TI PCM1792 DAC: change firmware in PC/soundcard

[hollowman]

May have missed this in the Archives ...
If one has a modern "high-end" PC sound card, or any computer sound section with a firmware-configurable DAC (like PCM 1792, etc.), how can one change this DACs parameters -- aka "User-Programmable Function Controls" -- (on-the-fly) in the computer? I haven't been able to find applicable drivers or plugins for my soundcard (Xonar) or software-audio-player (Foobar2000).
Ref: See Table 3: http://www.ti.com/lit/ds/symlink/pcm1792.pdf

 

[5th element]

With the Xonar cards the 1792 is controlled via the PCI bus interface chip over the I2C protocol. What the interface chip does to the 1792 is, as far as I am aware, determined by the ASUS software. ASUS already control the internal volume control on the PCM1792 and alter the oversampling rate with the sampling frequency. What is it exactly you would want to get access to?

 

[hollowman]

What is it exactly you would want to get access to?

The digital filter (fast/slow), sampling rate, etc.

I did find the 1792 User/Dev. Guide:
http://www.ti.com/litv/pdf/sleu037
(see pg. 1-5,1-6)
...and Demo/control software for Win.-based PCs:
Audio Converter - Audio DAC - PCM1792A - TI.com

I installed the Demo prog. but haven't figured out how to get it to DIRECTLY interact with the 1792.

 

[5th element]

You cannot access the 1792 on the ASUS card with the TI software, that is meant to work only with the TI demo board.

With regards to the sampling frequency, you can already change this with the ASUS control panel and the differences between the fast and slow roll off are largely accademic. Some hardware setups might benefit from the lower latency (slow) filters, but from a sound quality point of view, I doubt you'd able to hear anything. The slow roll off filter is still very steep in absolute terms, it's just less steep than the fast filter.

 

[hollowman]

Time domain vs. freq. domain perf.

You cannot access the 1792 on the ASUS card with the TI software, that is meant to work only with the TI demo board.

With regards to the sampling frequency, you can already change this with the ASUS control panel and the differences between the fast and slow roll off are largely accademic. Some hardware setups might benefit from the lower latency (slow) filters, but from a sound quality point of view, I doubt you'd able to hear anything. The slow roll off filter is still very steep in absolute terms, it's just less steep than the fast filter.

For some reason, the Xonar Control Center did not install on my PC. Now that I've done so, I do see the sample-rate control feature, but have no idea whether this is implemented in the 1792 or "in software only" via the main PC CPU ???
As far as fast/slow roll-off... it may not be JUST the freq-domain performance; i.e., sometimes the latter means better concomitant time-domain reconstruction. E.g., the "min. phase"or "apodizing" filters that have been getting a lot of attn lately. Wadia has used something similar since the late 1980's.

 

[5th element]

For some reason, the Xonar Control Center did not install on my PC. Now that I've done so, I do see the sample-rate control feature, but have no idea whether this is implemented in the 1792 or "in software only" via the main PC CPU ???
As far as fast/slow roll-off... it may not be JUST the freq-domain performance; i.e., sometimes the latter means better concomitant time-domain reconstruction. E.g., the "min. phase"or "apodizing" filters that have been getting a lot of attn lately. Wadia has used something similar since the late 1980's.

DACs are slave devices in that they don't have any say in the data rate, it is provided for them, not by them. When you select 48/96/192k in the ASUS control panel you are directly changing the sample rate being delivered to the DAC by the PCI audio interface chip.

Now some DACs do require that you alter the internal oversampling rate when the sampling frequency changes, the PCM1792 in default mode is not one of them however.

With regards to the fast/slow filters, yes the latency of the slow filter is less than that of the fast, so in theory, I think, would show less pre ringing. Without details of the ASUS drivers though, I don't know which ASUS use as standard.

 

[hollowman]

Upsampling?

DACs are slave devices in that they don't have any say in the data rate, it is provided for them, not by them. When you select 48/96/192k in the ASUS control panel you are directly changing the sample rate being delivered to the DAC by the PCI audio interface chip.

