XMOS-based Asynchronous USB to I2S interface

Yes many thanks are in order.
I’ve got mine hooked up to an ES9023 (my first DIY DAC) and its sounding really good.
Lucien’s board was a massive jump in SQ from the tenor based board that I was using to develop the DAC.
It’s taken the DAC to a whole new level.
A very nice board thanks Lucian.
 
Board Up and Running

Hooked my board to my Buffalo this weekend using external 5V supply. Using Mac and Pure Music, iTunes and Decibel. Everything came right up. This board is a keeper. Sounds really really good. Soundstage wider and deeper than my HiFace. Seems like I am hearing more detail than before. Excellent job Lucian
 
Hooked my board to my Buffalo this weekend using external 5V supply. Using Mac and Pure Music, iTunes and Decibel. Everything came right up. This board is a keeper. Sounds really really good. Soundstage wider and deeper than my HiFace. Seems like I am hearing more detail than before. Excellent job Lucian
The HiFace does not have analog outputs. Which S/PDIF-input DAC were you using with the HiFace that sounds inferior to the USB-to-I2S-to-Buffalo?
 
Flo...

If you are interested in trying something new with your Buffalo, I would suggest you try synchronous clocking: supplying the masterclock from the Wave IO. I love synchronous clocking with my B-II.

I suspect that Flocchini's comparisons were likely Hiface-SPDIF-Buffalo, unless he had an Evo which offers I2S out in addition to SPDIF.
 
If you are interested in trying something new with your Buffalo, I would suggest you try synchronous clocking: supplying the masterclock from the Wave IO. I love synchronous clocking with my B-II.

I suspect that Flocchini's comparisons were likely Hiface-SPDIF-Buffalo, unless he had an Evo which offers I2S out in addition to SPDIF.

Wavelength has been using two discrete clocks with the Sabre DAC in synchronous mode from the beginning when everyone else said it made no difference.
 
If you are interested in trying something new with your Buffalo, I would suggest you try synchronous clocking: supplying the masterclock from the Wave IO. I love synchronous clocking with my B-II.

I suspect that Flocchini's comparisons were likely Hiface-SPDIF-Buffalo, unless he had an Evo which offers I2S out in addition to SPDIF.

I was using a modified Hiface that does USB to I2S feeding the Buffalo II.
 
I am...

Wavelength has been using two discrete clocks with the Sabre DAC in synchronous mode from the beginning when everyone else said it made no difference.

Curious, who is "everyone"? Certainly there are quite a few audio designers who do not believe in using ASRCs: Gordon Rankin of Wavelength is one, others: Charlie Hansen (Ayre), Paul McGowan (PS Audio), etc. Usually, these designers just choose not to use the ESS chips and their built in ASRCs and instead choose a different DAC chip.
I have found the sound of the ESS 9018's onboard ASRC to be much better than that of the "usual suspects" (TI 4192, 4193 series, for instance), and the sound using synchronous clocking with the ESS is only just a little better...(this difference is small enough, thta I ma sure some might prefer using the ASRC).

Lorien: sorry for the off topic diversion-I only pointed out the option of synchronous clocking because the Wave IO has the ability to provide a low jitter masterclcok, and as such its users can take advantage of synchronous clocking if they so choose.
 
Russ White, for one, says there is no disadvantage to using a single crystal clock because of the Sabre's Time Domain Jitter Eliminator thingy. Dustin Forman of ESS says pretty much the same thing. Gordon disputes that and said Dustin actually agreed with him. An ASRC, no matter how good, always acts like a a filter and can not completely eliminate jitter. I guess it depends how fussy the user is about jitter.
 
Actually...

If one reads Russ White's most recent comments on the subject, he has noted that he really likes using the ESS 9018 with synchronous clocking, and he has noted that Dustin agrees.
It may be that in dialing in his own XMOS USB receiver, Russ has achieved low enough jitter to get synchronous mode sounding really good.
My experience is that the real issue here is not jitter per se:
Most DACs which use ASRCs do so because the ASRC does reduce measured jitter artifacts at the analog outputs (this is how Benchmark, and bel canto can get good measured jitter specs using adaptive USB).
The problem is, that the ASRC comletely resamples the data by complex mathematics, all the original samples are discarded and replaced with the new interpolations (unlike the integer based oversampling which occurs in the DAC chip itself, which preserves the original smaples). Additionally, the commonly used parts for ASRC are limited in processing power (unlike sophisticated SRC programs like Izotope, which use more complex math to achieve better results) and must operate in real time during playback. My opinion, is that the ASRCs used in many DACs, are simply not up to the challenge, and by their conversion, add significant (non musical) artifacts not present in the original data.
So, if one can provide a very low jitter data stream to the DAC, it is advantageous to use synchronous clocking, to avoid the problems caused by the ASRC, since we no longer need to "reduce" jitter at the DAC, as it is already very, very low.

To get back on topic: this is one reason why low jitter interfaces, like Lorien's, are so important: we can have low jitter at the DAC, without having to use "jitter reduction" techniques which cause other damage to the sound.
 
Lorien, a few questions.
Product looks like what I am looking for! The stock reference design from XMOS is cool but it is closed while I see your board as open and thus catering to my wishes of being able to tinker around a bit.

1) is the board populated & soldered? I have mostly seen empty boards in the start of this thread!

2) Can I connect to the I2S with a Philips-version of the I2S protocol, a TDA1541-43 based DAC or is it the Sony such as PCM58-63, or is that selectable?

3) I can't go higher than 18 bit in my DAC. That is no problem of course.

4) Can I improve the onboard clocks? (In the reference design XMOS L1 I could easily do that I think by piggy backing a doubled up TENT clock board.) I like to be able to do a later upgrade.

5) Does your XMOS implementation do selectable upsampling (&interpolation)?

I em'd you at ekits (at) luckit . biz, that is correct?
albert
 
Hello triode_al,
1) board will come fully populated and tested. Most of the parts are in SMT technology... some of them hard to solder without proper equipment.
2) WaveIO will source digital audio in 32-bit I2S format from which only 24 bits are used for audio data. From what I know, TDA will work but the data will be truncated.
4) I could leave the WaveIO card without oscillators mounted on it to be easier for you to test other master clock sources.
5) It's supposed to be bit-perfect so there's no processing techniques applied to the sound. WYSIWYG.
Yes, my email is as posted above.
Kind regards,
L
 
No intentions to half the price?
That's more in keeping with April 1st :rolleyes:
Not quite! Keeping the prices at affordable level for DIYers it's not new to me though 50% off does not even offer the chances to acquire all the parts needed to assembly the card... I could do that if I would choose ordinary parts and build this board in large quantities (10k or more)... so, from my perspective only XMOS have better chances to fulfill your demand :).
Kind regards,
L
 
Hello triode_al,

2) WaveIO will source digital audio in 32-bit I2S format from which only 24 bits are used for audio data. From what I know, TDA will work but the data will be truncated.

L

Unfortunately the LSB also gets mangled, so its not bitperfect for the older DAC's. Unfortunately I think this is a unirversal issue all asynch usb with the only exception being the proporitary Rankin firmware (not a plug more a cry for help as I hate the idea of hacking up a Legato to build an asynch USB-I2S TDA1541 DAC.)


Which leads me to unother question, those of you using the Xmos's clock to synchronously clock the Sabre DAC, do you have to share the xmos board's ground with the DAC's ground ?