I2S from Juli@ PCI sound card

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This is my first post. I have learned quite a bit from reading the various forums, and I am giving a little in return.

This post does not keep entirely in the spirit of building your own DACs but it is a simple DIY cable which gives I2S output from a PCI sound card at up to 24/96 and possibly 24/192. Flame me if you must.

I will get some diagrams posted, but for now just wanted to document this idea. Pictures of the Juli@ can be found on http://www.digit-life.com/articles2/esi-julia/index.html and the I2S pinout on http://www.diyaudio.com/forums/showthread.php?postid=179778

I have an ESI Juli@ sound card ( new from ebay about $129 ) and was using the spdif to drive an external DAC in the Box. I upgraded to a used Perpetual P-3A and wanted to investigate an I2S connection. I use foobar2000 and Kernel Streaming and flac files.

The Juli@ has an interesting design - there are two circuit boards connected by dual in line headers ( a 20 pin designated J3 and a 10 pin designated J7). The "digital" board has the PCI connection and the Envy-24 HT ( 1724 ) controller chip. The analog board has an AK4358 DAC and an AKM ADC chip as well. The "analog" board can be rotated 180 degrees to change from RCA unbalanced connectors to Balanced 1/4" Phono plugs. Therefore there are matching connectors on opposite sides of the analog board.

I downloaded the datasheets for the Envy and the AKM DAC. The 1724 is capable of multichannel I2S outputs, and the AK4358 has multiple I2S inputs. I figured the I2S interface was going through the pin headers from digital to analog boards.

I took a multimeter and buzzed out the connectors and the chips, and found the following signals:

J3 ( 20 pin header )

J3 Pin 1 -> 1724 PSDOUT[0] ( I2S data out) -> AK4348 SDTI1 (I2S data in)

J3 Pin 5 -> 1724 PSYNC ( I2S Word Clock ) -> AK4348 LRCK (I2S Left/Right Word clock)

J3 Pin 7 -> 1724 PBCLK ( I2S bit clock out ) -> AK4348 BICK ( I2S bit clock in )

J3 Pin 9 - > Xilinx on the digita board -> AK4348 MCLK ( I2S Master Clock in )

J7 ( 10 pin header )

J7 Pin 5 -> Digital VSS ( Actually several pins on J7 are VSS )

Here I had all the I2S signals ( Except DEEM as in Audio alchemy, but that does not exist on the Juli@ board )

Since there are two sets of headers, one set is on the top edge of the analog board, and has nothing plugged into it. I took an old keyboard cable with a 6 pin mini-Din connector and soldered the other end to wire wrap type pins, and plugged them into the signals listed above. The cable is shielded and the shield drain wire is connected to the mini-din connector shroud.

I simply broke off the sixth pin inside the mini-din to yield a 5 pin mini-din, and connected it up.

Believe it or not, this works perfectly with the P-3A DAC. I get I2S lock and can drive it with 16/44 up to 24/96 signals. It sounds great on my system, much better than the spdif connection. The only weirdness is the P-3A phase invert light is on all the time. Maybe I need to invert the word clock or something. I may send an email to Perpetual Tech and see what they say.

Now, I admit this is cheap and dirty. I should use better quality cable, connectors and probably buffer the signals with a 74LS244 or something. But it works and I am digging it for now.
 
Thanks for your reply.

I forgot to mention that I have to configure the Juli@ as Juli@ Ch12 or Ch1234. However spdif does not seem to work with Ch1234, only if I use Ch34 alone. Not sure why Ch1234 does not drive both together.

The specs say that 192KHz only works on coax digital output, but I am getting only noise through the DAC if I upsample to 192kHz. I am not sure if it is the upsampler, PCI card or DAC at this point. However Secret Rabbit Code with 96kHz upsampling works fine and the DAC shows 96kHz signal.

The cable is about 1 meter or so, to reach from the pc under the desk to the DAC on the desk. There is one foil shield wrapped around the conductors, and an uninsulated stranded drain wire inside the foil that is tied to the connector.

I will put a scope on the signals maybe this weekend and see if there is loading or ringing.
 
Juli supports 24/192 at the DACs, its I2S outputs offer this mode.

Since this specific card outputs stereo only, the pin header holds only one I2S channel from the ICE1724 controller. The remaining 3 I2S signals must be taken directly from ICE1724 pins, specifications e.g. http://alsa.cybermirror.org/manuals/icensemble/Envy24HT091DS.pdf . Of course the original Juli windows drivers do not support the other three channel-pairs, that is the closed-source system tax :)
 
phofman said:
Juli supports 24/192 at the DACs, its I2S outputs offer this mode.

