• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

exaU2I - Multi-Channel Asynchronous USB to I2S Interface

Hello, what format does this device use for 16bit / 44.1 khz ?

is it the Philips I2S format or the Sony/EIAJ format available on it's outputs?

I mean if I would like to use it in NOS (non over sampling mode) together with a philips TDA1541A DAC which accepts only Phlilips format I2S stream would this be possible? or does it make internally some kind of OS (oversampling) already to the signal?

Thanks

Hi, the device has an I2S output format and is capable of 44.1khz up to 384Khz, 16 to 32 bits. So yes, it also supports NOS DACs for 44.1Khz files.

What is EIAJ? Isn't that an analogic audio-video standard?
 
Hello, what format does this device use for 16bit / 44.1 khz ?

is it the Philips I2S format or the Sony/EIAJ format available on it's outputs?

I mean if I would like to use it in NOS (non over sampling mode) together with a philips TDA1541A DAC which accepts only Phlilips format I2S stream would this be possible? or does it make internally some kind of OS (oversampling) already to the signal?

Thanks

exaU2I uses the I2S format. There is no oversampling or any other signal processing. The driver and the device are designed to be as transparent as possible. You can check the Terminal functions section of the DIY guide to verify compatibility with your DAC - www.exadevices.com > exaU2I > D.I.Y. Guide.
 
thank you both for your answers...perhaps i wasn't that specific about my question;
as you all know there are 2 standards of the I2S stream that reaches a specific DAC depending of the various decoders used in cd players;
there are many postings on the internet forums about these 2 standars and conversions from one to the other and vice versa;
I'll give you a link where you should read the 2-nd posting, after this maybe someone who has the device may confirm which standard is on it's outputs

http://www.diyaudio.com/forums/digital-line-level/58564-eiaj-i2s-converter-vice-versa.html


PS.
as an example I give you the TDA1543 DAC which comes in 2 versions (and in some well designed applications is very impressive natural sounding in NOS mode with NO long term listening fatigue)

TDA1543 which is only working with Philips I2S standard
TDA1543A which accepts ONLY japanese input format (EIAJ), as opposed to I2S of plain TDA1543
 
Last edited:
I've been using the exaU2I board for about a week now. It's connected to a Twisted Pear Opus WM8741 board powered by a diy shunt regulated PSU.

I've tried this dac with both a standard PCM270x receiver and the async TAS1020 receiver. The exaU2I is far superior to both of these. With the Twisted Pear receiver I had a lot of problems with noise. This is not the case with the exaU2I. It's completely noise free at full output with my 105 dB/w/m sensitive speakers. The sound is really detailed, dynamic and natural. It is really the best transport I've heard. The jittery PCM270x made my ears bleed, but the exaU2I have none of the traits I associate with a jittery device. The asynchronous TAS1020 is also pretty good, but the exaU2I reveals more details and comes across a bit more relaxed. I've also experimented with 24/192 upsampled material with great results. This allows the WM8741 to be run in low oversampling mode, which sounds even better.

Highly recommended!

The plan is to get two more Opus boards and use the Foobar crossover plugin for a three-way active speaker system.
 
Hi Painkiller, nice to see others are hooking their dacs to the exaU2I. Thanks for the review also, I had basically the same experience comparing the exaU2I to the PCM270x.

Could you please test your WM8741 DAC with a 352.8Khz sampling rate file? I am very curious if your DAC would play that..
 
The WM8741 plays sampling rates up to and including 192 kHz. I tried 352.8 and 384 kHz but all I got was silence.

I guess you'd have to run the WM8741 in 8FS mode to be able to play higher sampling frequencies, but then it's not as straigtforward anymore. I'm intrigued to read about the AK4396 accepting higher sampling frequencies. The AK4396 is a really good chip. Even better sounding than the WM8741.
 
And this is a serious problem as far as your understanding of high-end digital audio goes. Textbook solutions which work just fine in other types of electronics are not necessarily suitable for serious audio.
There is nothing lacking in my understanding of high-end audio. If you care to give an example where digital isolation improves a situation without analog isolation, or even where digital isolation improves a situation with analog isolation, then I would be very interested. In the former case, I would say that the mistake is to not have analog isolation in the first place, such that the digital isolation is inferior to analog isolation. In the latter case, I would say that the analog isolation is insufficient if digital isolation can improve upon it.

But before we go any further, I would like to ask whether you have any electronics design experience at all, or if you are just reciting articles from popular audiophile literature.
 
