The sound board driver sets latency for the system. Opus will run w/ 128 samples latency. BII will lock at 256 but not solidly and crackles all the time. At 512 samples BII crackles on complex passages. At 1024, only the rare quiet tick on big sounds that are tough to record (such as opera choruses, the finale of Beethoven's 5th, big live concerts, etc.), whereas jazz and simpler acoustic programming are never a problem.
Frank
Last edited:
I could reproduce jkeny's finding on my I2S connection from SDTrans192 to Buffalo II even for 352.8 kHz/24 bit I2S signal. I don't think my devices are so lucky guys. SDTrans192 and Buffalo II DAC are running on separate battery power supplies and IVY III is connected another +-12 V power supply. Therefore, there is no common ground link between SDTrans192 and Buffalo II.
Dear Frank,
I think it's better to show more specific conditions of your system.
Have you observed wave shapes? Which voltage of I2S signals, +3.3V or +5V?
On what sampling rate do you suffer the noises?
Does your sound board output 16bit data or 24bit data?
Do you use a standard MCU on Buffalo II without any modification?
Do you set the input select switch on Buffalo II on I2S side?
What kind of sound board driver software do you use, ASIO ?
What kind of OS do you use on your computer?
Best regards,
Bunpei
Assuming that latency means adding some delay before the bits are delivered to the DAC, that is some strange behavior...
I found the 'handle'...
Thanks to Bunpei and others, I was spurred to a better understanding of the odd DAC behavior. Bottom line: the problem I supposed to be quality of I2S was really something else. Now I am not so concerned about the details of how to lift I2S from the sound board. Thanks, guys, for your patience and willingness to help! Read on if you want the story...
Before hooking up the scope to check the signal, I thought it best to decide on specific conditions that were either OK, or were problematic. The sound board and crossover software (Thuneau) run via ASIO. The player I use outputs via Windows Audio Session (WAS) in Win7 OS. I don't know much about how WAS and ASIO interact in software, but a) there is a separate latency for WAS, b) the units are milliseconds, and c) I had it set low @ 1 ms because I thought my new CPU and memory was easily fast enough. I discovered that increasing the WAS latency would then allow me to lower the ASIO latency. BII ran just fine with an ASIO latency of 128 samples if the WAS latency was way long - like 100ms. At 128 samples latency there was never a problem with HD sources, however, sound would still break up on some redbook recordings (ALAC format) that are routinely upsampled to 96/24. My best sounding compromise at the moment is 256 samples for ASIO and 4 ms for WAS. SQ "presence" is very noticeably increased - an effect I appreciate. Best of all, no heroic measures will be required to pull I2S off of the sound board when I re-box everything in the new chassis. ...back on track!
BTW, for the sake of curiosity, I removed the ground input from the BII and... Voilla! Sound!
Also BTW, lowering the ASIO latency helps the Opus boards even more than the BII.
Thanks and best regards,
Frank
Thanks to Bunpei and others, I was spurred to a better understanding of the odd DAC behavior. Bottom line: the problem I supposed to be quality of I2S was really something else. Now I am not so concerned about the details of how to lift I2S from the sound board. Thanks, guys, for your patience and willingness to help! Read on if you want the story...
Before hooking up the scope to check the signal, I thought it best to decide on specific conditions that were either OK, or were problematic. The sound board and crossover software (Thuneau) run via ASIO. The player I use outputs via Windows Audio Session (WAS) in Win7 OS. I don't know much about how WAS and ASIO interact in software, but a) there is a separate latency for WAS, b) the units are milliseconds, and c) I had it set low @ 1 ms because I thought my new CPU and memory was easily fast enough. I discovered that increasing the WAS latency would then allow me to lower the ASIO latency. BII ran just fine with an ASIO latency of 128 samples if the WAS latency was way long - like 100ms. At 128 samples latency there was never a problem with HD sources, however, sound would still break up on some redbook recordings (ALAC format) that are routinely upsampled to 96/24. My best sounding compromise at the moment is 256 samples for ASIO and 4 ms for WAS. SQ "presence" is very noticeably increased - an effect I appreciate. Best of all, no heroic measures will be required to pull I2S off of the sound board when I re-box everything in the new chassis. ...back on track!
BTW, for the sake of curiosity, I removed the ground input from the BII and... Voilla! Sound!
Also BTW, lowering the ASIO latency helps the Opus boards even more than the BII.
