FORGETTING the Digital to Analog conversion part, what is BEST Digital source?

[marce]

Most digital interfaces will do bit perfect these days...

 

Noise is not such an issue with digital, the signal will get through. Noise on systems in general, you need measurements and investigation to see if its a problem.

Jitter is only a problem at the point of conversion, so re-clocking clears that up.

Me I have a NAS that anyone can use to access the music library on their phones, laptops or whatever and a stand alone PC that just runs my squeezebox server, shoved away in a corner somewhere.

From what I have seen of audiophile servers etc. just the same as a PC with fancy clothes, not seen anything exciting or spectacular. I know some will squalk, but I do low noise instrumentation layout, mil, aerospace and medical so I know what it takes to get a really low noise system, whether that level is needed for audio I wonder, look at the figures you can get out of on board sound cards connected directly to a PC motherboard

[marce]

4 minutes ago, beerandmusic said:

 

Ok, i am not wanting to limit this topic to USB only.  I want to know if there is any method to deliver the digital data accurately and if it can be measured such that it is not debatable?  So PC's are out in your opinion to accurately deliver the 1s and 0s...what is your suggestion for perfect digital transmission?

 

I know i can copy a digital file from one computer to another computer and do an md5 on it to verify that the digital contents are 100% accurate.  The fact that you can copy a digital file to a usb drive with 100% accuracy using ANY cable, I don't understand why a digital file cannot be copied to a dac with 100% accuracy?  Maybe Dacs should be built with an internal HD where the source file can be compared to the destination file within the dac and verify that it is working with a 100% accurate file before the conversion?

 

 

 

PC's do perfect bit transmission, no problem what so ever.

Eye diagrams, network analysers, you can even simulate it all beforehand if you want... bit perfect data transmission these days is easy. Its all around you.

[marce]

11 minutes ago, beerandmusic said:

 

So are you suggesting that even using my asus vm60 s/pdif connector, if this digital music file is sent to a w4s dac which has a femto clock upgrade, that the digital data will be 100% accurate regardless of what cable i use?

 

If the cable is within the SPDIF spec then YES.

 

Timing at the point of data conversion is critical, if you buffer and re-clock your data then yes there is no problem. Getting the data their is usually not an issue, just timing when it is converted.

Femto clocks in audio... always gives me a giggle, especially when they are on a separate board with a piece of wire to the main board (all clock mods that involve external wires are not going to do much, I could go on about signal integrity but...)

[marce]

All this is covered by EMC engineering and testing, nothing special. How much noise comes out of your computer is measured and covered by CE FCC etc. 

PC load processing Ethernet or USB is not that different, both interfaces have transceivers that handle the bones of the data transfer... Be it an a PC or an audiophile server, both being digital by nature will have some noise present (with or without SMPS's as the main supply) called simultaneous switching noise (we have to study and look at this stuff for doing really quiet systems).

What worries me is that audiophile DACs cannot handle this noise, a bit worrying that considering the levels they can achieve with sound cards plugged into the computers motherboard...

[marce]

10 hours ago, beerandmusic said:

 

1. When you say "fed by an external ethernet renderer"

would a Synology NAS serve that purpose well?

 

2. and "at some distance" from the DAC, what would you recommend?  Don't they normally recommend short usb cables?  3',  6', longer?

 

P.S.  I do highly regard your input, so please don't think that just because I am questioning you a lot, that I do not trust you... (wink)

The noise travels down the cables, for radiated its not going to come out of the case unless it is a true audiophile design.

[marce]

7 hours ago, beerandmusic said:

if noise wasn't an issue, then there would be no need for "usb toys" or special cables.

It is stated that there is zero issue with the transfer of the digital file.

I don't need a special cable to copy a bit perfect file to an external hard drive via the usb port.

The dacs should be able to implement a way to receive the file perfectly regardless of the cable and discard the noise...so that all is left is the designer's method of guestimating the analog out.

WELL!

[marce]

Galvanic isolation DOES NOT mean noise isolation, many moons ago I posted a layout that shows what has to be done for noise isolation (it was in reference to the Jitterbugs terrible layout) which has been copied once or twice... But galvanic is to isolate direct current flow paths it can still let noise through if not implemented properly.

