Can System Noise be saved to HDDs, SDs, etc. in digital audio files?

[marce]

10 minutes ago, Blackmorec said:

Ralf11 et al. There’s a lesson here. Look at the above thread. With respect to ALL posters,  just another stinking pile of acrimonious garbage. A shame because the original question was interesting, is not well documented or researched, yet can have tremendous impact on all our systems, where buffering is implemented as a means of preventing noise travelling through a system. 

You may have alternate views on things but your name (at least in my head) has become synonymous with flippant, glib, silly answers that contribute nothing but disrupt threads....just another form of noise.....Forum noise, if you will. 

So why not turn over a new leaf....if you do or don’t believe in something AND you have good solid reasons, facts, evidence and experience not to, then post your reasons and add to the value of discussions, rather than simply disrupting and polluting threads with worthless one- and two-liners. 😉

And its not Ralf11 alone...so et al, please take note....you know who you are. 

The facts of the matter are that system noise is not stored with a digital file...  Unlike recording an analogue file where superimposed noise on the signal is recorded as part of the signal.

As to researched, system noise and its effects are extremely well researched, I posted some links on the thread this has evolved from. Maybe some basic research on signal integrity/EMC would be enlightening. Of course to say this has no bearing on audio would be silly, Henry Ott (a respected expert on EMC) has done AES presentations and his book will be found in the reference section of many books on low noise design). Then there is the information all over the IC manufacturers sites covering low noise design. When people are using femto ampere op-amps noise can be a big issue, again what does have to do with audio, well the techniques learned doing the realy sensitive stuff feeds down to other electronics. And so as we push the boundary our knowledge expands.

Buffering is not implemented to prevent noise travelling through a system, isolation using opto isolator, isolation IC's or some other form of galvanic isolation to isolate two power domains completely and prevent noise currents flowing between the two domains. As a lot of this noise is high frequency care has to be taken to avoid capacitive coupling between the power domains. A non galvanic basic buffer has the same noise present at both input and output as they share the same GND, internally. 

[marce]

1 hour ago, sandyk said:

 

 An appeal to Authority that you don't like from others.

 

So what ?

6 correctly performed separate DBT sessions say otherwise, as do several more prominent and appropriately qualified members  in the last 4 1/2 months alone.

Two of them are also way more qualified in their respective areas than an amateur like yourself.

No a reference of information, why can't you discuss this like an adult...

 

[marce]

45 minutes ago, Blackmorec said:

Let me summarise your post as a 1 liner:

We don’t know how it happens; so it doesn’t

 

That about right? 

 

No that is totally wrong, we know it doesn't happen, if you have any proof other than some anecdotal evidence or even some idea of mechanism that could cause noise to be recorded on 1 bit at a time please provide it.

49 minutes ago, Blackmorec said:

 

When digital was first released, it was pushed as “perfect sound forever”. That was because the ‘noise’ that had plagued analog from the get go was no longer transmitted and could be entirely excluded from digital recordings and replay. So job done. Noise excluded. Noise was no longer a problem

What we didn’t know at the time was anything about the deleterious effects HF noise had on all digital replay. We simply didn’t know.  It was there, it negatively impacted sound quality but we didn’t know about it. The first indication came when people heard it and couldn’t explain why digital had an edge, a hardness, harshness that became known to this day as “sounding digital”  

 

 

Digital noise was known and understood then, so noise was still an issue in the design process.

 

53 minutes ago, Blackmorec said:

 

 

Now in your post, we’re back where we were 50 years ago. The HF noise is excluded so we don’t have a problem.  Yet audiophiles the World over clearly hear that making improvements at the front end of the chain brings SQ improvements from the speakers, despite bit perfect files along the whole chain.

 

The thread is about noise being recorded with the digital data, nothing to do with system noise or any claim that noise is not an issue, it is dealt with during the design process. The issue of noise in systems is a different topic.

[marce]

33 minutes ago, sandyk said:

 

Agreed.

However, if noise was embedded in the file we wouldn't be able to REGENERATE it to very close to that of the original, which very few of us are able to verify anyway, unless we have access to the original Master copy, whether  Digital Audio  or Digital Video.

