Head-Fi.org › Forums › Equipment Forums › Sound Science › Quality of Digital-out is always the same regardless of manufacturer?
New Posts  All Forums:Forum Nav:

Quality of Digital-out is always the same regardless of manufacturer?

post #1 of 25
Thread Starter 
Hy,
in an ideal world every soundchip should give the same results if he processed a digital signal.
For a PC it shouldn't matter if I bought a high-quality soundcard or a cheap Asian model - because they just pass the signal through their digital-out.
Since we live in the real world, I wonder if there is some limitations.
Is a digital-out signal always the same regardless of the manufacturer?
Could I conclude that I could buy the cheapest available soundcard with SPDIF or COAX and hook that up to an expensive DAC with no loss?
post #2 of 25
In a perfect world, my Geo Prism would run with me neighbor's F430...Yeah...



Anyway, you should check out the Computer-As-Source Forums:

Computer Audio - Head-Fi: Covering Headphones, Earphones and Portable Audio

NK
post #3 of 25
In a broad sense, no, not always.

In a more esoteric, audiophillic sense, not really.

Some of the 'cheap asian chips' in computers have the s/pdif output clock locked at 48khz. This is a relic of a microsoft specification that is more than 10 years old.

Contrary to popular delusion, manipulating the sample rate can only degrade the purity of the digital signal.

Increasing the sample rate by a large margin can make the DAC produce less noise, and the benefits outweigh the distortion added by the resample, provided the margin is at least 2x and the resample is done in a proper manner. The ESS Sabre DAC uses this effect to it's benefit, resampling to truly crazy-high rates.

Increasing the sample rate by a thin margin introduces aliasing, which is very bad. I'd rather listen to a cheap dac with a straight 44.1khz signal than an expensive dac with 44.1 upsampled to 48.

A lot of arguments are made about coaxial vs. optical, but they're all based on research done on the state of the art in optical transmitters in 1992. optical transmitters are way better now. These arguments are now mostly horsepucky.

Not all coaxial links are transformer coupled, and this can cause either or both ends to misbehave in ways that are worse than an analog link. A 120hz (or 100hz if you're in europe) ground loop overlay can wreak havok on your nice clean digital transmission.

If you're up for a little DIY, you can add an inline pulse transformer fairly easily.

In the more esoteric sense, there is the problem of jitter. This basically refers to the bitstream not streaming at a constant rate, potentially with little gaps in it.

20 years ago, the major source of jitter was CD and DAT transport mechanisms. It was a mechanical issue as well as a matter of the speed and quality of the circuits around the transport.

With good transports, memory based sources, and modern technolgoy, jitter isn't nearly the problem it was 20 years ago.

The amount of source jitter you typically get now is usually within the means of clock recovery in the DAC.

Beware of people making fuzzy-headed arguments about the tonal qualities of physical s/pdif or toslink transmission hardware. The physical layer doesn't know when the data words begin or end, and is thus unable to discriminate in it's offenses.

Also beware of people who mix metaphors between analog systems and digital systems when describing the potential pitfalls of the digital connection itself. They're talking out their posteriors.
post #4 of 25
In addition, Toslink transmitters and receivers being "underpowered" which I mean supplied woithout sufficient supply voltage decoupling, produce jitter and errors by themselves. This is actually very common in budgetary and mid-end digital audio devices. On the other hand, some tweaks might improve an entry level CD or DVD in terms of being a reliable digital transport. However, in most cases, you don't get the 100% 1's and 0's delivered unchanged and without fluctuations.
post #5 of 25
Quote:
Originally Posted by majkel View Post
In addition, Toslink transmitters and receivers being "underpowered" which I mean supplied woithout sufficient supply voltage decoupling, produce jitter and errors by themselves. This is actually very common in budgetary and mid-end digital audio devices. On the other hand, some tweaks might improve an entry level CD or DVD in terms of being a reliable digital transport. However, in most cases, you don't get the 100% 1's and 0's delivered unchanged and without fluctuations.
Yeah. But all that crap about the optical transmitter not having enough bandwidth for the audio signal is utterly meaningless in 1999, let alone 2009.