I should've been clearer....what I meant was what is delivered to the Xonar's decoder DSP. Does the Xonar control center simply act as a multi-position switch (it detects native sample rate of certain file or disc). So, is some software-based processing done betw. Asus and Windows/Intel? I doubt it, but audio players like Foobar do feature their own plugins that work in pure software only.
I think it's more likely as you suggest: hardware acceleration only (i.e., in the sound card). On that note ... can one effectively upsample non-high-rez data: e.g., send typical 16/44.1 RedBook data to 1792 in 96k or 192k? I haven't tested this yet -- maybe others have (subjective) and even taken some measurements (objective; graphs; RightMark)???
I have a MusicalFidelity outboard D/A that offers 96 or 192 as switchable upsamples ... via some dedicated asycn chip (curiously, one can't turn upsampling off completely). It also uses a TI DAC, I think.
Oh ... FWIW ... on the MF, 96k sound better to my ears.

 

[counter culture]

Why not just use a software sample rate converter?

 

[hollowman]

Why not just use a software sample rate converter?

If one has a dedicated "audio computer", this makes sense. Not the avg. Joe who is an obscene and vulgar multi-tasker ...

 

[dmills]

Why not, the upsampling does not have to be particularly low latency, and even a fairly long polyphase filter is not **that** expensive to compute.

The upsampler could run in the disk IO thread which can easily have a MB or so of buffer between it and the card.

I see no reason not to do this on the main CPU even of a very conventional multitasking box, a few percent of the CPU is a lot of mac operations.
Not that I think it will make much difference, my hearing is shot past about 15Khz or so, so 44.1K is **way** more then I really need.

Regards, Dan.

 

[counter culture]

You don't have to do it in real time. Just upsample file to file. Play the upsampled files. If you can hear any difference. This will give you a chance to find out. Rightmark?

Duh.

 

[5th element]

I should've been clearer....what I meant was what is delivered to the Xonar's decoder DSP. Does the Xonar control center simply act as a multi-position switch (it detects native sample rate of certain file or disc).

The switch in the control panel directly affects the rate that the hardware runs at. This overrides what any other software, including what's in the windows control panel, might be set to. The only exception to this is if you are using an ASIO driver, which bypasses everything in the control panel.

You do have to watch out with regards to sample rates. Depending on the settings you will either get the soundcard doing the upsampling, which it does a good job of, or windows upsampling, which it sucks at.

 

[hollowman]

Just took the Sample Rate all the way up from 44.1k (where its been all along AFAIK) to 192k. Definite improvement -- a keeper!
Yeah, that sucks that ASIO + Xonar Control is not doable (or is there a workaround?). For older Win this may be an issue, but AFAIK, Win 7 and later run much cleaner. Does this mean ASIO on newer Win OS not as "necessary"? You guys tell me.

EDIT: Researched this topic a bit. There was topical disc. thread on Hydrogenaudio not too long ago. The thread-starter noted 192k works best -- for him anyway -- with 16bit originals (he posted RMAA results to that effect). I haven't tested any of the in-between rates yet.

 

[5th element]

You're not supposed to be able to use the ASUS control panel with ASIO, that's the point, it bypasses everything so that nothing gets in the way. This means that no additional processing of the data stream occurs between the software producing the sound and soundcards interface chip. It gives a direct path, both in terms of software and any configurable processing within the soundcard itself, from start to finish.

 

[hollowman]

ASIO vs. Xonar

You're not supposed to be able to use the ASUS control panel with ASIO, that's the point, it bypasses everything so that nothing gets in the way. This means that no additional processing of the data stream occurs between the software producing the sound and soundcards interface chip. It gives a direct path, both in terms of software and any configurable processing within the soundcard itself, from start to finish.

So how "direct" is the Xonar Control Center software? It could be written in a way that it achieves ASIO-level "purity" and offer some hardware-level control (like Sample Rate). I've heard that ASIO "directness" is much less of a deal with newer versions of Windows (Win 7 and later; indeed, the ASIO home page has not been updated since Oct. 2008...so no new version since v2.9) -- apparently some folks in Redmond were concerned enough about audio in Windows to clean up the OS's audiostream path.
IAC, even on my now-dated XP machine, Xonar at 192k -- playing 16/44.1 files -- is an important improvement over ASIO.

