ego. Maybe it's even a good thing. Breaking things usually makes you get more curious for the underlying basics. I know I did, still do sometimes.
Maybe even he'll eventually compare them and finds out the differences fade away after a few weeks. Can't hear a difference at all. Next you know, when someone else who wants to try this here on diyaudio, he'll be there saying it's all a hoax, buts are bits. Imagine that. But, gotta give him a chance to right?!
Completely disconnected from reality. In fact, so brain damaged that any answer is futile.
A VHF signal, from an antenna on earth that makes it through the atmosphere, makes
it to the moon, is scattered back from the moon's dusty surface, travels the 380000 Km
a second time and makes it back to the antenna, experiences a path loss of about
230 dB. But I'm sure you could still hear it, at least in sighted testing. The additional
70 dB included.
What is the color of the sky on your home planet?
In the effort to inspire learning, I'm going to suggest you find your "proof" somewhere besides my posts.
The principle is we can analyze and quantify any and all changes to a waveform, and display them in both time and frequency domains.
And...
We cannot analyze perception.
So...
The cognitive gap is scaling measurements to human perception. That means we can measure everything about the wave presented to the ear. Hearing is perception of that wave.
In developing electronics that impact the quality of audio, the first level is to analyze what you have now, then analyze what you have after changes.
And again, the software, REW, is more than capable, and is free.
- And spending even more money in Japan, for unwarranted mods, instead spending towards legitimate improvement.
- After completing the mods, since ego is a bitch, we'll have a triumphant report that "it sounds so much better", maybe peppered with the "wife in the kitchen" type of story.
- Depending on how literary attractive the story is, a bunch of other innocent amateurs may jump in and attempt the same type of mod, spending money on unwarranted changes, when that money could be spent towards legitimate improvement.
- Meantime, existing or wannabe snake oil merchant swindlers are encouraging this type of stories, while slobbering about their next product, the "audiophile network card" *). This is exactly the kind of stories they need 10% facts and 90% BS to propagate yet another crapola myth.
*) Like a $400 audiophile network switch Introducing EtherStream, Audiophile Network Switch from Fidelizer Audio | Fidelizer.Audio (20 bucks at your next computer shop). Or even better, the 435 euros NET Card FEMTO - JCAT . next-gen computer audio (10 bucks at your next computer shop). Of course, a couple of audiophile network cables, for 300 euros a pop, cannot be missed LAN Cable - JCAT . next-gen computer audio
Last edited:
Jaddie, first and foremost, I was generalizing, on the measurement part maybe taking it a bit too far, and maybe I wasn't talking about you;-)
But, somehwere I did read someone saying that phase noise had to be measured to compare the standard clock to the new, XO, coul've been that wasn't you who said that, in that case of course the remark wasn't directed at all at you.
Having said that, I certainly also wasn't looking to chase people out the door, or off the thread, my intent was to pile up the list of good intentions, most if not all excellent advice and just try to make it a bit more directed towards the OP's needs and less about being right, technically sound or sane.
I sincerely hope you reconsider and join back in!
Btw were you suggesting to use a sound card to measure the output of OP's dac and look for differences?
Could work.
Among others, I'd be interested in finding differences wrt common mode noise, capacitively coupled by the transformers used in the network, at least this could explain, maybe, some people's claims that optical networks do so well.
I won't break my own rules, so won't say someone needs a really expensive measurement device to prove this of course;-)
Maybe it could be measured with a mod on your sound card, but I doubt it would generate usefull data. Also, these things might behave differently ina measurement setup than in OP, or anyone's stereo setup.
Okay, and I tend to agree, I get where you're coming from, some people will do things in spite of the information presented, but it doesn't justify the attitude by telling everyone that he knows better then the rest of us.
Last edited:
I didn't start this debate, I started a DIY post on a DIY forum about some modifications to a NIC, and asked for help.
You and some others then started saying how useless it was, how could I think it would improve the sound, and so on, usually motivated with "bits-are-bits" type of argument which isn't applicable in this case (I am not trying to make the NIC more fault tolerant).