[marce]

42 minutes ago, barrows said:

 

@marce, on close coupling of components: I always remember John Atkinson's measurements of one of the MSB DACs, which stacked its DAC box directly on top of its separate power supply box: doing so induced AC frequency noise into the DACs analog output (the components in question were designed to be stacked this way), of course you may have been referring to HF noise re: chassis shielding.

Bad design or lack of EMC testing... Not good.

[marce]

15 minutes ago, beerandmusic said:

hmmm...i thought galvanic isolation separated the noisy 5v bus.

If not so in your opinion, what "toy" does work if i have a noisy usb? 

S/pdif sounds great, so maybe a Schiit Eitr, but then i lose DSD capability...

From the internet:

Galvanic isolation is a principle of isolating functional sections of electrical systems to prevent current flow; no direct conduction path is permitted.

 

Even though its touted around a lot on audio sites at the moment on its own it DOES NOT mean noise isolation. I can only comment on the noise isolation if I can see the interface, either a photo of the layout, Gerber information or some other visual check. Off which I would do for free.

[marce]

As Opus has stated the noise will find any capacitance and cross it, we are talking high frequency noise here, so the main coupling is via capacitive coupling.

That said all DACs should be tested for conducted immunity so this should not be a problem, it is pretty easy to engineer the interface to avoid the noise causing problems... if it does then the equipment is inadequately designed....

[marce]

Its conducted noise into the DAC, should be catered for, we do it all the time on a variety of things that are going to have wires plugged into them, audio and instrumentation and of course the mil stuff where the requirements are tighter. Though I do agree with unbalanced it can be a PITA, balanced for analogue, LVDS for digital is best.

[marce]

I would like to see some proof of the differences in noise outputs between standard equipment and some audiophile stuff, as this is the only way to determine how good or bad a device is regarding EMC compliance.

All this talk of noise but NO measurements, the only way to control noise, is know what noise there is, then you can design the correct filters, that's the only way, anything else is guessing.

[marce]

11 hours ago, beerandmusic said:

 

This is the "first" time i heard and understood that.  It was my understanding that the pc will transfer the binary file perfectly and that usb toys, cables, or isolation, would resolve any noise problems.  It doesn't sound "bad", just not as clean as s/pdif does....but if that is the case, then i will just forget about using a pc with usb dac, and look at DLNA or network server for my main system....I never understood before this posting that the noise wasn't able to be managed by dac circuitry.

Again, do some research on how quite internal PC sound cards can be.

And again USB DACs should be able to handle the noise, otherwise their design is flawed, the whole point is plugging a USB port in.

[marce]

Hi Barrows

(I don't just do high speed, I actually do more analogue/digital stuff mainly aerospace/mil these days.

If there are all these differences between systems and it is noise (as I put forward on another thread) then it could be an EMC problem, this would also explain the many differences between systems with different cables, components etc.

What do we do:

Cry...

If you have been involved with any EMC testing  you will know what a pain it is, little things make a difference, the wrong length cable can suddenly become a 1/20th wavelength antenna if the system noise has the correct frequency content. So to solve it we will have to measure every system and permeation of every system we can configure, with every possible cable....

I do think that this could be the only cause, as the previous paragraph shows we are in a bit of a dilemma, what I would look at though is more robust front ends on USB DACs, I do think that many would not pass a conducted immunity test from some of the boards I have seen.

I will have a look at John measurements.

[marce]

10 hours ago, barrows said:

Do you know which is better (I do not).  The tiny transformers in Ethernet jacks, SiO2 chips for USB isolation, Opto isolators for USB, or GMR (basically transformers on a chip) for USB.  Plenty of USB DACs have full galvanic isolation.  None of these things, IME, do enough to make noise a non-issue.

 

My DAC uses GMRs, but it still sounds better when one gives it the lowest noise, best Is USB signal possible.

 

I suspect Beer may actually have some settings in his computer which are compromising his USB performance, this happens pretty easily, especially on Windows.  These will not happen with Ethernet (at least not with DLNA) because the computer is just sending the file and not engaging the onboard sound engine.

