Jump to content

Sign in to follow this  
jolon

Why digital seems to be affected by power and cables

Recommended Posts

The observation is very audible, so, how are the observed differences measured, realtime!

Test the cable for it's specified characteristics and note if the cable meets those specs. Many do not. Apply a known data stream to the cable and report any lost information.

Have you heard/observed audible differences between USB and optical cables, yourself, when using the same components? It appears to be quite a common occurrence and there are lots of USB cables on the market :)

No, but then I don't use USB for audio. Too many potential trouble spots. As for digital interconnects, as long as the impedance characteristic is correct (many cables fail this basic test) and the connections secure and appropriate (RCA connectors are NOT appropriate - BNC should be used), then the differences between cables disappears.

Data transmission differences(errors) should not occur with USB because USB transmission is carrying checksum packaged digital data! So a question is... what measurable changes to the data could there be?

Asked and answered.

Share this post


Link to post
Share on other sites

If there are audible differences between digital interconnects, then those differences WILL be measurable. Probably data is being lost and insufficient correction data is included. There can be no other explanation.Differences between any tow audio products is ALWAYS measurable. The trick is to figure out which measurement is applicable.

Zaph, I believe that most of the USB cable differences are not due in any part to data corruption or loss...packets are packets and they either make it they don't (they are not re-sent in USB audio protocol if there is a transmission error)...if too many packets are lost then we hear something akin to static.

However, the USB cable provides an electrical link between components without which data transmission would not be possible, and the thinking (and proof by measurements) goes that it is how this electrical connection is handled in the components at either end of the cable and by the cable itself that influences the electrical noise in the dac (mostly ground plane noise) that influences the DAC clocks which are also electrically attached to this common ground.

Saying all this, what seems to be the very best solution to this at this stage in my setup is a meticulously terminated and well made USB cable that actually manages to accurately meet the 75 ohm impedance requirements of the USB standard (a rare thing) and a DAC that has a USB interface that not only also meets those standards but is cleverly designed to avoid things like galvanic isolation that are shown to increase DAC noise on the ground plane from the USB interface.

This post has been drastically oversimplified, but hopefully I made it clear that it is not a data change because of this digital cable that causes a measurable change and the DAC output, it is the electrical inter connectivity of the components in the audio chain that causes the measurable change in the analogue output of a DAC.

Cheers,

Anthony

Share this post


Link to post
Share on other sites

As for digital interconnects, as long as the impedance characteristic is correct (many cables fail this basic test) and the connections secure and appropriate (RCA connectors are NOT appropriate - BNC should be used), then the differences between cables disappears.

BNC connectors are much more suitable (than RCA connectors) for analogue connections also...better impedance matching is possible and in all likihood a better physical connection is made as well.

Share this post


Link to post
Share on other sites

Saying all this, what seems to be the very best solution to this at this stage in my setup is a meticulously terminated and well made USB cable that actually manages to accurately meet the 75 ohm impedance requirements of the USB standard (a rare thing) and a DAC that has a USB interface that not only also meets those standards but is cleverly designed to avoid things like galvanic isolation that are shown to increase DAC noise on the ground plane from the USB interface.

Pretty sure the USB standard is 90 ohms.

I agree with your post though - this reflects my experiencs too :)

Share this post


Link to post
Share on other sites


Pretty sure the USB standard is 90 ohms.

I agree with your post though - this reflects my experiencs too :)

 

You could well be right - I pulled that number from my memory, which is notoriously unreliable according to my wife.

Share this post


Link to post
Share on other sites

So, the sound of USB audio is determined by an electrical analogue scenario, other than the USB checksum packaged digital transmission, which appears to affect the DACs analogue output.   So, it is not a digital issue but an analogue issue... Bugger and Crikey, no wonder we can't measure it!

  • A question then is ... why does USB not lose data with a plug and play HD environment although the sound of USB external CD/DVD burn can be changed using different USB cables where no DAC is involved?

 

I have not found BNC connectors to sound any good and I have not found any issue with appropriate RCA connectors which are of a very high standard nowadays :)

 

Your posts suggest that SPDIF for audio is still better than USB without the need for continual software updates (which is becoming more common) ... maybe/perhaps :)

Share this post


Link to post
Share on other sites

 

 Sorry if the physics is beyond your educational framework.

 

 

In my opinion, that's a bit harsh and or sarcastic and not a statement I would make without measurements or proof.

 

I have made a few probably unscientific measurements with my system and they have provided guidance and insight, but with so much stuff available for us types that are striving for sound improvements, I am guided by my ears in the long run and hopefully a little bit of common sense.... but mainly a heck of a lot of listening.

 

Thanks for a great thread and a great read.

 

Cheers,

Steve

Share this post


Link to post
Share on other sites

@@steve u, It was a sincere statement given that many people don't have the background to read such a paper and understand the math. That's all. I did not write that it is beyond Ping's education only that it might be, which, not knowing what the facts are, might be true. I get asked about measuring quarks and gave the correct reference. The way the question was asked is not an indication of a great familiarity with the subject if it was indeed a serious question. I was serious in pointing out the counter intuitive nature of Nature and there is no validity in people's suggestions that their intuitive experiences correspond to anything verifiably real within the limits I have already enunciated. Note that this is not to doubt the reality of the experience itself, only the validity of conclusions drawn from experience. Reality does not care what we believe about it. 

Edited by dr_carl

Share this post


Link to post
Share on other sites


@@steve u, It was a sincere statement given that many people don't have the background to read such a paper and understand the math. That's all. I did not write that it is beyond Ping's education only that it might be, which, not knowing what the facts are, might be true. I get asked about measuring quarks and gave the correct reference. The way the question was asked is not an indication of a great familiarity with the subject if it was indeed a serious question. I was serious in pointing out the counter intuitive nature of Nature and there is no validity in people's suggestions that their intuitive experiences correspond to anything verifiably real within the limits I have already enunciated. Note that this is not to doubt the reality of the experience itself, only the validity of conclusions drawn from experience. Reality does not care what we believe about it.

Nicely said. That brought a smile to my dial.

Steve.

Share this post


Link to post
Share on other sites

I've done quite a number of measurements on SPDIF interfaces, BNC connectors, cables and RCA connectors.  I've done this using Time Domain Reflectometry with a 25ps resolution, and oscilloscopes with 500MHz bandwidth.  Here's a summary of the findings:

 

1. Most (decent) SPDIF interfaces are galvanically isolated using pulse transformers at each end of the link.  Just about every manufacturer, right through to high end gear get this wrong.  Pulse transformers in general do not like working at 75 ohms, and to get a good, square rise time without overshoot and ringing needs quite a bit of care.  With the right resistor network at each end you can get about 1ns rise time with no ringing.

 

2. Many high end digital SPDIF cables are not 75 ohms - I've measured anything from 72 to 83 ohms characteristic impedance.  Item 1, and 2 explains why different digital cables sound different in most systems.  If you get 1 absolutely correct (and have a really good 75 ohm environment at each end to work with), and then use a good professional grade (cheap!) 75 ohm cable - job done.

 

3. Much to my surprise, a particular RCA connector set measures significantly better than any 75 ohm BNC plug/socket (and I tested several brands).  We're talking impedance discontinuities *inside* the connector itself.  But - the WBT 110 plug and 0210 socket is a perfect 75 ohm environment to the limit of my measurement capability.  Could have knocked me down with a feather!  All other RCA's I've tested are completely lousy.  So the order of merit is WBT 110/0210 followed by 75 ohm BNC.  Any other brand RCA - forget it.  By the way, the inherent limitations of commercial BNC's, whether 75 or 50 ohm led Tektronix to develop their own BNC connector that they could trust to do the job correctly.

Share this post


Link to post
Share on other sites

Nice to see some measurements, rather than mention of quarks and dark matter :)

 

I'm intrigued how digital transports seem to be affected by power and cables in a way, that to my ear, sounds like an analogue effect.

