Which sound card has an SPDIF output comparable to an expensive CD Player?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
analog_sa

analog_sa

Member
Joined 2002
momitko said:


I was of an opinion that the interference caused by this kind of PSU must influence only the analog part in the sound card without degrading the digital part.
Could this not be so?
Click to expand...


The digital part is just as sensitive to PS induced jitter and noise, both radiated and transmitted along the PS lines. You can improve the shielding of the soundcard as well as the power. Additional decoupling, chokes and regulators, especially for the clock circuit will no doubt help.
 
P

phofman

Member
Joined 2005
Decent cards isolate their coax SPDIF output with a transformer (the black little box next to the SPDIF cinch). I have not seen the transformer on motherboard-embedded cards. Supposedly it can be made diy:

http://www.hardwarebook.info/S/PDIF_output#Isolation_transformer_for_S.2FPDIF_output

The Terratec card uses the same chip as Chaintec 710, i.e. ICE1724/Envy24HT. Some cards push SPDIF output from the chip through a few logic gates in parallel as an output driver, before feeding the isolation transformer (e.g. Prodigy 192).

Noise from the SMPS affects SPDIF output as well - it is all about exact timing of the edges (i.e. clock jitter, jitter in logic level detection caused by power supply noise). Envy24-based cards differ in quality of power supply stabilizers and filters. I guess these factors constitute differences in SPDIF output quality of the cards based on the same chip.
 
P

phofman

Member
Joined 2005
Hezz said:


Yes, I think it would. Also, if you can keep the PC with only a 1 meter SPDIF cable this helps also. I imagine PC SPDIF jitter is orders of magnitude higher than in a CD transport even though it is reclocked in memory because of the crazy way the signal gets to the sound card. This is one reason USB is supposed to be better as it has a shorter path from memory to the USB port.
Click to expand...

PCI sound cards have their own crystal clock which basically controls the pace of the upstream flow. The timing of data before the sound card has absolutely no impact on the output jitter, as long as enough data is available in that part of memory which the card reads itself via Direct Memory Access. The card reads some data and throws an interrupt, telling the driver/OS that it has taken its data (food). The job of the software/OS/driver is to put enough food to that part of memory before the card runs out of food. If it happens, you get an audible glitch/dropout/xrun, however you call it. Some cards can be configured to throw interrupt up to every 2 seconds which gives the system plenty of time to provide enough data for the next cycle.

On the other hand, the majority of USB cards running in adaptive mode do not have a completely independent clock. They cannot influence the pace of data coming from the USB controller and have to fine-tune its clock via PLL to align with the incoming data flow, adding to the jitter. Technically, adaptive USB is worse than PCI. Only asynchronous USB cards can tell the USB controller to speed up or slow down, thus allowing them to have independent low-jitter clock. The only asynchronous USB card I know about is E-MU 0202/0404.
 
P

phofman

Member
Joined 2005
SndScape Cinema along with SndScape Odeum are based on ICE1724, have two crystal clocks (for 44.1kHz and 48kHz families). Therefore, they are able to output SPDIF up to 192kHz, unless their windows drivers would be for some obscure reason limited to 48kHz which I honestly doubt.
 
M

momitko

Member
Joined 2008
Decent cards isolate their coax SPDIF output with a transformer (the black little box next to the SPDIF cinch). I have not seen the transformer on motherboard-embedded cards. Supposedly it can be made diy: http://www.hardwarebook.info/S/PDIF...S.2FPDIF_output
Click to expand...

Well, I have an SPDIF input transformer in the external DAC rightr before CS8412. Although my Terratec card only has a Toslink SPDIF output and input some time ago I made a DIY coaxial output like the simple schematic shown on the web page that you refer to in order to find out whether or not coaxial sounds better than optical. At first time it seemed to me that the sound improved but then I noticed that the sound became a litle bit more irritating and I removed the coaxial connection and switched to Toslink again.
To my ears Toslink sounds a little bit more natural without irritating digital artefacts.
I believe that SPDIF transfomer is not a great solution for PC because it does not isolate PC ground from the DAC's ground and all HF interferences on PC ground easily migrate to the DAC's ground.
I believe, that's what I heard in my experiments.
It seems that coaxial is very bad for sound on PC and should not be used there.
 