  Many members use a USB Regen or ISO Regen to do this with USB audio.

 IOW, improved Signal Integrity.

 

14 minutes ago, PeterSt said:

Let's hop over to the next incarnation of whatever has been an audio forum.

 

B/S

 

We can predict what is next.

Excellent contribution to the thread.

It is a claim that has been made regarding audio files with the same check sum, so bit identical, therefore discussing it on an open audio website seams appropriate.

[marce]

2 hours ago, sandyk said:

 

 Simply giving a reference to a Textbook is NOT a discussion.

 Even the Textbooks used in Primary schools get regularly updated.

Usual pathetic answer, don't bother unless you are going to contribute to the thread in a more positive manner.

[marce]

1 hour ago, sandyk said:

 

 Some can even make audio sound a little different by changing a  simple setting in the player S/W  while the output remains bit perfect.

bit perfect to what or are you claiming two identical digital files contain different information.

[marce]

1 hour ago, Blackmorec said:

Are you unfamiliar with the concept of drawing parallels? 

The idea that 2 distinct things, while different are clearly very similar in many respects? 

 

Nope...

[marce]

How would the data read modulate the jitter or rise time of the signal. to all extents the data would be identical to the system if it were operating at such a  level where system noise was that critical the system is broken.

[marce]

2 hours ago, jabbr said:

 

The system is working as described. 

 

For example in a PLL or any error feedback system, the further away a signal is from the center, the larger the error signal: is that a voltage or current? Power draw? Consider a situation in which the error correction circuitry requires current draw.

 

 

in normal operation this current draw is pretty steady and should again be catered for in the design of the storage media, though erratic current draw due to errors is usually a sign of imminent failure.

[marce]

42 minutes ago, manisandher said:

 

Sure. It seems to me that it's the most likely mechanism responsible for audible differences at the point of D-to-A conversion too. But yes, not the topic of this thread.

 

Mani.

agreed and this could be a thread on its own, and it could be interesting discussing the issues and what is out there in the world of D/A and A/D, I would be interested myself if it was a civil thread.

[marce]

37 minutes ago, jabbr said:

 

Is the current draw steady? Of course it’s not variable  to the point where there are bit errors, but in CMOS electronics, there is switching noise and that doesn’t cause bit errors either. One could use constant current logic, or ECL which is balanced and this is used for high speed designs, but for the everyday CMOS/TTL logic, there is switching noise. I’m not suggesting that this doesn’t all work exceedingly well, rather that the systems work well despite the noise.

 

Again, none of this noise survives copy. None of this noise should be considered embedded in the logical disc.

Steady in that the current profile over a certain time is steady, there are going to be small variations, but for reliability the PDS is designed to handle the maximum current and then some. If you do not cater for max. current draw the supply will dip, power supervision IC's will start getting twitchy and things will go downhill fast... I would not be suprised if all disk manufacturers these days use MTBF software and design for reliability, Amstrad and Seagate provided some incentive.

[marce]

29 minutes ago, John Dyson said:

Any device in the system can produce noise.  As a circuit designer, one has to remember that even an old CMOS gate doesn't draw a continuous current.  The current draw tends to spike during transitions.  As was mentioned before, there is a logic family that has less 'glitching' -- an ECL derivative.  They tend to be semi-balanced circuits --- similar to a differential amplifier -- and even though they are fast, tend to draw more constant current.  Some of the higher speed MOS and schottky technologies produce lots of glitching, and requires careful consideration for bypassing and ground current.

 

So, chiming in on HDDs or any large scale circuit or HW, their current needs will tend to vary significantly.  Any time there is a variation in current requirements at the speed near the audio frequency range (which means pretty much all frequencies above DC), the change in current draw can be reflected into the audio signal.   Higher frequencies are more challenging because of the typical high currents needed to charge even small capacitors (that is, large current surges), and the surges are very fast.  The voltage developed across even a short conductor when the current surge is fast will be significant.   It isn't necessary that the single current surge itself be directly in the audio frequency range, because the pattern of the surges can average into the audio range.  Also, the surges*impedance can directly cause nonlinearities in analog circuits.

