Let's just imagine for a moment that someone establishes via proper listening tests that USB cables can be reliably distinguished and the 'better' cables actually do deliver a 'better' signal (as far as I know that hasn't yet happened so at this stage it is just a hypothetical question). What would it mean?
Options:
1. USB chips are not properly implementing the CRC check, so accepting corrupted packets and turning them into audio. (incompetent/fraudulent chip designers)
2. USB DAC designers are relying on timing from a system which does not supply timing (incompetent DAC designers, who do not understand USB)
3. USB DAC designers simply use far too small FIFO buffers (incompetent/cheapskate DAC designers, hoping to save a bit of money)
4. USB DACs have lousy digital circuits (or power supplies) generating interference which is is made worse by retransmissions (as above)
5. USB DACs have poor grounding arrangements so they are relying on grounding via the USB cable. (as above)
See, there are sound engineering reasons why different USB cables could sound different but they all point to incompetent electronics, not some audiophile mumbo-jumbo. What some punters don't seem to realise is that there is an absolutely fundamental difference between SPDIF and USB as a DAC interface method: SPDIF delivers bits in real time so has to get it right on time all the time with only minimal jitter; USB can deliver a bag of bits whenever it feels like it so no timing information whatsoever can be derived or deduced from packets. Thus there is a (tiny) influence of cable quality on SPDIF, but no possibility of an influence on USB (in both cases, assuming normal engineering requirements have been met).
I have no idea which of my five suggestions may turn out to be true. At present I have no reason to conclude that any of them are true, as at present there is no established audio phenomenon to be explained - just a pile of anecdotes from people who appear not to understand basic IT technology.
Options:
1. USB chips are not properly implementing the CRC check, so accepting corrupted packets and turning them into audio. (incompetent/fraudulent chip designers)
2. USB DAC designers are relying on timing from a system which does not supply timing (incompetent DAC designers, who do not understand USB)
3. USB DAC designers simply use far too small FIFO buffers (incompetent/cheapskate DAC designers, hoping to save a bit of money)
4. USB DACs have lousy digital circuits (or power supplies) generating interference which is is made worse by retransmissions (as above)
5. USB DACs have poor grounding arrangements so they are relying on grounding via the USB cable. (as above)
See, there are sound engineering reasons why different USB cables could sound different but they all point to incompetent electronics, not some audiophile mumbo-jumbo. What some punters don't seem to realise is that there is an absolutely fundamental difference between SPDIF and USB as a DAC interface method: SPDIF delivers bits in real time so has to get it right on time all the time with only minimal jitter; USB can deliver a bag of bits whenever it feels like it so no timing information whatsoever can be derived or deduced from packets. Thus there is a (tiny) influence of cable quality on SPDIF, but no possibility of an influence on USB (in both cases, assuming normal engineering requirements have been met).
I have no idea which of my five suggestions may turn out to be true. At present I have no reason to conclude that any of them are true, as at present there is no established audio phenomenon to be explained - just a pile of anecdotes from people who appear not to understand basic IT technology.
Last edited:
The "audio" phenomena - an ‘applied’ scientist way to get rid of issues that there are no answer to be found in the books or on the internet - most of us are namely ‘applied’ scientists.
6. USB DAC designers don't understand how to do decent mixed-mode PCB design.
Mixing anything digital with a fast-changing edge with analogue circuitry on a PCB is an art as well as a science.
Most DAC (and other) designers just read the chip manufacturer's application notes and do roughly what they say, without understanding the issue as a whole, and that involves grounding, power, EMI (inductive and capacitative coupling, both from internal and external sources) etc. - every part of the signal chain from end-to-end.
Always been a problem. Even in so-called hi-end devices...
Mixing anything digital with a fast-changing edge with analogue circuitry on a PCB is an art as well as a science.
Most DAC (and other) designers just read the chip manufacturer's application notes and do roughly what they say, without understanding the issue as a whole, and that involves grounding, power, EMI (inductive and capacitative coupling, both from internal and external sources) etc. - every part of the signal chain from end-to-end.
Always been a problem. Even in so-called hi-end devices...
Last edited:
This is the typical argument-less discounting. The very point is the "audio phenomena" was NEVER turned into credible issue by presenting objective data by so called audiophiles. Intentionally.
If you and others relying just on your ears listened to the calls for credible and solid data, it is possible some phenomena would be already taken seriously and studied rigorously. Not just talked about.