Recorded DSD looks alot like stored Analogue so hopefully when played back it will sound like analogue reproduction!

 

Perhaps a difference between digital and analogue is the quality and accuracy of the stored harmonic content which provides the realness and accuracy of timbre, prat, decay, leading edge control, and other artifacts that allows us to determine that a note/tone is emanating from this instrument or that instrument.    

 

I would suggest that recorded analogue would contain more real harmonic content than digital.   The recorded digital may have removed low energy real harmonic content as noise, or as simply not required, as part of the compression process to reduce storage requirements, especially re low energy upper mids/highs harmonic content. The missing low energy upper mids/highs harmonic content from the stored DSD recording, re translation and playback, could easily account for why analogue sounds more natural than digital :)

 

Fortunately, with continual improvements, re digital playback and reproduction, especially re low energy real harmonic content, digital playback sound is sounding more and more analogue in its presentation :)

Edited by Ping

Share this post


Link to post
Share on other sites

Indeed - it has taken more than 30 years' development, and massive advances in silicon in that period (Moore's law = 150,000 over that time) to get from Philips "Perfect sound, forever" hyperbolic claim for digital, for it to be close to the sonic presentation of analogue.

 

1983:  IBM-XT PC, 20MB hard disc, 64k RAM, 8MHz clock

2015:  I7 series, 6 cores 12 threads, multi-TB hard disc, 16GB+ RAM, 4GHz clock

 

Which spans the commercial evolution of digital recording and playback.

Share this post


Link to post
Share on other sites


Here are some real measurements of connectors.  In each case true 75 ohm cable was used, terminated in the mating connector.  The chassis connector was terminated by a 75 ohm 1% surface mount resistor soldered directly across the terminals.  The scales are all the same, so it is a real comparison.

 

First a WBT0144.  Although a high cost esoteric RCA, there is a massive reflection indicating that it is nowhere like 75 ohms on a nanosecond timescale.

 

post-153047-0-34194500-1449501448_thumb.

 

Second is a high quality 75-ohm BNC.  The reflection discontinuity is half as long as the 0144 at 500ps, and although the downward going transition is better, there is also an upward going bit, indicating that the impedance starts off lower than 75 ohms and then transitions to higher than 75 ohms.  In a sense this is a slightly academic measurement because 90-odd% of digital equipment manufacturers fit 50 ohm BNCs!

 

post-153047-0-83029600-1449501825_thumb.

 

Finally is the RCA WBT 110 plug and 0210 socket pair, which WBT say is 75 ohms.  That is actually correct!  In this case there is hardly any impedance discontinuity at all - you could guesstimate it at +/-10millirho - which indicates that it is within about 1.5 ohms of 75 (2%!) on a nanosecond timescale, and is clearly much better than a 75 ohm BNC connector pair.  That is bewilderingly good.  Colour me surprised.

 

post-153047-0-64902100-1449501899_thumb.

Edited by CraigS

Share this post


Link to post
Share on other sites
Guest Doppler Effect

With tertiary studies in both electronics engineering and psychology, as well as many years in various engineering positions with experience from audio equipment design to data, information, and control systems, my take on this is fairly broad and i'd like to think well-informed.

 

I'll start by pointing out that I believe there is a reliance on obfuscation in some marketing endeavours.  Further, I frequently encounter arguments against measurement that are based on a shallow (for want of a better word) understanding of the phenomena under investigation.  Often this is tied in with a claim that a subjective descriptive term does not lend it itself to measurement.  Often this is simply that there is a measurable parallel, which may or may not be known and understood by the person making the remark.  Even pointing out this parallel between the subjective and objective can be fraud with debate, made worse by some engineers being poor at communicating across various educational levels or not being inclined to (after all, what is to be gained?)  So we have discussions involving various levels of expertise, various assumptions of expertise (both in regard to self and other), and just to make sure it remains a bit confusing we have the fact that there is, as always, much to be studied and learnt.