Thankfully some others have helped me and now I know how to proceed, just have to wait for a shipment from Japan, and figure out clock-input to a chip 🙂
But that's the whole point, isn't it? There isn't a uniform jitter threshold for all listeners, as is true for most other audio error parameters as well. Yes, at some level, conversion jitter will be inaudible for essentially all listeners. That level, however, is currently in dispute is all. Your assertion that jitter is not relevant below levels that provoke hard bit-errors doesn't make that assertion universally true. It's your opinion. Fine. But that's all it is. As you yourself had pointed out, jitter audibility isn't simply a matter of it's amplitude, for jitter is a complex structure. Which leads me to wonder why you now seem to be countering your earlier statement?
First, my initial post was in response to your prior statement, repeated here:
"If you don't change the bits, you don't change the resulting sound. I agree that if the data arrives at the DAC bit-perfect, the it will be bit perfect without the tweaks. But, if it's arriving bit-perfect, then what are we trying to change?".
You are clearly suggesting that jitter is a non-issue unless it provokes bit-errors. Bringing up other error parmenters, such as noise, is an attempted evasion. One kind of error being lower in power than some other kind of error doesn't automatically render the smaller error inaudible. Surely, you know that.
I have already made quite clear that I've been referring to conversion jitter. As I've also previously made quite clear, as long as bits remain in the digital domain, jitter does not matter, so long as it isn't severe enough to provoke bit-errors over a given data transmission link. That is basic. It DOES matter at levels below that which would cause transmission bit-errors in systems where domain conversion (ADC/DAC) is taking place. Digital audio is, of course, such a system.
The key question is, how do you know whether a given instrumented jitter measurement is low enough for the ear? Is this merely an judgment of how 'nice' a jitter plot appears to the eyes on a display screen? Is that any less an arbitrary yardstick than is the ear? More reapeatable yes, less arbitrary, I don't think so. Yes, measuring jitter is trivial with the proper equipment. Eliminating it is impossible. Reducing it to inaudibility, however, is the design issue. Audibility is in the ear of a given listener. You may feel that your ears don't perceive jitter's affects, that's fine. Others, however, obviously feel they do perceive it's affect. I didn't read anyone declaring that you do indeed hear the affect of jitter when you feel that you don't. Why then do you declare that they do not hear it when they feel that they do?
Last edited:
You understand to down to a certain level, but you omit where it can get more complicated.
For the frequency domain, your claim applies only if waveform changes are stationary over the time of the FFT and the time of any averaging that may be done to attenuate displayed noise. IME, not all audible effects of RFI/EMI on dacs are sufficiently stationary to measure with a soundcard and or scope. There is also hysteresis distortion which can be hard to measure at low levels and which is nonstationary, but its still audible according to Purifi: This Thing We Have About Hysteresis Distortion - PURIFI ...just a few examples of where it can get more complicated.
The foregoing is for (mostly) nonlinear distortion.
Linear distortion is harder to measure at very low levels, particularly if it is somewhat dynamic (which makes it weakly nonstationary). Take the case of audible effects due to interconnection cables, as one example: AES E-Library >> Physical characteristics of analog audio cables and their effect on sound quality
If one only thinks of examples such as where REW and a sound card are easily applied, one may conclude that's all there is. As Daniel Kahneman tried to emphasize: WYSIATI: A machine for jumping to conclusions
If you had formal education in electronics you would understand that changing a xtal connected to the microcontroller of a PCIe ethernet card would not make any changes to anything connected to it.
Isn't it nice when wrapping BS in a pseudo scientific language is debunked?
"Linear distortions" are defined (and not everybody accepts this definition as anywhere relevant or useful) as changes in the frequency response or phase of a linear device (that is, without any non linearities). Accept this definition or not, fact is, changes in the amplitude/phase are among the easiest measurements we can exercise on a device. Phase change measurements are at the core of the LIGO experiment, detecting phase changes after what is essentially a +430dB parametric amplifier.
To add insult to injury, all audio amplifiers are minimum phase systems, where the amplitude and phase response are exactly coupled - there's no need to measure separately, do one and you got both.
"Non stationary", in this context, is just throwing an interesting word, for the artistic effect. As a wild guess, if you mean by "non stationary" a time variant system/amplifier, then I'll let you worry about such in audio; other sentient humans have better things to do than to worry about the impact of the moon shining on their washer/dryer in the laundry.