Opto isolation would be best.....

[marce]

7 hours ago, jabbr said:

 

 

Yeah I tried reclocking a switch but can't say it makes a difference. If anything a good switch in between a noisy PC is better. There are cheap LPS which are easy to use. 

Why?

[marce]

9 hours ago, NOMBEDES said:

 

@barrows  I see that you are a computer expert:   "Design/Build Consultant with Simple Design /Sonore".  Good on you!

 

What about the rest of us that do not have your education, work experience and skill?  What about the rest of us that had to sell off a W4S DAC 1 because it would quit working with each so called "upgrade" from Cupertino?  My opinion only; "computer audio" is for experts such as yourself, or the serious hobbyist who likes to experiment with all manner of DACs, USB Cables, external clocks, and god knows what else.   I like to kick back play a CD or an LP and enjoy the music.  I am also unconvinced that high resolution files sound much different than redbook.  If a recording is well done, it should sound good on the common CD player and on a high resolution platform.  A poor recording will sound even worse on a high resolution file and given the state of the recording art industry today, you have a rather good chance of getting some horrid recordings.

 

I do have a modest computer audio system which does work, most of the time!

 

Interesting...

One set of views from someone who has a vested interest in selling audiophile gear.  I understand now the emphasis on audiophile gear... whereas some of just work in the mundane world of electronics (albeit often very low noise design).

 

 

 

 

[marce]

13 hours ago, jabbr said:

 

I swapped the clock awhile ago in a very cheap fiberoptic switch ($60-70). Figured that since the chip itself is standard, and the SFP modules supplied by me, that the cost cutting was likely in the crystal. There are reports from hackers that these things matter. At least for fiberoptic, I don't notice any difference, and haven't looked at the issue in several years.

 

I know it shouldn't matter but don't like to dismiss without personal experience.

Interesting, actually I agree, worked on some comms stuff, long distance and there the clock is critical, wonder what difference it would make in a domestic situation... I know for long distance it minimises the number of packets lost as timing becomes critical at the receiving end especially for signals over copper. Re-doing my new house from scratch so instead of basic switches etc. from the local catalogue I'll look round and see what the alternates are. Tried cadging some liquid cooled stuff but they wouldn't give away a free system...

Fibre optic, not an area I have looked at a lot lately, but that has given me an idea. Any info you have on home fibre network would be greatly appreciated. No copper, distributed nodes, no interference, fast... Hmmm

[marce]

15 hours ago, NOMBEDES said:

 

+1.  "The difference between running, say, a standard consumer computer via USB to a DAC, vs a good purpose built audio source (edit) is not subtle."

 

Agree.  The home computer is a poor audio source.   

I do agree using the best low noise source is the optimum way, but isolation is easy, use wireless or an optical link and you can totally isolate the main source, be it PC, NAS whatever. Then have a simpler set up at the DAC to receive the data and transmit it to the DAC. 

 

[marce]

11 hours ago, bobfa said:

There are so many places for signals to go wrong.  Here is an interesting reference to watch:

 

 

 

Do a search for "signal integrity" "Eric Bogatin" "Dr Howard Johnson" "henry Ott" "Ralf Morrison" and you will find there is so much experience and information about getting the signals right, just watching that video on different continents shows how far we have got with digital signal transmission.

[marce]

18 hours ago, seeteeyou said:

 

http://asoyaji.blogspot.com/2017/01/duculon.html

 

 

Not to be picky but there is no use using a super duper 40+MHz crystal, then putting in a huge impedance mismatch with the connectors used to pass the clock signal to the target board, at the very least SMA connectors or similar should be used as Sotm do. The way that is wired up negates any benefit of the clock. The return  (shield) of the co-ax has been split of and made into a tail to fit in the connector... Attention to detail is critical with clock signals, not just a fancy crystal, the whole signal path is critical and the shorter the better (ie the crystal next to the device it is clocking).