I wonder if I can get some opinions about a nutty idea I just had.

Is there any merit in feeding my stereo DAC that has eight DA chips with eight I2S streams via the exaU2I?

I have the Audio-gd REF7 ºÍ§Ó*µ响. I really like the sound signature of this eight piece PCM1704 R2R mono chip but it does seem to be very sensitive to jitter.

The REF7 is fed by SPDIF and has a jitter reducing DSP module ([FONT=&#26032]http://www.audio-gd.com/Pro/dac/DSP1/DSP1ENspecs.htm Two-channel Digital Interpolation Filter and data in-phase processor for digital audio),[/FONT]that takes I2S, upsamples if you like and splits I2S into left and right channels for the mono PCM1704's (four on each side).

I guess Foobar can use VST (Virtual Studio Technology) and DSP's to create four left only channels and four right only channels.

If these were taken out of the exaU2I by identical wires to feed each chip discreetly would it work?

If it makes music I wonder if you might care to speculate which will sound better.
A)a single I2S stereo signal to the DSP-1
B) Eight discreet I2S mono signals - one per PCM1704

Cheers
 
Last edited:
There is nothing lacking in my understanding of high-end audio. If you care to give an example where digital isolation improves a situation without analog isolation, or even where digital isolation improves a situation with analog isolation, then I would be very interested. In the former case, I would say that the mistake is to not have analog isolation in the first place, such that the digital isolation is inferior to analog isolation. In the latter case, I would say that the analog isolation is insufficient if digital isolation can improve upon it.


I am not sure if I understand you completely, but I recall the designer of the HRT Music streamer stating that he thought ground isolation of the incoming digital stream was of the utmost importance, it is rather easier to clean and isolate the power supply grounds from the computer. Mind you his device was USB bus powered so maybe he had a vested interest in saying that.
 
I wonder if I can get some opinions about a nutty idea I just had.

Is there any merit in feeding my stereo DAC that has eight DA chips with eight I2S streams via the exaU2I?

I have the Audio-gd REF7 ºÍ§Ó*µ响. I really like the sound signature of this eight piece PCM1704 R2R mono chip but it does seem to be very sensitive to jitter.

The REF7 is fed by SPDIF and has a jitter reducing DSP module ([FONT=&#26032]http://www.audio-gd.com/Pro/dac/DSP1/DSP1ENspecs.htm Two-channel Digital Interpolation Filter and data in-phase processor for digital audio),[/FONT]that takes I2S, upsamples if you like and splits I2S into left and right channels for the mono PCM1704's (four on each side).

I guess Foobar can use VST (Virtual Studio Technology) and DSP's to create four left only channels and four right only channels.

If these were taken out of the exaU2I by identical wires to feed each chip discreetly would it work?

If it makes music I wonder if you might care to speculate which will sound better.
A)a single I2S stereo signal to the DSP-1
B) Eight discreet I2S mono signals - one per PCM1704

Cheers

Maybe exa065 should have answered this - but I take my chances and I hope exa065 corrects me if I am inaccurate or wrong...

Your question started me thinking and as I have experimented a bit with the exaU2I I believe both PCM1704 and other DAC´s could work as mono DAC´s with a correct configuration of the exaU2I ASIO driver :D

The PCM1704 could be driven directly by the exaU2I as you can set up the ASIO driver to output Left channel for both channels of I2S channel 1, then Right channel for both channels of I2S channel 2.
In this case you could have full 384k sample speed support as it uses the 4 channel mode of the exaU2I.

If you map I2S channel 3 and 4 like channel 1 and 2 you would double the number of outputs and could drive additional DAC chips directly, but now the 8 channel mode of the exaU2I are used and the maximum sample rate are lowered to 192k.

In the case of the PCM1704 it should handle samplerates up to 768k, but I do not know if this 768k in as in mono or stereo.
When using the exaU2I in 4 channel mode with support for 384k the PCM1704 will be feed with up to either the full speed or the half speed of the supported sample rate speed.
This will also act as a "kind" of upsampling as the PCM1704 now will receive the double rate of the data as two Left or Right samples (identical) will be sent to the DAC, but without any damaging upsampler with dither or any digital filters with damaging pre and post ringings etc..