But neither you nor others will present any credible results. Unless you were a major exception.
Perhaps I can reiterate what we found when using my (probably should be patented) self-randomizing, but consistent A/B testbed...
10 trials ... or less, for easily discernable differences.
25 trials ... or so, for small differences
100 trials ... or more for subtle, near-no-differences (to get +15/-15 results)
150 trials ... or any number, to get less than #trials/sqrt(#trials) no-difference confirmation.
Indeed - the N/sqrt(N) (which equals just sqrt(N)!) thing is a powerful reminder that it takes more than the response of one's significant other, one's father-in-law, one's cool neighbor and one's own supposedly abstractly-unbigoted ear. It takes maybe 10 people, each doing 10 to 20 trials, trying to guess whether they like something better than something else, without having ANY information as to whether they're listening to A, or B ... to work this out.
IN FACT - the thing to really remember is that you have to get 2×sqrt(N) "positives" to meaningfully say that you've detected "something".
And don't forget DF96's 5 conjectures. (Conjectures which only need considering - at all! - if the 100+ trial confirmation indicates a real difference!)
GoatGuy
10 trials ... or less, for easily discernable differences.
25 trials ... or so, for small differences
100 trials ... or more for subtle, near-no-differences (to get +15/-15 results)
150 trials ... or any number, to get less than #trials/sqrt(#trials) no-difference confirmation.
Indeed - the N/sqrt(N) (which equals just sqrt(N)!) thing is a powerful reminder that it takes more than the response of one's significant other, one's father-in-law, one's cool neighbor and one's own supposedly abstractly-unbigoted ear. It takes maybe 10 people, each doing 10 to 20 trials, trying to guess whether they like something better than something else, without having ANY information as to whether they're listening to A, or B ... to work this out.
IN FACT - the thing to really remember is that you have to get 2×sqrt(N) "positives" to meaningfully say that you've detected "something".
And don't forget DF96's 5 conjectures. (Conjectures which only need considering - at all! - if the 100+ trial confirmation indicates a real difference!)
GoatGuy
Last edited:
AB or ABX testing cannot be used as it simply does not work.
It does not work due to the same electromechanical reasons that causes cables to behave differently.
This is my opinion after decades with measurements and trials.
I'm so sorry you think so.
AB/ABX testing absolutely can work even for USB cables ... but it requires building a very special box, with very special electrical layout and characteristics. A reasonable amount of "overdesign" insures that the box will be adding very little to no color of its own to the mix. And will be highly symmetric.
But one has to be a GHz qualified digital designer to accomplish this. Not something you regularly find these days in the DIY audio world. I wouldn't even hint at being qualified, myself. Never got much past 50 MHz (HAM 6 meter band).
GoatGuy
AB/ABX testing absolutely can work even for USB cables ... but it requires building a very special box, with very special electrical layout and characteristics. A reasonable amount of "overdesign" insures that the box will be adding very little to no color of its own to the mix. And will be highly symmetric.
But one has to be a GHz qualified digital designer to accomplish this. Not something you regularly find these days in the DIY audio world. I wouldn't even hint at being qualified, myself. Never got much past 50 MHz (HAM 6 meter band).
GoatGuy
Ha ha ha ha ha ha ha ha, no its science, its been going on for many years, and the rest of your post is just the audiophile justification of why everything commercial is rubbish... Sorry I have been doing mixed signal layouts since 1985, I have worked on hundreds of projects with numerous engineers, all strive to do the best design they can, some of the cheap boards from e-bay etc I see on here being modded may fit into the category you present, not enough layers etc. But mostly a designer will design to the best of his abilities. (Henry Ott is a good guide on all this).
RAYCTECH, please enlighten me (us) on your cable knowledge that you have mentioned before on other threads in the past.
nope, that doesnt add anything to DF96's list, it would still need to result in one of the mentioned phenomena to be considered. i'd say its just science, there are well outlined techniques to deal with mixed signal design for more challenging fields than audio
No, it can be done much more simply than that; I gave one example above, you can probably think of two or three other ways. If you get a null result, even using a faith-based audiophile as the listener, don't expect that you'll shake his faith, but you will have learned something interesting for yourself.