So now we come to subjective evaluation of the resulting product, and a point to make regarding biases: they are a real thing.  It is true that acknowledging them provides us stronger tools for critical evaluation, but this by no means offers immunity.  When I am designing products for KARRI acoustic I neither excuse or exempt myself from these.  I design based on sound engineering and psychoacoustic principals and understanding, and then compare and contrast subjective and objective data.

But here is the crux.  Its that those of us in design need to be mindful of what we are designing.  What good is it to hear our customers telling us that something provides them a nicer experience if we don't listen to that and interpret it?  If someone tells us that a certain item provides them with a more organic experience, for example, and as engineers we are confident that the internals are unlikely to be accounting for that difference, then surely it is upon us to determine what is!  Expectation bias?  Sure, absolutely, but is that telling the full story?  In food circles it has become commonplace to use the phrase "we eat with our eyes".  If there are visual styling elements, to call on only one example, that are involved in the full sensory-peerceptory experience, then isn't it valuable to respond with products that are attuned to this?  I think it is.

Although I will add that I don't believe this should extend any kind of laissez faire in regard to disseminating pseudoscience.  Quite the opposite, it is merely saying that we should broaden the net of science in designing products.

 

 

TL;DR: Listening to customers say "I like this" and answering with "no you don't" is neither good design nor good business.

Edited by Doppler Effect

Share this post


Link to post
Share on other sites

 

 

 

TL;DR: Listening to customers say "I like this" and answering with "no you don't" is neither good design nor good business.

 

 

A great quote from a very balanced and thoughtful post there Karri.  One of the things that I find very interesting is the idea (seemingly demonstrated in this thread) some have developed that subjectivity and science are mutually exclusive.

 

Having spent some time with several great Australian speaker designers, one common theme is that all utilise a combination of measurement, analysis, and subjective listening in designing, and particularly tuning of their speakers.

Share this post


Link to post
Share on other sites

(seemingly demonstrated in this thread.

I'd argue otherwise.  Maybe I'm getting old but I've seen more decent discussion in this thread than I've seen in a long time.  :)

Share this post


Link to post
Share on other sites


Here are some real measurements of connectors.  In each case true 75 ohm cable was used, terminated in the mating connector.  The chassis connector was terminated by a 75 ohm 1% surface mount resistor soldered directly across the terminals.  The scales are all the same, so it is a real comparison.

 

First a WBT0144.  Although a high cost esoteric RCA, there is a massive reflection indicating that it is nowhere like 75 ohms on a nanosecond timescale.

 

attachicon.gifWB0144 TDR.jpg

 

Second is a high quality 75-ohm BNC.  The reflection discontinuity is half as long as the 0144 at 500ps, and although the downward going transition is better, there is also an upward going bit, indicating that the impedance starts off lower than 75 ohms and then transitions to higher than 75 ohms.  In a sense this is a slightly academic measurement because 90-odd% of digital equipment manufacturers fit 50 ohm BNCs!

 

attachicon.gif75 ohm BNC TDR.jpg

 

Finally is the RCA WBT 110 plug and 0210 socket pair, which WBT say is 75 ohms.  That is actually correct!  In this case there is hardly any impedance discontinuity at all - you could guesstimate it at +/-10millirho - which indicates that it is within about 1.5 ohms of 75 (2%!) on a nanosecond timescale, and is clearly much better than a 75 ohm BNC connector pair.  That is bewilderingly good.  Colour me surprised.

 

attachicon.gifWB0110 TDR.jpg

Yes and interesting, and the winner is SPDIF via RCA and more importantly, to my ears, SPDIF via RCA sounds better, ie for me, more analogue :)

Edited by Ping

Share this post


Link to post
Share on other sites

Needs the specified RCA's though.  Any other RCA's are lousy from a measurement perspective.  But if the send and receive environment are not 75 ohms on a nanosecond timescale the connector is a secondary thing.  Just as a for instance, this is a well thought of pulse transformer with a 75 ohm resistive loading.  Note that the scales are different - 100mrho vertical and 5ns per div horizontal

 

post-153047-0-96840400-1449564248_thumb.