Thanks and best regards,
Frank
Last edited:
Hi Frank,
Thanks for sharing.
It seems Windows Audio Session (WASAPI) in exclusive mode and ASIO are designed to do the same thing, that is, bypass the windows mixer and send the data straight to the device driver.
If I remember correctly, ASIO was developed to bypass Kmixer in windows xp and WASAPI exclusive will also bypass the windows mixer in Windows Vista/7.
If your player can set WASAPI exclusive, and your drivers support windows 7 you might want to try not using ASIO
Thanks for sharing.
It seems Windows Audio Session (WASAPI) in exclusive mode and ASIO are designed to do the same thing, that is, bypass the windows mixer and send the data straight to the device driver.
If I remember correctly, ASIO was developed to bypass Kmixer in windows xp and WASAPI exclusive will also bypass the windows mixer in Windows Vista/7.
If your player can set WASAPI exclusive, and your drivers support windows 7 you might want to try not using ASIO
Right - the 2 channels that feed the crossover software are (in this case unnecessarily) available to a DAC. However, the signal is unfiltered and as such unsuitable for bi-amping. The player software is only stereo or 5.1 with no interface for frequency or timing control.
Frank
ahh brain fart, you're right, I was thinking of USB, have been putting together USB->I2S modules. if it was differential it wouldnt be so prone to noise and need such short cable runs
Yes. You can even use differential drivers and make it a differential signal
You will want to carefully manage latency between the signals.
Thak you. I used before 485 (I believe that was number) transmiter but that was for S/PDIF signal and it worked perfect. I never worked with I2S though.
Yes. You can even use differential drivers and make it a differential signal
You will want to carefully manage latency between the signals.
As you may already know, PS-Audio advocates I2S on LVDS/HDMI connector & cable. Its reference schematic is available here.
http://www.fidelix.jp/img/PS-HDMI.jpg
The standard employs HDMI connector and cable. However, it is not HDMI signal at all. It is a differential transmission of I2S signals.
As far as I know, CAPRICE DAC by Japanese Fidelix (Rx side), Wyred 4 Sound DAC 2 (Rx side) and SDTrans192 Rev. 3.0 (Tx side) have implemented this standard other than PS-Audio.
In my case, I connected SDTrans and CAPRICE with 50 cm HDMI cable and found it feasible for up to 352.8 kHz/24 bit.
Hi Bunpei,
You were so kind to help me get my hands on a set of these fidelix I2S/HDMI transmitter/receiver boards and I plan to use these as an umbilical cord between my DAC (a dual-mono Buffalo II) and a yet to be constructed/implemented "digital input selection and DAC control unit". A great feature of the fidelix implementation is that the HDMI cable also carry the I2C signal so the DAC operation can be controlled/monitored remotely.
One question: somewhere on the documentation I received together with the fidelix modules I recall to have read "DSD/I2S" transmitter/receiver", suggesting that these module can transmit/receive also DSD signal (and possibly SPDIF). Have you tried this or only the I2S format?
You were so kind to help me get my hands on a set of these fidelix I2S/HDMI transmitter/receiver boards and I plan to use these as an umbilical cord between my DAC (a dual-mono Buffalo II) and a yet to be constructed/implemented "digital input selection and DAC control unit". A great feature of the fidelix implementation is that the HDMI cable also carry the I2C signal so the DAC operation can be controlled/monitored remotely.
One question: somewhere on the documentation I received together with the fidelix modules I recall to have read "DSD/I2S" transmitter/receiver", suggesting that these module can transmit/receive also DSD signal (and possibly SPDIF). Have you tried this or only the I2S format?
I also tried a DSD connection between ElectrArt USB Audio Board and Buffalo II using a pair of these interface boards and got successful results for both DSD64 and DSD128.
You can tap DSD signals from your SACD player and extend those signals to ES9018 based DACs with 50 cm or 1m HDMI cable without problems using these boards.
I have not tried it for S/PDIF.
Thank you very much for this post. That's the way to do it. Now if I read correctly there are boards that performs the conversion? Do you know where to look for it?
A Japanese technology oriented vendor, Fidelix, provides a board level component.
http://www.fidelix.jp/img/I2S-HDMI.gif
http://www.fidelix.jp/img/I2S-HDMI%20PCB.jpg
The receiver module incorporates an isolator device and its power can be supplied from a transmitter side.
Its price is 15,000 JPY. Their contact address is info@fidelix.jp