PC - The Perfect Source ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
BlackCatSound said:
You may want to read the PCM2707 datasheet.



Its slaved to the PC.

http://focus.ti.com/lit/ds/symlink/pcm2707.pdf

Page 8 has the block diagram.

Perhaps we should get the defintions straight.

It says. "It recovers the clock from USB data". The chip itself is clocked by
an on-chip PLL.

On a Master/Slave setup, which runs from my understanding synchrounous, Fifos on both ends are synched by a master clock. Jitter is more or less not existent. This is a complete different thing with the 2707.
 
The PCM2707 is clocked by its onboard PLL, the PLL is synced to the PC via the USB.

Any time you do this you will get jitter as the USB is asynchronous and will be jittery.

With PLLs jitter in = jitter out.

In the case of PCI, the I2S side of the FIFO and the DAC would be clocked by a precision clock which is totally unconnected with the PCI side of things. The FIFO provides a buffer between the two clock domains.

No PLL means the jitter performance is determined only by the precision clock.

The PC just keeps the buffer full, and as long as the buffer doesn't run empty the timing of the PCI side of things doesn't matter at all. The jitter on the PCI side could be huge and it won't matter.

You could actually do this with a USB interface but all the USB interfaces available have PLLs in them as they are all designed to be slaves.
 
Ok, but is there any way to extract the I2S signal from a pci audio card or even directly from the pci slot?

As far as I know the next best option would be to use the AES/EBU signal from a pc card like the lynx aes16. Any other cards that offer this output format that are cheaper than the lynx?
 
i2s

most of the via envy chipsets output i2s and are used in a lot of lower cost, but decent cards. I got a no brand card which is essentially the same as a chaintech av710, cant remember which chip it uses specifically, but i2s could be tapped off the controller chip before it goes to the dac. There are a couple of circuits on this forum for interfacing i2s externally which could be added on a bracket above the soundcard. this same procedure has been posted here several times. No drivers would have to written for this, just a bit of careful soldering.

Yes soundcheck, in that thread on diyhifi John swenson had made measurements which he suggested showed the usb drifting with the PCs motherboard clock, which I think was the spact mechanism. Do you think offline upsampling to 48khz is combating this somehow? I dont doubt it is a good solution - heard it from too many people too many times, including John, for there not to be something in it. I have some ti samples and will try it when I have time.

The main points against using an internal dac are the EMI and the psu quality. Could the card not simply be shielded and have its own low voltage supply? Havnt seen any detailed diy solutions like this, but there was a linux audio page which suggested surrounding the card with foil lined card which is grounded to the case. EMI from the cpu is apparently directly proportional to the amount of heat generated by the cpu. The EMU cards have very decent dacs internally.

Would be really cool to have a linux distro specifically for playback, but I think perhaps dynebolic has everything needed set up out the box. Will have a look at this Aplay, but one of the reasons for my interest in PC audio is for an improved interface to a large music collection, preferably remote control only - would prefer to swap cds than use the command interface or load in single tracks at a time. I use audacity on windows, great features, but doesnt seem to support asio. For free asio recording I use Krystal, might be worthwhile auditioning this.
 
Re: i2s

rossco_50 said:
The main points against using an internal dac are the EMI and the psu quality. Could the card not simply be shielded and have its own low voltage supply?

The PSU side of things is a total doddle to solve.

The vast majority of internal sound cards just don't bother doing anything with the PSU as it cuts into their profit.

Even something as simple as a BLM or inductor in the PSU rail will help a LOT. Better still sub-regulate as well.
 
Re: i2s

rossco_50 said:
most of the via envy chipsets output i2s and are used in a lot of lower cost, but decent cards. I got a no brand card which is essentially the same as a chaintech av710, cant remember which chip it uses specifically, but i2s could be tapped off the controller chip before it goes to the dac. There are a couple of circuits on this forum for interfacing i2s externally which could be added on a bracket above the soundcard. this same procedure has been posted here several times. No drivers would have to written for this, just a bit of careful soldering.

Yes soundcheck, in that thread on diyhifi John swenson had made measurements which he suggested showed the usb drifting with the PCs motherboard clock, which I think was the spact mechanism. Do you think offline upsampling to 48khz is combating this somehow? I dont doubt it is a good solution - heard it from too many people too many times, including John, for there not to be something in it. I have some ti samples and will try it when I have time.

The main points against using an internal dac are the EMI and the psu quality. Could the card not simply be shielded and have its own low voltage supply? Havnt seen any detailed diy solutions like this, but there was a linux audio page which suggested surrounding the card with foil lined card which is grounded to the case. EMI from the cpu is apparently directly proportional to the amount of heat generated by the cpu. The EMU cards have very decent dacs internally.

Would be really cool to have a linux distro specifically for playback, but I think perhaps dynebolic has everything needed set up out the box. Will have a look at this Aplay, but one of the reasons for my interest in PC audio is for an improved interface to a large music collection, preferably remote control only - would prefer to swap cds than use the command interface or load in single tracks at a time. I use audacity on windows, great features, but doesnt seem to support asio. For free asio recording I use Krystal, might be worthwhile auditioning this.


Ross.