 

[5th element]

So how "direct" is the Xonar Control Center software? It could be written in a way that it achieves ASIO-level "purity" and offer some hardware-level control (like Sample Rate). I've heard that ASIO "directness" is much less of a deal with newer versions of Windows (Win 7 and later; indeed, the ASIO home page has not been updated since Oct. 2008...so no new version since v2.9) -- apparently some folks in Redmond were concerned enough about audio in Windows to clean up the OS's audiostream path.
IAC, even on my now-dated XP machine, Xonar at 192k -- playing 16/44.1 files -- is an important improvement over ASIO.

From the tests and measurements that I've performed, they suggest that the Xonar's processing is minial, or close to none when you've got the 'direct audiophile' button selected.

Windows upsampling is horrendous though and I'm talking objectively about W7 here. The Xonar's is very good. If windows is handling any of your upsampling then you owe it to yourself to ensure that Windows isn't doing it.

ASIO directness wasn't specifically done to ensure bit perfect data transfer, it was originally designed and intended specifically to give very low latency playback. This is essiential with real time music synthesis, ie when you've got a MIDI keyboard plugged into the computer with the computer handling the sound. If you use windows to do this you press the key and the sound comes out what sounds like seconds later. Use ASIO and most of this goes away.

I personally use foobar with the SoX plug in set to 192k. I mostly use direct sound, making sure that windows is set to 192k as is the Xonar. If I want to do any critical listening though I'll switch over to ASIO. I can't really tell any difference but it makes me feel better that windows is truly out of the loop.

 

[hollowman]

From the tests and measurements that I've performed, they suggest that the Xonar's processing is minial, or close to none when you've got the 'direct audiophile' button selected.

'direct audiophile' button? I don't know what that is--at least for my ST version. Where is it?
EDIT: Okay, went thru the manual ... I think this is Hi-Fi "HF" DSP mode. It may be called something diff. in other versions???I had everything else "off" anyway, so pretty direct w/ or w/o this button, methinks.
Anyway ... About ASIO and latency. Audio via ASIO can be clicky and poppy. Latency can be a good thing in, e.g., a dedicated audio computer -- with proper software -- as it can buffer the datastream. But now with other inline buffers, like the one on the HD, not so sure ... also not sure how RAM memory w/ or w/o ASIO interact.
In the "old" days and on "older" (non-PC) gear, this was sometimes (and loosely) called re-clocking. For my Xonar ST, I'm using a 9-year-old XP system, and I way prefer Xonar at 192k over ASIO.
Not sure how the Xonar handles sample rate ... I assume the SRC is built into the AV100 DSP chip (as opposed to contracting out the job to an 'old'-fashioned / dedicated SRC chip).

I use Foobar, but I haven't experimented with its plug-ins extensively enough to offer opinion.

 

[hollowman]

Clock freq. for Firewire vs. Red Book

The Ayre QB-9 an outboard USB D/A uses two clocks:
22.5792MHZ for 44.1k/88.2k data, 24.576MHz for 48k/96k
So, the Xonar's sole 24.576 XO may not be ideal for most audio.
The 24.576 seems to also be optimized for Firewire.
See:
Crystal oscillator frequencies - Wikipedia, the free encyclopedia
That said, Xonar's manual notes its use of double floating point sample-rate conversion, so "integer division" or "binary division" issues from that 24.576 clock may not be that important.
But all that said, I just installed the latest n' greatest Uni Xonar 1.70 drivers and played around with the ASIO settings ... and that 192k sample rate in Xonar Center still stands out as the best software upgrade.
It does make some mathematical sense: setting the decoder to a STANDARD "integer division" freq. of the 24.576 clock (i.e., 192k or 96k), and feeding that "clean" math to 1792. But I'm not sure how much of this and/or concomitant up-/oversampling is related to my perceived sonic improvements.