Asynchronous Sample Rate Conversion - Page 2 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Source

Digital Source Digital Players and Recorders: CD , SACD , Tape, Memory Card, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 25th February 2004, 10:39 AM   #11
diyAudio Member
 
CheffDeGaar's Avatar
 
Join Date: Feb 2003
Location: Safe and sound
Hi Werewolf.
Nice thread you started here. Were(wolf) you a teacher ? I've almost all understood . Very "pedagogic" (sorry, dunno the english term...) Your posts are very clear, even for a non-english speaker like me.

Just a point for me please :

Quote:
Originally posted by werewolf
So, an interpolator is really an ideal low-pass filter. Quick example ... let's say I just want to "fill in" one sample between each input sample. There's really two conceptual steps in the process: first, "zero stuffing". We essentially consider the input signal to be coming twice as fast, with a zero inserted between each input sample. Then, the 2x rate signal feeds a low-pass filter ... output is an interpolated sequence
What would be here the cutoff frequency of the ideal low pass after interpolator ? 1/(2xrate) ?

Sure some diagrams would help... But darn good explanations - for the moment

Keep it up, and thanks for sharing your knowledge.
__________________
/Cheff - Falling feels like flying, until you hit the ground
  Reply With Quote
Old 25th February 2004, 12:17 PM   #12
diyAudio Member
 
annex666's Avatar
 
Join Date: Jul 2002
Location: England
I have to admit that I'm no expert and I'm probably missing something very fundamental, but can you not simply convert from 44.1KHz to 48KHz by oversampling the digital input at a multiple of fs_out and then use an FIR (or some other form of digital filter) to remove the imaging components created by the modulation of the original analogue signal and the 44.1KHz sampling rate.

Will this approach work or will it cause some other strange modulation between the 44.1KHz and 48KHz signals?
__________________
...if it ain't broke don't fix it - make it BETTER!
  Reply With Quote
Old 25th February 2004, 03:13 PM   #13
diyAudio Member
 
Join Date: Jan 2001
Location: Scandinavia
Werewolf,

Nice to see someone as knowledgeable as yourself on this forum. I have on quick related question:

When using an SRC (which I am pretty much forced to with my Behringer DCX2496), how do I maximize jitter performance given that input will almost always be 16 bit 44.1KHz and that outputs will always be 96KHz?

I am thinking specifically about putting a master clock in the Behringer (near the DAC, or near the SRC?) and clocking the source remotely from this master oscillator.

Would I get just as good performance if I had a stable clock in the source, and a stable separate clock in the target, or is there benefit from going this route when using an SRC. I am trying to understand what the critical performance improving factors are.

Any thoughts on this?

Petter
  Reply With Quote
Old 25th February 2004, 03:36 PM   #14
diyAudio Member
 
Ouroboros's Avatar
 
Join Date: Dec 2003
Location: Nottingham UK
I guess you could use a CS8420 to receive the SPDIF and use it to rate convert before feeding the Behringer.

There's a CS8421 coming out soon, which is the async rate converter without the driver/receiver bits.
  Reply With Quote
Old 25th February 2004, 09:33 PM   #15
diyAudio Member
 
Join Date: Jan 2001
Location: Scandinavia
The Behringer has integrated async samplerate converter that outputs 24/96 into the DSP core.

Petter
  Reply With Quote
Old 25th February 2004, 10:03 PM   #16
diyAudio Member
 
Join Date: Feb 2004
Location: Austin, TX
Thanks guys I'll post a few answers before we progress with the tutorial, cool?

Cheff : The ideal interpolator is an ideal LPF, with a cutoff at half the input sample rate. This is far from obvious in the time domain ... it is STILL a topic of endless debate & confusion ... but is very obvious in the frequency domain. And of course the simple result pertains EQUALLY to both domains.

Annex : Can't really do what you're suggesting either. In the Asynch world, no amount of finite integer interpolation of the Fs_out "tciks" will give perfect alignment with Fs_in ticks, any more than the other way around. Basically, simple integer interpolation of EITHER rate will never give perfect time alignment with the OTHER rate ... because they are asynchronous.

Now I know it sounds like I've painted myself in a corner ... but please be patient for another post or two

Petter : sounds like you want to jump right to the conclusion! I'm not planning to hit the JITTER implications for a little while yet !
But, since you asked ...