 

WIth todays circuitry, where it is so very dynamic -- and even HDDs power up and power down -- it is important to make sure that power supplies are stable, that the power sources have low impedance, and the ground currents are not reflected into the analog audio signal.  These are wickedly more tedious things to do nowadays instead of back in the 1970s/1980s when multi-amp current surges only happened in the audio power amplifier.  Nowadays, digital circuitry (esp CPUS -- HORRIBLE) can surge relatively high currents for short intervals.  (I was amazed when the DEC Alpha chip drew maybe something like 10A back in the '80s.)  Now, chips surge in the 100's of Amps -- that is a lot of opportunity to create noise.  Even little (but fast) digital circuits can do some serious glitching.

 

So, anything that changes current draw (including fast current gltiches in logic chips) can create audio noise.

 

John

which is all catered for with today's power delivery systems, otherwise , especially if the noise levels required are minute.

[marce]

11 hours ago, John Dyson said:

 

 

 

 

There is often no one single solution to noise problems -- the problems can be a combination of a lot of botches, and each noise issue itself can be a combination of botches.  (Okay, not so severe as to be a true 'botch', but rather non-ideal design.)

 

John

Nope, there is engineering solutions, there is power and signal integrity, there is EMC engineering there 

 

11 hours ago, John Dyson said:

So, the current handling (for circuit stability, let alone ground/power noise effects) has to be managed at both the macro and micro scale.  A 'perfect' power supply will not necessarily fix supply/ground current noise problems in a circuit with non-ideal layout.

 power supply! you have to design a power delivery system, that include decoupling capacitors. A bad layout or a non ideal layout these days should not happen, unfortunately sometimes it does, the problem has to be sorted or the kit will not function correctly.

 

11 hours ago, John Dyson said:

For example, a traditional CMOS gate might draw uAmps when on or off.  As soon as you change the state, then the current will glitch instantaneously.  If you change the state repeatedly, then the average current usage will go up rapidly.   Back in the day, people would be surprised that their CMOS circuit might used more current than an equivalent LS circuit.  The current draw depends not only on the bias currents, but also the frequency of state change.   All of the various current draws in a segment of a circuit can start merging into a larger scale change.  These larger scale (multi-chip, not necessarily whole board) changes can cause audio interference just as the little glitch does.

 I mentioned simultaneous switching noise earlier, plenty of info on its effects and how to solve them.

 

11 hours ago, John Dyson said:

There can even be troubles if a design is too periodic in physical structure -- this stuff is also not easy to point-to-point wire (in fact, would be rather ludicrious to point to point wire most circuitry nowadays.)   Look at the 1st old color TV designs -- they make me cringe, but they worked (sort of.)  :-).

Point to point is never used these days for today's design, multiplayer PCB's are the way, at least 4 layers allowing at least on contiguous GND plane for return current paths.

Analogue TV's have been superseded by digital TV's both NTSC and PAL had issues, digital is far better and solved most of the issues.

 

11 hours ago, John Dyson said:

he term 'holistic' approch comes to mind... YUCK, but unfortunately true.  You'll sometimes see a pattern of bypass capacitors being used along with careful grounding design.  These all make a symphony that improves the noise performance without causing other troubles.  (Just throwing large capacitances around not a good thing, just like not enough is also not good.  People get paid to understand these things and solve these problems.)

Again, there is a whole wealth of information on PDS's, decoupling capacitors, how to place decoupling capacitors, creating planar capacitance, controlling the noise within power islands. All these issues are solvable with good engineering practices, there are far more sensitive analogue/digital/analogue than audio done, there are many techniques you can use in the design to mitigate and control the designs own inherent noise. Understanding that current travels in a loop and will always try and get back to its source, and thus understanding return currents is a start, same with the current in digital switching. Same with PCB's, construction methods (HDI), materials, manufacturing have all improved over the years to cater for the higher speed designs. The majority of boards I work on are 8 layers and up, more often 12 these days as more GND planes are used, some designs will be nearly 50% GND planes and some RF/microwave may be up to 75% of the layers a GND plane. 

Doing a single DAC design these days is heaven, a single converter makes it far easier to isolate the digital side from the analogue side, I would imagine most modern well designed DAC's have low noise...