I think integrating a shielded, selfpowered, precision clocked DAC into the PC, would be a nice DIY project. On top you'd connect a T-AMp right to the DAC output. The DAC and T-AMP might be powered by a nice battery.
You'll have a complete perfect sounding system in one enclosure!
My system looks pretty similar today. I am just lacking the PCI-I2S- card.

One more LInux hint: Configure yourself tmpfs, a kind of virtual ramdisk by
entering below line in /etc/fstab

tmpfs /tmp tmpfs defaults,size=1028m 0 0

The value (1028) shows the ramdisk size in MegaBytes. Depends on your available RAM of course. I got 2 Gigs.
The great thing. The allocated size is dynamic. This is IMO much better than a static ramdisk, which allocates the RAM on a permanent basis.
Copy your tracks over there. And play it from tmpfs. Let me know how it sounds! ;)

Finally. I regard 48khz samplfreq superior over 44,1 and offline upsampling with R5Brain pro superior over any realtime algorithm I tried ( in PCM2707/USB context). (Drop me a mail if you're interested to learn more about it.)



Klaus
 
I regard 48khz samplfreq superior over 44,1 and offline upsampling with R5Brain pro superior over any realtime algorithm I tried ( in PCM2707/USB context).

Well, what do you think about offline upsampling and use of a ASRC?
I intend to use a 24/192 dac (ad1955 or wm8740) and I was wondering what kind of upsampling would give the best results (assuming I'l feed the dac with a I2S input extracted from an audiocard).
There are, I think 3 options: ASRC (but this maybe would complicate things as Ross and blackcatsound showed me), offline upsampling with a software like R5Brain (thanks soundcheck) or the foobar dsp plugin, and finally upsampling with the settings from a M-audio card or similar (although I am not very sure about this. Where is the upsampling beeing done: software in the drivers or in the envy codec?).
 
I found the following sound card today. maybe Audiotrak is reading this thread. To summarize:

- power supply filtering, electromagnetic shielding, and OpAmps on sockets for swaping.

- I can see that this card can be modded by replacing the 2200uf capacitors and some other stuffs with jung SuperReg, and swaping in OPA627.


To provide stable high level output signals, Prodigy 7.1 HiFi uses a fine selection of high quality capacitors from Sanyo (SP series) and from Nichicon (up to 2200uF RU series / Nichicon ES series) to optimise the power supply of the analog and digital components on the card and to optimise the audio output quality. The special design of Prodigy 7.1 HiFi with its 4 layer PCB cuts off electromagnetic waves from the internals of the PC – there won’t be problems with an unstable power supply caused by the systems CPU or the HDD.

A separation between the analog and the digital part using an aluminium cover that shields the sound against electromagnetic waves is a major part of the Prodigy 7.1 HiFi design. Even better, the OpAmps on the card are installed on sockets, providing the possibility to change them to the users preference.

http://www.audiotrak.co.uk/home_theatre/prodigy71hifi/prodigy71hifi.shtml
 
Well I found something like that to...

Auzentech X-Meridian

The shield of the prodigy is a nice touch though. Auzentech also has swapable op-amps. The pro's (compared to prodigy) would be the use of a new audio controller, the CMI-8788 and maybe better DAC's than the wolfonson, 4 pcs 24-bit/192kHz AK4396VF (120dB-part spec.) DACs for 7.1channel output. (24-bit/192kHz in 7.1channel playback)

I also like the simplicity of the layout.. It eventualy alows me to easily extract I2S signal, but I guess that maybe an external dac is not that necessarely..

A nice review is here, I hope others will apear on the net...

review
 
You could look to replace the crystals on the soundcard, but think you will find it difficult to find the right frequency in low jitter versions (I tried for c-media, but vendors are not interested in low volume) - most often there are two crystals to cover the different formats supported by the card. You might try Kwak clocks based on the existing crystals, but the clock generators on the cheap chipsets, c-media etc, might be the quality bottleneck here.

I hope in the future to set up a master slave situation with the PC through optical spdif. The jitter performance can be improved by synchronous reclocking, and optical isolates the dac from the pc electrically.

Put the master clock next to the dac, divide it down to whatever you want the output of your soundcard to be synced to, 44.1khz etc. send the divided clock as a spdif signal to the spdif input of the soundcard, the soundcard will sync to the incomming master clock and output via spdif to the dac. At the dac, reclock the soundcard ouptut with the masterclock. I've read it doesnt really matter (within reason) what jitter happens in the spdif to/from the dac as the same clock is used at either side of the interface. Therefore there would not be much advantage in using i2s in this instance. Please correct me if Im wrong.

as I mentioned before, the m-audio audiophile 2496 seems to sync to spdif input and is probably the cheapest (most linux friendly) card to provide this feature without hacking. I believe the linux alsa drivers will allow you to do this through the envy24 control panel.

I have a cheap card with an envy chip - it has spdif in and out. It certainly doesnt natively support sync to master clock, and Im unsure whether this is a hardware feature or software feature. However, given that on linux all envy cards use the same control panel it may be possible to use the linux drivers to force the card to do this. If I cant do this in software, I could remove the crystals and then tap the incomming spdif from the reciever chip and feed the external clock signal to the chip in place of the crystals. As would be done when applying this diy methiod to a cd player.

You could always do this with a Cmedia card which has spdif input. I have seen CMI8*** (cant remeber exact model) based cards spdif in/out for around 10 euros. Alternatively save up for an RME or Lynx and be done with it.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.