I've always hated the S/PDIF format, for the simple reason that it's technically backwards ... just so that a single cable can be used between a transport (source) and a DAC. We all know the BEST way to manage the system timebase is to put the cleanest (highest Q, lowest jitter) clock source known to man, right next to the component that CARES about how clean the clock is ... the DAC. And then, use whatever means necessary (simple cable, FIF0's, etc.) to SLAVE the transport to the DAC ... instead of the other way around. So, for minimal jitter, here's the best ways to go (increasing order of better performance):

1. Use a PLL-based clock, recovered from the S/PDIF stream, to clock the DAC.
2. Use an Asynch SRC to receive the S/PDIF stream. This allows a local (to the DAC) clock to time the DAC, and of course provide data to the DAC on the local timebase. I think this is the best option available that's still compatible with S/PDIF. It will be the ultimate intent of this thread to compare and demonstrate the superiority of this technique to option #1.
3. Slave the transport to the DAC ... probably requires, horror of horrors, more than one cable between the source (or transport) and the DAC (or processor). I know at least a couple versions of this technique are discussed frequently on this board

Now if the Behringer unit in question has a Crystal CS8420, it is most probably already using option #2.

But I promise, we will discuss the jitter impacts of ASRC, after we develop two things : the conceptual plan, and the more detailed implementation architecture
  Reply With Quote
Old 25th February 2004, 10:14 PM   #17
diyAudio Member
 
Da5id4Vz's Avatar
 
Join Date: Feb 2003
Location: lost in america...
Send a message via AIM to Da5id4Vz
Quote:
just so that a single cable can be used between a transport (source) and a DAC.
Does this mean that you prefer SDIF-2?

**I reread your post it sounds like youd prefer SDIF-2 with reverse WC back to the transport?**
  Reply With Quote
Old 25th February 2004, 10:18 PM   #18
diyAudio Member
 
Join Date: Feb 2004
Location: Austin, TX
gotta claim ignorance Dr. Strangelove ... what's SDIF-2? Is there a new (well, new to an old guy) standard for sending the word clock back to the transport?

I know my beloved Wadia does this, or something similar ... ?
  Reply With Quote
Old 25th February 2004, 10:48 PM   #19
dddac is offline dddac  Germany
diyAudio Member
 
Join Date: Jul 2003
Location: Wiesbaden
Hi W-Wolf...

nice reading stuff, nice theory. But imo what counts is how it sounds. I have built few versions with ASRC the last 4 years and these are my 2 key findings:

1. If you do asynch resampling and then apply a digital filter (like 8420 followed by SM5842 or DF1704 for example) you can do almost any rate you want, totally asynchroneous, I have used 46.875kHz, 70,31kHz and 93.75kHz The clocks driving this was also clocking the DAC. All with relatively the same result: more air round instruments, better soundstage pin point. I liked best the last one, but they were all close....

2. I tried this as well with the Non Oversampling dddac1543 and it did not work well !!! Indeed you get the extra air etc, but tremendous amount of IM distortion, brrrrr with female voice you could actually HEAR the extra tones . Then I moved to the Tent XO Clock with 11.2896 MhZ, so in fact this is stays a sample rate conversion, but very very close to the original rate. And yes, I had again exactly the same benefits as described in #1.

The sound improvement is very easy to hear also in blind test (you need only a A-B switch, so indeed very easy, no cable plugging etc)

The practice is already here, so interested to read your theorie now. Still wondering what conclusions you will be drawing and if there is a hand-on part as well. I gues you have been doing some workshopping as well and not only been number crunching ??

doede

ps: as one example (I have some more test pictures) I attach a sinus of 1kHz, ARS-ed with a 47kHz clock in a non oversampling DAC.... clearly this will not sound pretty...
Attached Images
File Type: jpg sinus 0db 1khz fft 12mhz clock.jpg (22.7 KB, 3059 views)
  Reply With Quote
Old 25th February 2004, 10:54 PM   #20
dddac is offline dddac  Germany
diyAudio Member
 
Join Date: Jul 2003
Location: Wiesbaden
ok, another one, could not stop my self

this is the spectrum of a 11kHz sine wave from a nono dac, a-synchroneous resampled with ~ 47kHz........

pretty nasty eh?

when resampled with a close rate, no issue...

tc

doede
Attached Images
File Type: jpg fft sinus 11khz 12mhz clock.jpg (20.0 KB, 2943 views)
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
PMD100 Max Sample Rate? Filburt Digital Source 7 23rd May 2008 05:30 AM
Sample rate into BB PCM1798? Asgard Digital Source 14 22nd July 2006 01:54 PM
DSO Bandwidth vs sample rate bzo Parts 3 23rd February 2004 04:53 PM
New sample rate converters from TI hifiZen Digital Source 3 16th March 2003 12:42 AM


New To Site? Need Help?

All times are GMT. The time now is 08:18 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2