I'll admit: I use transformer-coupled coaxial 'cause it allows me to easily have a 3-meter link over reasonable-quality coax.

Optical can have only two advantages.

1: No potential for ground loop 'cause there's no electricity.

2: Maybe you ran out of coaxial inputs on your dac, or maybe the source only has an optical output.

It's neato, but, i can't make the cables at home without expensive gear.
post #6 of 25
Thread Starter 
Thank you all for your infos!
I dont care if it is optical or Coax since I dont believe that my ears are that superior to be able to hear the slightest jitter.

From what I have read now, I deduct, that the digital signal is NOT just passed thru, but rather processed and therefore depends on the quality of the soundcard/ soundchip.
That means the higher the reputation of the manufacturer the better should the sound be.

Currently I use a C-Media card. I think its a CMI-8738 or something.
Now I need a USB-Soundcard. My Denon amplifier handles both Optical and Coax- so I wondered if I could just take the cheapo and am set.
But I would like to make my own buying decision based upon scientific facts instead of: 'I feel that this card sounds good- I recommend it to you!'. You understand what I mean, right?

So it is really ONLY about the card. That it also depends upon the CD-Drive, cables connected etc... that's something later will be taken care of.
post #7 of 25
Sure, but there are no ideal world...
post #8 of 25
Quote:
From what I have read now, I deduct, that the digital signal is NOT just passed thru, but rather processed and therefore depends on the quality of the soundcard/ soundchip.

But what about transports with bit-perfect output? They should pass the signal unaltered by any DSP the soundcard might have?
post #9 of 25
Quote:
Originally Posted by musician View Post
From what I have read now, I deduct, that the digital signal is NOT just passed thru, but rather processed and therefore depends on the quality of the soundcard/ soundchip.
That means the higher the reputation of the manufacturer the better should the sound be.

Currently I use a C-Media card. I think its a CMI-8738 or something.
Tell me, what have you read that leads you to believe that, for example, the s16le pcm stream from a ripped CD is processed on it's way out the digital output on your c-media card?

Keep in mind that the same hardware is used to output compressed dolby digital & dts signals, and that processing these in any way, without a software tool that is aware of the compression, would probably render them mute.


Quote:
Originally Posted by musician View Post
Now I need a USB-Soundcard. My Denon amplifier handles both Optical and Coax- so I wondered if I could just take the cheapo and am set.
But I would like to make my own buying decision based upon scientific facts instead of: 'I feel that this card sounds good- I recommend it to you!'. You understand what I mean, right?

So it is really ONLY about the card. That it also depends upon the CD-Drive, cables connected etc... that's something later will be taken care of.
Well, the cheapest i know of is the turtle beach audio advantage micro - which comes with optical out. It's also based on a cmedia chip. I don't know how well it's digital output works. I have one, I've just never tried it's digital output.

The TBAAM's analog performance is . . . from a technical perspective, well, poor. It measures badly. But people say it sounds good. Go figure.
post #10 of 25
Thread Starter 
Quote:
Originally Posted by ericj View Post
Tell me, what have you read that leads you to believe that, for example, the s16le pcm stream from a ripped CD is processed on it's way out the digital output on your c-media card?
You yourself wrote in a very mystic way
__In a broad sense, no, not always.

In a more esoteric, audiophillic sense, not really.

Some of the 'cheap asian chips' in computers have the s/pdif output clock locked at 48khz. This is a relic of a microsoft specification that is more than 10 years old.__
Which means that it IS processed/ altered on its way out.
I have chosen to go with an M-Audio Transit regarding the USB output.
Still I would like to have some more insight in the way digital sound is handled from the source to the digital signal. Any insights?

TIA
post #11 of 25
Quote:
Originally Posted by musician View Post
You yourself wrote in a very mystic way
__In a broad sense, no, not always.