250 dB. Its a logarithmic scale. Do you even start to comprehend what that figure means? Mr Magnus, you are making a fool out of yourself here.
//
Some people who knows this area extremely well certainly don't agree with you. Just ask any designer of streamers.
But this picture is very applicable in cases like this. I think I know, and I don't know if you know, or only think you know. So instead I experiment, listen and decide for myself. I know how to remove (or at least reduce) bias but you are free to believe the improvement I claim to hear is 100% bias.
I am not trying to prove anything, I am just enjoying my hobby 🙂
Attachments
This is hilarious, I suspect these so called "designer of streamers" are not electronic engineers.
Rules of Thumb for Music and Audio - Benchmark Media Systems
Rule 7: 0 dB SPL is the threshold of normal hearing.
A sound at 0 dB SPL is just loud enough to be heard. It is roughly equivalent to the sound of a mosquito flying 10 feet (3 m) away. This threshold of hearing is frequency dependent and listener dependent. Most listeners with normal hearing have maximum sensitivity at about 3 kHz. This sensitivity tapers off above and below 3 kHz. Use the 0 dB SPL threshold for quick approximations of audibility especially when considering mid-range sounds.
SPL is an abbreviation for sound pressure level. It is measured in dB relative to a reference pressure level that corresponds to the threshold of hearing.
Rule 8: 130 dB SPL is about the highest peak loudness that we can tolerate.
Very brief exposures at this level can cause permanent hearing damage. OSHA allows less than 2 minutes of exposure per work day at 130 dB SPL. The threshold of pain may be as high as 140 dB for some listeners but this threshold also varies with frequency. Curiously, the threshold of pain tends to get lower after our ears have sustained permanent damage. Some listeners may experience pain at much lower levels.
In the context of a home audio system, there is no sane reason for having the ability to produce peak sound pressure levels above 130 dB SPL at the listening position. If your sound system is hitting your ears with 130 dB SPL on very short peaks it is almost certainly playing at ear-damaging levels.
In a very large room or an outside venue, a sound system may need to produce sounds above 130 dB SPL (at the speakers) in order to achieve reasonable levels at the listening position.
- - - - - - - - - - - - -
Then you feel/think/know/guess that yet 110 dB is available hearing dynamics...
Troll.
//
Rule 7: 0 dB SPL is the threshold of normal hearing.
A sound at 0 dB SPL is just loud enough to be heard. It is roughly equivalent to the sound of a mosquito flying 10 feet (3 m) away. This threshold of hearing is frequency dependent and listener dependent. Most listeners with normal hearing have maximum sensitivity at about 3 kHz. This sensitivity tapers off above and below 3 kHz. Use the 0 dB SPL threshold for quick approximations of audibility especially when considering mid-range sounds.
SPL is an abbreviation for sound pressure level. It is measured in dB relative to a reference pressure level that corresponds to the threshold of hearing.
Rule 8: 130 dB SPL is about the highest peak loudness that we can tolerate.
Very brief exposures at this level can cause permanent hearing damage. OSHA allows less than 2 minutes of exposure per work day at 130 dB SPL. The threshold of pain may be as high as 140 dB for some listeners but this threshold also varies with frequency. Curiously, the threshold of pain tends to get lower after our ears have sustained permanent damage. Some listeners may experience pain at much lower levels.
In the context of a home audio system, there is no sane reason for having the ability to produce peak sound pressure levels above 130 dB SPL at the listening position. If your sound system is hitting your ears with 130 dB SPL on very short peaks it is almost certainly playing at ear-damaging levels.
In a very large room or an outside venue, a sound system may need to produce sounds above 130 dB SPL (at the speakers) in order to achieve reasonable levels at the listening position.
- - - - - - - - - - - - -
Then you feel/think/know/guess that yet 110 dB is available hearing dynamics...
Troll.
//
No, I am sure you think you know more about digital audio than guys like John Westlake, Swanson, Rankin, Watts and so on.
Sound pressure - Wikipedia
0 dB is used as a reference for the quietest level that can be heard. You'd have to be umpteen kilometers away in order to get a level of -200 dB relative to that... by which time you'd go crazy trying to understand why you crossed such a distance to prove to yourself that you couldn't hear 0 dB from that far away.
- Status
- Not open for further replies.