 

An interesting article on current returns amongst other stuff.

http://www.x2y.com/filters/TechDay09kr_hpa_Track2_1_Precision_Analog_Designs_Demand_GoodPCBLayouts _JohnWu.pdf

 

[marce]

10 hours ago, sandyk said:

 

Unlike some, I don't keep referring people to textbooks by Henry Ott etc.

Not just textbook, there is ongoing research into all the issues, it is you who chooses to ignore the information, why I don't know... All you do is post negative comments about textbooks... Sad

[marce]

If my posts are against the forum rules I would expect Chris to reprimand me...

Again you don't answer in the spirit of the thread, but derail it by yet another angry rant and personnel attack. You act like the forum is your own personal domain and spend all your time telling others what they post is not acceptable here...

 

1 hour ago, sandyk said:

It would be a shame if just for once that you said something directly on the subject instead of referring non technical members  to a textbook that they would need to purchase, perhaps to refer to just once, then put it back on a shelf or in a cupboard never to be used again.

You don't have to buy the textbook, there is information all over the web, including AES presentations. 

 

[marce]

35 minutes ago, John Dyson said:

Note when I was speaking of power supply -- i was answering someone who had implied that a central power solution can fix the problem.  Believe me -- I understand the issue of bypassing, but it also cannot just be done willy-nilly.  For fun, do a nice, orderly layout with totally symmetrical bypass organization -- sometimes works, sometimes disaster.  (A totally periodic design can really cause troubles if done imprudently.)

 

I have read a lot of so-called experts that totally miss the point of bypassing.

 

When I was speaking of point to point -- I thought that I denigrated it?  Amazed that they could get color TVs working with it.

 

Regarding more sensitive than 'audio' -- I am speaking from an RF perspective -- essentially the same requirements as needed by audio.  I love using the fact that RF uses 12-14bit a/d that provides more dynamic range than needed by audio -- esp when people claim that greater than 16bits is needed for listening.  (Audio processing is a different story -- but that is off subject.)

 

John

I did all the UK City and Guilds exams in analogue TV at the local tech collage on a night, just passed part 3 as digital YV,s came out... LOL

We use this for critical power:

https://www.quadrasol.co.uk/eda/eda-products/cadstar-power-integrity-advance/

 

the rf/microwave stuff is usually comes from microwave office or similar and is transferred to the PCB as a DXF.

Decoupling is fun, with BGA's tend to use caps directly on the pads on the opposite side with filled through via's, its all in the detail and getting a low impedance loop through the cap and pins. I have seen decouplers at the edge of a BGA, connected with 0.1mm tracks several mm's long!

[marce]

Awesome, of course with correctly trained pigeons we could have parallel data transfer, what worries me is the effect of feed on the data, i have heard that some have experienced data differences, with the same leg tags! between corn fed and free range fed birds... Cher Ami has said in double blind feeding sessions she noticed the messages had more "chirp" when powered by a corn fed chicken. Scientist are working on synchronised pigeon data transfer and believe with some genetic engineering and a physic master clock, data speeds of at least 5 bits an hour will be achievable  by 2034. PAC, pigeon to analogue trials were unsuccessful, luckily Duck to analogue (DAC) tests are showing good results...

So good night and to sign of in QuakerCluckerdecimal:

Quack quack no-quack quack cluck cluck no-cluck quack cluck!😀

[marce]

17 minutes ago, esldude said:

If any fans of Sean McMullen haven't read his Great Winter trilogy you might like it.  An old war left orbiting satellites that vaporized any electrical devices that try to function on earth.  In time computers are developed that don't use electronics or electricity.  Initially trained people do all the functions in a modern computer.  This idea is developed pretty far along.  Used by kings in war, signals from data centers transmitting via networks of blinking lights (non-electrical lights like old lighthouses use).  Its good steampunk post apocalyptic  fiction if your taste runs to such.  

 

Added to wish list on kindle, it £15 for the trilogy, looks good. I'm a big fan of Peter F Hamilton, but also love the classics I can get on kindle, my number one technical piece of kit, it lives with me.