[marce]

17 hours ago, barrows said:

@marce, I am in the process of putting together a new (DIY) DAC.  There is separate re-clock/isolation board for the USB interface, which can provide masterclock to both the USB interface and the DAC (chip, on a separate PCB).  The masterclock signal is distributed to the DAC board via a short u.fl coax cable, with proper micro BNC jacks on the boards.  Considering a cable length of <10 cm (and assuming good PCB design and trace routing) is u.fl appropriate for masterclock distribution to the DAC?

I understand that having the XO and final re-clock (POTATO FFs) on the DAC PCB and sending it back to the USB interface would be best, but I do not have that option.

And, BTW, always appreciate you being "picky"!

Yes is the short answer, sent you a PM with a longer answer.

[marce]

12 hours ago, beerandmusic said:

 

marce, you seem to have a lot of knowledge, and seem to be objective more than subjective in thinking (smart budget minded), and dont buy into marketing hype....I am just curious what you use for your system?

 

If you are going to replace a clock with an off-board clock then you need to do it correctly, there is no use doing half a job, getting the clock signal from the clock to device is critical, get that wrong and you have wasted your time, signal integrity wise, so no point doing it in the first place. In the case illustrated the wiring was not optional so any benefits of the clock were negated by the signal path.

Went into detail on Cornans thread I believe, at the moment some ATH-AD900X plugged into whatever is handy, computer, squeezebox...

Got some home made OB's, some Curvy Changs and some KEF 104/2s  floating around, a few other bits of kit, audiolab a pair of SET class A valve monoblocks. Most stuff is in storage whilst I sort out my new house, once that is done... dunno yet still deciding, some Magnepan's or more likely ATC SCM-40A's.

[marce]

1 hour ago, barrows said:

No.  Remember, my example was limited to only one part of the picture.  To make a high performance product one applies this approach to every aspect of the design, so there would be very many other "upgrades".  Typical part cost to retail price ratios for manufacturers are anywhere from 5x to 10x the part cost.  For example, say you have the option of using a fifty cent regulator or a 5 dollar regulator in part of a product.  The fifty cent regulator adds .5*5=$2.50 to the retail price, the 5 dollar regulator adds 5*5=$25 to the retail price, now consider that you might have 30 of these in the product, now you have added 25*30=$750 to the retail price.  But this is just for regulators, if your product is going to be the best it can be, or even close, most every other part would also be upgraded with the associated costs.  And, the better regulators typically require additional surrounding components (usually capacitors) to work, so that is more cost.  These are just examples, there is much more to it.

Again, i suggest you take a look at the cost of high performance measurement gear from Audio Precision, as an example of what no BS high performance electronics really cost.

 

Now I am personally not so sure about products costing >$10K or so... and when we get to six figure things like dCS I just do not know.  but the $5K to $10K DACs out there seem reasonable to me (not all of them, of course).  You just are not going to get the best possible performance at lesser costs for a manufactured product.

I totally agree here, there is a relationship between a product and its BOM up to a point. In barrows case its a well engineered product with the best parts for that level of product. Top end engineered designs cost money, I know because I work on a lot of no cost designs where failure is not an option, be it life or mission critical. I have worked on all levels or products over the years from commercial where price is a big factor, consumers don't want to pay a lot for their toys but expect all the trimmings, to billion dollar projects such as flight refueling, AN-VIC-5 and other stuff that I can't mention..  Just having some proto-type flexi-rigid PCBs done amere snip at about £5K each, plus the cost of design, we have been working on and off on this project for 2.5 years. The board I am working on now is only 150x250mm it will take me a month to lay out at least,  plus the rf engineer and EE's time on the project...

[marce]

35 minutes ago, beerandmusic said:

 

you understand niche boutique pricing/marketing, not volume pricing/marketing.  My suggestion is the bigger players understand bigger dollars in volume, where profit margins don't need to be so high.  You are free to believe whatever you want as well, and that too will not make it so.  I am offering you answers from consumer's standpoint and where the real money is, is in volume, not niche marketing.

There is not the volume in audio, high volume these days are assembly runs off 5,000+ units and even that is pretty small, some consumer goods are made in batches of 20,000 up. Try negotiating deals for components, big outlay, all to be catered for in the final price, never mind cash flow a big issue with all businesses.