In a more esoteric, audiophillic sense, not really.

Some of the 'cheap asian chips' in computers have the s/pdif output clock locked at 48khz. This is a relic of a microsoft specification that is more than 10 years old.__
Which means that it IS processed/ altered on its way out.
I have chosen to go with an M-Audio Transit regarding the USB output.
Still I would like to have some more insight in the way digital sound is handled from the source to the digital signal. Any insights?

TIA
A lot of sound cards, particularly older sound cards, process it by default.

Primary examples would be nearly all motherboard sound chips prior to about 2005 (and still some of them today), and all of the SB Live and Audigy boards. Possibly not Audigy 2.

I believe that your CMedia card is among those that pass the stream unmolested. It is, on paper anyway, bit perfect.

Cards that don't process the stream are common, and not hard to find, but you do have to actually check.

The ALSA sound card list, even if you don't use linux, will generally say if a card has built in spdif and whether it's locked at 48khz.

For motherboard audio, you just have to read the manual and find out what codec chip it uses, and then look up the specs for that codec. I used to have an HTPC running on a board that hit the market in late 2002 that had bit-perfect on-board spdif.

Another budget option for you, if you're sure you need usb, is to troll ebay for a used M-Audio Sonica. Not the Sonica Theater, but the regular Sonica. This has an optical output as well, and MAudio makes pretty solid gear. Unfortunately you don't see the Sonica come up all that frequently.
post #12 of 25
Quote:
Originally Posted by ericj View Post
This is a relic of a microsoft specification that is more than 10 years old.
Isn't it because that is what DVD uses? I leave my soundcard at 48khz even though it can do bit perfect 44.1khz because after experimenting with sampling rates I just don't hear any difference. Whether it is a measurable difference matters not if I can't hear the difference.
post #13 of 25
all depends on the equipment.

my on board digital out is stuck at 48khz and is full of distortion if i listen to anything other than 48khz

had to stick to the m-audio in order to get the bit perfect output to my dacmagic.

recently purchased a QED coax cable, gonna test it against my monoprice optical
post #14 of 25
I have both optical and coax going to my DAC from 2 computers and I don't hear a difference. One soundcard is Chaintech AV-710 and the other is X-Fi.
post #15 of 25
Quote:
Originally Posted by milkweg View Post
Isn't it because that is what DVD uses? I leave my soundcard at 48khz even though it can do bit perfect 44.1khz because after experimenting with sampling rates I just don't hear any difference. Whether it is a measurable difference matters not if I can't hear the difference.
I don't remember what microsoft was on about with the ac97 specification. DVD may have had something to do with it. More than likely it was about being lazy and cheap. There were probably long boring meetings where people in suits droned on about business-class audio or some nonsense.

If it was about DVD support, there could not have been any reason why codecs that are strictly AC97 compliant resample everything going out to 48khz.

If it was about making cheap southbridge audio interfaces that were wire-compatible with dozens of different cheap codec chips from dozens of vendors with a software interface where you push whatever the heck you like through the controller and don't have to mess about with selecting the correct signal clock, suddenly it all makes sense.

Especially if you consider that 48khz is the maximum clock in the ac97 spec. upsampling everything to the max is good, right? RIGHT?

It results in playback that is reliably mediocre with a minimum of effort for the chipset makers and OS coders. Why pci sound card manufacturers jumped on the bandwagon is a mystery. Maybe they figured it guaranteed easy driver work and cheap codec parts.

Which brings up a point - most old motherboard audio does a terrible job of resampling in hardware and you get better results if you use a high quality software resampler.

The SB Live was criticized for not resampling to audiophile quality. The SB Audigy also resamples all output to 48khz - it just does a far better job than the Live, and lightyears better than most old realtek chips.

It's probably a happy accident that you can stream dolby digital over these interfaces.

What i don't understand is why many AC97 chips can receive spdif at 6 or 8 different sample rates but only send at 48khz.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Quality of Digital-out is always the same regardless of manufacturer?