M

momitko

Member
Joined 2008
SndScape Cinema along with SndScape Odeum are based on ICE1724, have two crystal clocks (for 44.1kHz and 48kHz families). Therefore, they are able to output SPDIF up to 192kHz, unless their windows drivers would be for some obscure reason limited to 48kHz which I honestly doubt.
Click to expand...

When I was looking at the specs I didn't see 44.1 anywhere on the page. That's why I made an assumption that the card does not have 44.1. Honestly, I have no idea.
 
P

phofman

Member
Joined 2005
momitko said:


I believe that SPDIF transfomer is not a great solution for PC because it does not isolate PC ground from the DAC's ground and all HF interferences on PC ground easily migrate to the DAC's ground.
I believe, that's what I heard in my experiments.
It seems that coaxial is very bad for sound on PC and should not be used there.
Click to expand...

The transformers I have seen on PCI cards always output to a cinch isolated from the PC ground. It provides galvanic isolation, otherwise it would be a useless item on the bill of material. However, on my Prodigy 192 it can bend and touch the ground plate if forced enough. To be used with care, as usual :)
 
P

phofman

Member
Joined 2005
momitko said:


When I was looking at the specs I didn't see 44.1 anywhere on the page. That's why I made an assumption that the card does not have 44.1. Honestly, I have no idea.
Click to expand...

It is basically a regular Envy24-based card extended with the proprietary TG Link (some kind of serialized multiple I2S, clever stuff). I would be surprised if it did not work with all the frequencies Envy24 supports, i.e. from 8-12kHz up to 192kHz.

The major problem is the outdated PCI interface - many new MB do not have a PCI slot, while there are very few PCI-express sound cards on the market. Plus I have not seen any with two crystals, to support 48kHz as well as 44.1kHz families without the jittery PLL.
 
M

momitko

Member
Joined 2008
The transformers I have seen on PCI cards always output to a cinch isolated from the PC ground. It provides galvanic isolation, otherwise it would be a useless item on the bill of material.
Click to expand...

As far as I understand galvanic isolation is achieved when input and output signals are isolated through a transformer but their grounds are not isolated via transformer. Am I not right here?
 
P

phofman

Member
Joined 2005
The "signal" requires two wires. In SPDIF it is the ground/shield and the live wire/coax core. The output winding of the isolation transformer has two ends - one goes to the shield, another goes to the core. If you hook the shield to the PC ground, no more galvanic isolation.
 
M

momitko

Member
Joined 2008
I’ve been wondering about the output winding.
Looking at the standard input connections of any DAC with SPDIF transformer in front of the digital receiver IC I see that in the input winding of the transformer the core is connected to one end of the transfomer and the shield is connected to another end of the transformer and this point is connected to the chassi of the DAC as well.
From this I made an assumption that the grounds of two devices are connected even in galvanic isolation and only Toslink provides complete galvanic isolation.
Is this not correct? I do not have an electric engineer degree. So I am just guessing…
 
P

phofman

Member
Joined 2005
You are right about the DAC side. However, the coax shield is NOT connected to PC chassis on the source side, only to the coax transformer output winding. Therefore the two chassis are not connected via the shield. They may be connected via the ground plug in the socket but that is a different story :)
 
Franz Gysi

Franz Gysi

Member
Joined 2007
Hi Danny

I did not any measurements, but comparisons (Studer D730, Shigaclone, Squeezebox).

My conclusion: the SPDIF implementation on the SndScape Card is very good.

Absolutely impressive is the TG-Link. It would be nice to have this modules on the market:

www.sndscape.cn/download/pdf/tg-link to i2s module v11.pdf

TG-Link to I2S. This would allow to add TG-Link Input to homebrew DAC's.

The disadvantages from this soundcard:

- manual only in chinese language. But enough pictures and the driver software is easy to understand.
- TG-Link is limited to 1.5m length

Regards

/Edit
It is worlds better than every USB solution I tested up to now (Digigram, Yellotec, SuperPro, DIY PCM2706).
Franz
 
P

phofman

Member
Joined 2005
TG Link, unlike SPDIF and USB, does not involve complicated clock recovery. It is some kind of multiple I2S via balanced lines. I still do not understand why it took so long for this kind of solution to appear. All the components used have been on market for decades.
 
danny_66

danny_66

Member
Joined 2005
So TG-link seems the way to go, hopefully we'll see more from this new technology in the future :)

About USB I can say it's great for connecting printers, cameras, keyboards... but for audio it's worthless.
Firewire is better, but not much support outside the pro-audio world.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom