Testing audiophile claims and myths - Page 132

Quote:

Originally Posted by xnor 

I think expensive (digital) cables are equally absurd, but that's just my opinion.

 

As for resending: that's simply not true. It's called USB audio streaming for a reason, it's a real-time system.

 

In isochronous transfer mode the USB bus sends frames every 1 ms (= 1000 Hz). If your computer can handle real-time audio streaming (which any properly configured PC can) each frame will contain the next 1 ms worth of audio data. If not there will be plainly audible glitches.

 

If a frame contains an error there is absolutely no attempt to resend anything. The sender doesn't care what the receiver receives, it just pushes the data in an as constant rate as possible. In fact, the receiver could not even tell the sender to resend anything because the communication is unidirectional.

 

The link you posted contains some nice marketing material, but that's about it..

Ok. I am trying to justify the reason why I think it to be better (and I am quite sure of that). It seems you have some knowledge on this, so please correct me if I am wrong as I do want to learn more on this.

Ok so it has error checking, but not error correction. How does it help if it detects an error?  Windows is not a real time system from what I know.

Quote:

Originally Posted by dan.gheorghe 

Ok so it has error checking, but not error correction. How does it help if it detects an error?

You have to keep in mind that errors are extremely rare even with the cheapest of USB cables. Think of 1 flipped bit (1/8th of a byte) in a few months of nonstop (24 hours a day, 7 days a week ..) playback of audio.

That's why correcting errors is of no concern. What's of concern for streaming audio is that data arrives at a constant rate.

That is true. They keyword here is Deferred Procedure Call (DPC) latency. Windows has a queue of such DPCs. Filling the next frame with audio data is the job of one of those DPCs. Now if the queue is too long or the calls take too long to complete the audio data cannot be provided in time. Again, the result is clearly audible crackling noise or at least an annoying *click*.

There are several free DPC latency checker tools like "LatencyMon" or "DPC Latency Checker". They also offer some advice how to fix those problems. Once the latency is fine data will be available and sent every millisecond. Isochronous transfer mode guarantees access to USB bandwidth.

Misconfigured PCs and bad drivers are the main reason why people are having audio glitches with USB DACs. Internal soundcards are usually more resistant to DPC latency.

And it should be noted that the cable plays no role in this process. Provided that it is not broken, and built to the very generous USB specs, and will still transmit data, it does not affect this transfer of information. 

But Burson Conductor with its Asynchronous USB receiver should be pretty much immune to all these timing errors issues.

Quote:

Originally Posted by Andrew_WOT 

But Burson Conductor with its Asynchronous USB receiver should be pretty much immune to all these timing errors issues.

As for errors that's not true. The only difference with async is that there is feedback from the receiver so the sender can make small adjustments in the data rate.

Jitter is usually not a problem with either type of synchronization.

I heard that $70 USB cables also make the photos off digital cameras look better - more vibrant colours in the midrange and bluer skies. That's how it looks to me and I trust my eyes.

I know you're just kidding, but the mind is quite powerful. Once bias creeps in you could very well hear differences. Differences that are not there in reality, but in your head.

I find these debates a bit interesting, because there is often some nugget of truth somewhere in there that has long since been covered in a massive mountain of crap. For example, once upon a time, I had this giant spool of wire labeled "speaker wire" that was actually aluminum. Of course, upgrading from the 5 cent per foot to the 30 cent per foot ofc speaker wire I currently use made a difference. It is more conductive, less resistive, etc, etc. But, that doesn't mean that $100 per foot wire will be at all superior. I think some are conditioned to accept this notion of diminishing returns, and some feel that "you just can't hear that level of quality." But, that's not at all the case. Once upon a time, people used what they had lying around as interconnects, speaker wire, etc. I will admit that I even had a pair of interconnects made out of speaker wire. Yes, switching to real interconnects made a big difference, they were far less capacitive. Once the connector meets the required spec for the equipment, then it doesn't matter if it's $1 or $100,000, they should function nearly identically.

Regarding USB cables. I once found an audible difference. I had this really crappy USB cable that came with a really cheap charger. The cable did not have the USB logo, and basically did not meet the USB standards of impedance, capacitance, and shielding. Copying data with it had errors, etc. It was intended basically to be a cheap knock-off that was capable of delivering the power from the charger. When I replaced it with a REAL USB cable, everything worked fine. I would imagine that using a similar POS cable with audio would have some pretty crappy consequences. I remember that if the (cordless) phone rang, using it to transfer data was impossible. But, that's not a $70 upgrade, monoprice sells excellent USB cables for ~$1.

Yes, more conductive, less capacitive cable will work better. But this isn't a case of diminishing returns. Either they meet the spec for USB, or they do not. And it's literally 1s and 0s, how can the cable color, flavor, or massage the sound? I mean, unless you were previously using some similar knock off cable that had interference and all kinds of drop outs, there's no conceivable way to argue that.

If we can all agree that what travels over USB is digital, which it sounds like we can, then I propose a test. If a cable can impact the 1s and 0s of audio, it can impact the 1s and 0s of data. Windows tells you data transfer rates pretty precisely. I would like someone to try this. Take a file, any file and transfer it to an external harddrive 1000 times, recording the mean transfer rate for each transfer. Do so to the same drive with both cables. If there is a significant difference (based upon a 95% alpha, and I would like to see raw data - I'll do the analysis for you), then I give you my word, I will audition the $70 USB cable.

Quote:

Originally Posted by scootsit 

If we can all agree that what travels over USB is digital, which it sounds like we can, then I propose a test. If a cable can impact the 1s and 0s of audio, it can impact the 1s and 0s of data. Windows tells you data transfer rates pretty precisely. I would like someone to try this. Take a file, any file and transfer it to an external harddrive 1000 times, recording the mean transfer rate for each transfer. Do so to the same drive with both cables. If there is a significant difference (based upon a 95% alpha, and I would like to see raw data - I'll do the analysis for you), then I give you my word, I will audition the $70 USB cable.

I almost wrote something like this above but I think an easy counterargument is that the USB audio "protocol" if you like (not sure what the right word is) differs from the mode used for file transfers. The point is still good though that if it can do file transfer OK (i.e. not millions of error corrects, should error correction be available) then it is likely to be similarly reliable for streaming audio, i.e. error rates of the one in many billions.

The correct terms are isochronous transfers (aka streaming real-time transfers) for example used with audio devices and bulk data transfers for example used with external hard disks.

-> Isochronous guarantees bandwidth but not data integrity.

-> Bulk guarantees data integrity but not bandwidth.

Has anyone run blind tests with an equipment that had burn in and a brand new counterpart?

for headphones it is next to impossilbe

few serial production headphones of the same model match in frequency response to better than AB/X hearing thresholds, even those crowing about matched L&R drivers are mostly showing how poor in absolute terms "good" production tolerance is

then you can't even put the same headphone on your head accurately enough to avoid frequency response changes between sessions - if Tyll's InnerFidelity reproducibility plots with the dummy head is to be believed

Quote:

Originally Posted by uchihaitachi 

Has anyone run blind tests with an equipment that had burn in and a brand new counterpart?

Blind, no, but objective, yes:

http://www.innerfidelity.com/content/evidence-headphone-break

Thanks for the link! Very informative