44.1k -> "upsampled" to 88.2k
48k -> "upsampled" to 96k
88.2k -> "upsampled" to 176.2k
96k -> "upsampled" to 192k
176.4k -> "upsampled" to 352.8k
192k -> "upsampled" to 384k
352.8k -> "upsampled" to 705.6k
384k -> "upsampled" to 768k

The PCM1704 have a pin to select inversion of the data so balanced outputs are easily implemented.

In the case of all other current output DAC´s this ASIO setup configuration could also be used and effectively create mono DAC´s out of a stereo / 2 channel DAC with the "upsampler" effect without the damaging pre and/or post ringing digital filter...
It is just to sum the current outputs of the two channels of the stereo / 2 channel DAC´s before the I/V stage...
This will in addition give a ca. 3dB improvement in noise / THD as a bonus..

Also voltage output DAC´s can use this setup, but the left and right channel on each DAC must then be summed to get the benefits.
The implementation of the outputs of the actual DAC´s may require different approaches to make it work...

As @kazap very good question gave me the "solution" to implement the PCM1704 (and other DAC chips) without any upsamplers, DSP´s or digital filters and with a effective 2x upsampling without the damaging effects of upsamplers, DSP´s or digital filters and then use my remotely controlled variable gain JFET I/V stage as volume control I have a near perfect setup I apparently must make as a prototype for evaluation..
 
Last edited:
In the case of all other current output DAC´s this ASIO setup configuration could also be used and effectively create mono DAC´s out of a stereo / 2 channel DAC with the "upsampler" effect without the damaging pre and/or post ringing digital filter...
It is just to sum the current outputs of the two channels of the stereo / 2 channel DAC´s before the I/V stage...
This will in addition give a ca. 3dB improvement in noise / THD as a bonus..

I have for some time had the upcoming Arda AT1401 DAC in mind for a possible solution...
But the AT1401 will not be able to be used with the above mentioned benefits as it extracts only half the samples out of the data stream.
This feature makes it easier to implement as it does not require a DSP or digital filter with dedicated left and right channel outputs, but as said it eliminates the possible benefits....
 
The optimal setup with the exaU2I and the PCM1704 may be as follows:

1. Galvanic isolation of the USB, and a local 5 volt power supply for the exaU2I.
2. Connect the PCM1704 chip inputs directly to the exaU2I by bypassing the GMR I2S galvanic isolation and with its own + and - 5 volt power supply.
3. Use the automatic LSB extension setting of the ASIO driver and the PCM1704 will get "correct" 24 bit data (with 16 bit sources) and the Bipolar offset servo of the PCM1704 will deal with the LSB offsets below 23 bit...
This is mostly important when using the PCM1704 in the inverting mode (balanced setup) as 16 bit data then will create a 1/2 LSB error (at 16 bit level) that the PCM1704 cannot correct due to the Bipolar offset servo works below 23 bit resolution.

With this setup the only clocks in use are the low jitter clocks of the exaU2I..
My math (calculator) indicates a possible theoretical improvement of the jitter in
 

Attachments

  • dig_PCM1704.gif
    dig_PCM1704.gif
    6 KB · Views: 429
Last edited:
The optimal setup with the exaU2I and the PCM1704 may be as follows:

1. Galvanic isolation of the USB, and a local 5 volt power supply for the exaU2I.
2. Connect the PCM1704 chip inputs directly to the exaU2I by bypassing the GMR I2S galvanic isolation and with its own + and - 5 volt power supply.
3. Use the automatic LSB extension setting of the ASIO driver and the PCM1704 will get "correct" 24 bit data (with 16 bit sources) and the Bipolar offset servo of the PCM1704 will deal with the LSB offsets below 23 bit...
This is mostly important when using the PCM1704 in the inverting mode (balanced setup) as 16 bit data then will create a 1/2 LSB error (at 16 bit level) that the PCM1704 cannot correct due to the Bipolar offset servo works below 23 bit resolution.

With this setup the only clocks in use are the low jitter clocks of the exaU2I..
My math (calculator) indicates a possible theoretical improvement of the jitter in

Thanks for the intriguing design ideas.
Its even more complicated then I had imagined.
I wonder what your math indicated, in terms of a jitter figure, with your design?
 
Thanks for the intriguing design ideas.
Its even more complicated then I had imagined.
I wonder what your math indicated, in terms of a jitter figure, with your design?

It appeare that a whole section was removed by accident when I edited the post...

A calculated improvement in jitter of up to 60 dB...
It may not be measurable by more than a few dB´s..
But I expect it would affect the fidelity...