 

Share this post


Link to post
Share on other sites

Needs the specified RCA's though.  Any other RCA's are lousy from a measurement perspective.  But if the send and receive environment are not 75 ohms on a nanosecond timescale the connector is a secondary thing.  Just as a for instance, this is a well thought of pulse transformer with a 75 ohm resistive loading.  Note that the scales are different - 100mrho vertical and 5ns per div horizontal

 

attachicon.gifRG179 + SC +75.jpg

You're making me regret selling my Tektronix 7854.

Share this post


Link to post
Share on other sites

I'd argue otherwise.  Maybe I'm getting old but I've seen more decent discussion in this thread than I've seen in a long time.  :)

Not sure what you mean. I didn't say anything about the quality of the discussion in the thread.

Share this post


Link to post
Share on other sites

What is the symbol period in spdif?

 

There seems to be an oscillation in the WBT0110 measurement.

 

Would be interesting to see the same measurements on a network analyser

Share this post


Link to post
Share on other sites

Needs the specified RCA's though.  Any other RCA's are lousy from a measurement perspective.  But if the send and receive environment are not 75 ohms on a nanosecond timescale the connector is a secondary thing.  Just as a for instance, this is a well thought of pulse transformer with a 75 ohm resistive loading.  Note that the scales are different - 100mrho vertical and 5ns per div horizontal

 

attachicon.gifRG179 + SC +75.jpg

No not really, the true high quality 75-ohm BNC socket/plug looked just as a crumby as the other WBT socket/plug connection, ie a bad double ringing rather than just a bad single ringing :)

 

I am sure (very certain) that you haven't tried all RCAs :)

Share this post


Link to post
Share on other sites

No not really, the true high quality 75-ohm BNC socket/plug looked just as a crumby as the other WBT socket/plug connection, ie a bad double ringing rather than just a bad single ringing :)

 

 

All pictures are at the same scale.  The first picture of the WBT0144 shows about 100mrho deviation, That indicates that the impedance is not 75 ohms at all over a 1ns timescale - it actually falls to 61.4 ohms (-18%).  The BNC initially drops to 66.5 ohms and then rises to 82.9 ohms (-11.3%+10.5%).  The  WBT 110 plug and 0210 socket socket stays within 1.5 ohms of 75 (+/-2%)

 

So you are right - the performance of 75 ohm BNC's are little better than vanilla or esoteric RCAs (other than the reflections are of shorter duration with the BNC), other than the particular WBT's that are specified at 75 ohms and perform superbly.

 

However, like I said these effects are swamped by the typical implementation of a pulse transformer circuit.  The single image above shows an increase of impedance peaking at 102 ohms (35.3%) and lasting for 5 nanoseconds.

 

Why is this important?  Because rise time aberrations in an SPDIF signal make clock recovery more difficult and adds to timing jitter.  Once you get it right, absolutely right, it makes a major difference to sound quality.

Edited by CraigS

Share this post


Link to post
Share on other sites

OK - here is a pulse transformer with an optimised matching network.  This is on the same timescale and vertical sensitivity as the connector tests.  I have it fed via a 75 ohm BNC, and all you can see is the connector itself - compare and contrast to a pulse transformer terminated in 75 ohms (which is what *everyone* does).

 

I might just remove the BNC and fit the good WBT and see what further improvement that makes

 

post-153047-0-89249400-1449569925_thumb.

 

Share this post


Link to post
Share on other sites

OK - in went the 75 ohm WBT RCA.  Same scale as above.  Convinced?

 

 

post-153047-0-02715700-1449571392_thumb.

 

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

  • Classifieds Statistics


    Currently Active Ads

    Total Sales (Since 2018)

    Total Sales Value (Last 14 Days)

    Total Ads Value (Since March 2020)
  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...