On USB (like Ethernet) design, the specification is well documented, as have all computer interfaces, both PC (and thus common knowledge, SATA, USB, RS 232, Firewire etc etc) to more industrial based such as G64, VME, Profibus etc etc. All levels of the design are covered from low level programing to PCB layout. Most common modern interfaces have bespoke chips designed to perform the connection between the system and the output to connector. Thus all equipement plugged in to a modern computer standard interface these days works, why because it takes more effort to ignore all the rules, design aids, layoyt guides than it does to follow them and know it will work.
I do a lot of work with both audio and instrumentation analogue/digital designs, and the most destructive and problematic noise to deal with is noise in the target analogure range (20-20kHz audio), digital noise, controlled and under certain limits (EMC, testing design keeps this down despite the audiophile view of digital noise) has little or no effect on the analogue quality (some microscopic heating of inductors, and caps with high ESR, and I do mean microscopic). Noise in the target frequencies has a huge impact and if often the hardest to cure, though a general first fix is to beef up any grounding system with multiple point low inpedance straps, cos excessive star grounding always leads to a liitle escape route for some frequencys, audio is problematic when considering how to layout the grounding for a system as it covers the range where the return path is changing from path of least resistance to path of least inductance, makes things fun
On digital noise, hands up all those that use power line communications?
That is digital noise superimposed on your mains round your house, now calculate the loop areas, and where your equipment is placed, neat to a wall with a mains cable in it, or in the middle of the room!
Well... I too am an EE who for the last 30 years or so has been designing mixed-mode systems (though I've been mainly digital). Those designers with a true feel for what they do have often referred to the "art" of laying out boards - its a science too, as I stated, but the accumulated wisdom of years "at the coal face" is certainly a strong part of what they do - no auto-router can completely replace a good layout engineer yet (though some are pretty good).
I also feel you plainly didn't read or probably didn't understand what I wrote in the post - I have no truck with audio-foolery - I thought I made that clear
I also don't recall stating that "everything commercial is rubbish".... Perhaps you should calm down a bit? Have a cup of tea, maybe?
well maybe you EEs might call it art, but I wouldnt, I have an art/design background. those that turn technical fields into art usually do so by bending the rules/riding the boundaries while creating something that works and works well, or they write their own rules and it works well. I'm not sure either of these methodologies could be said of an EE
its a turn of phrase and I know what you mean, but i'm not sure its appropriate here, even when some effort is made to make the PCBs look nice/symmetrical, its aesthetic appeal is an attribute of the design, not a means to an end.
its a turn of phrase and I know what you mean, but i'm not sure its appropriate here, even when some effort is made to make the PCBs look nice/symmetrical, its aesthetic appeal is an attribute of the design, not a means to an end.
Sorry yes I did miss read, 25+ years at the coal face and still employed, though many changes of location. Though I must admit PCB layout is a dying art, most I know and see these days are getting rather long in the tooth, including myself. Yet it is at one of its most exciting points with the increasing high speed logic, 3D chip stacking, high speed RF and my favorite HDI (High Density Interconnect).Is that strictly true? In the isochronous USB transfer mode, the timing is derived purely from the rate the packets arrive. Yuk. So conceivably it might just be possible that the cable could affect the degree and frequency at which the DAC's PLL has to adjust its timing. In a more sensible system, e.g. asynchronous mode, the cable would have no effect whatsoever, short of total failure to get the information across.
actually "isochronous" is a 'reserved slot in time' which is reckoned by the pair of devices, transmitter and receiver. The transmitter controller (or higher level logic) can calculate from its buffered bitstream the optimal elapsed time between packets, and the receiver agrees to make that opening-slot available to receive a packet. In practice, little things (like other packets) get in the way, but that's no real matter - the buffer at the receiving end typically has 4+ packets in advance of play, in waiting. It therefore doesn't experience any "resynchronizing" of clocks or anything else. In conventional DAC playback, there is no provision for making up data that is missing. When the transmission rate is either close-but-higher, or close-but-slower than the receiver's clock, there certainly is the possibility of data drop or interpolation (respectively). Very smart DACs have precisely controllable synthetic clock circuits that can adjust the play rate to match the transmission stream by even as much as ±2% ... though of course such adjustment comes also as a shift in tone. This is not desired, for most music. Changing though by a couple of parts-per-10,000 is inaudible, even to persons with finely honed perfect pitch. And that is usually the on-the-ground, real-world difference in clocks between unsynchronized, but certainly crystal-referenced independent systems.
GoatGuy
GoatGuy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.