I asked because it says that the digital output supports 96k with optical spdif and 192k only with coaxial spdif:
http://www.esi-audio.com/products/julia/

I wonder what the drivers (win) would do using i2s.

Also, there is any pin that could be used for wordclock input?
The nice thing of the Juli@ compared for instance to the maya44 is the use of two different crystals for the internal clock.
 
Telstar said:
I asked because it says that the digital output supports 96k with optical spdif and 192k only with coaxial spdif:
http://www.esi-audio.com/products/julia/

The limiting factor is maximum transmission rate of TOSLINK. The older models run up to 6MHz, 32 x 2 x 96000 = 6 144 000 bps. It is not a fixed limit, my Prodigy192 TOSLINK with the same specifications runs at 192kHz.

The output TOSLINK connector is hooked in parallel to the SPDIF isolation transformer, there is no software limit.


I wonder what the drivers (win) would do using i2s.
The drivers themselves do not care whether the original DAC or your DAC is hooked to the I2S output of ICE1724 (unless they detect presence of the analog daughter board which is easy to circumvent). The problem is the windows driver does not configure ICE1724 for the remaining 3 stereo channels, and defines only one stereo device in the sound layer. Arranging that in linux is a matter of changing several lines in the driver.


Also, there is any pin that could be used for wordclock input?

Juli can be clocked by signal recovered from incoming SPDIF. If you want a regular wordclock input, check out e.g. http://www.infra-sonic.com/site/p_PCI_quartet.php . You can locate the appropriate pin of Juli's SPDIF receiver AK4114 too, both cards use the same controller.


The nice thing of the Juli@ compared for instance to the maya44 is the use of two different crystals for the internal clock.

All ICE1712/24-based cards have two crystals, incl. Maya44.
 
Infrasonic Quartet

phofman said:

Juli can be clocked by signal recovered from incoming SPDIF. If you want a regular wordclock input, check out e.g. http://www.infra-sonic.com/site/p_PCI_quartet.php . You can locate the appropriate pin of Juli's SPDIF receiver AK4114 too, both cards use the same controller.

All ICE1712/24-based cards have two crystals, incl. Maya44.

Not sure why in the picture the second crystal is not shown, but you are right, according to VIA specs.

According to this site:
http://envy24.svobodno.com/

The three cards using both the envy24ht(or s) and AKM 4114/4112 DIT are the Juli@, the Quartet and m-audio audiophile 192.

The Quartet looks the best of the tier: it has the advantage of having driver support (in Vista) for external clock, ready to use bnc wordclock input, and also a mic preamp. Sounds ideal for my needs.

Price and availability are both very good too, thanks to Auzentech distribution. Only pity is the pci interface (wonder if they'll ever make a pci-e version).

Are there any problems with this card that you know?
 
Re: Infrasonic Quartet

Telstar said:
The three cards using both the envy24ht(or s) and AKM 4114/4112 DIT are the Juli@, the Quartet and m-audio audiophile 192.

Plus the cards supporting the MI/ODI/O addon card, such as Prodigy192, Waveterminal 192, etc.

[/B]
The Quartet looks the best of the tier: it has the advantage of having driver support (in Vista) for external clock, ready to use bnc wordclock input, and also a mic preamp. Sounds ideal for my needs.
[/B]

It certainly looks good, thoughtfully designed for needs of semi-pro musicians.

[/B]
Are there any problems with this card that you know? [/B]

The card sample just arrived from the Korean manufacturer. While working on its linux driver, I hope any potential problems should be revealed. But honestly, I do not expect any.
 
Yes and in the same way the original poster did. Except you don't have the luxury of the connector thingy that the Juli@ has. Instead you have to solder wires directly onto the PCB at the correct locations.

Look for a Delta 1010 (not the LT - the full version with the breakout box). All the I2S lines are brought out to a DB25 connector. Pricey even used, but certainly the easiest approach. I2S going through a connector won't be as pure as would be ideal, but if you're good with the iron you can remove the connector.
 
anyone buffering the sigal with say a buf634 or diamond buffer? wondering if that makes any sense. currently i'm using a modified breakout cable on my upgraded RME 9632 (power supply/filtering/decoupling) to supply spdif and AES with BNC and XLR on a cryo silver twin coax cable. but I have just recently installed an ethernet neutrik connector on the i2s lines of my dac and wondering if I should have a play with the julie@. I had mistakenly thought the card was a PC and linux only thing and since i'm on mac this isnt any use, at least not my G5 anyway. certainly looks worth a shot, is multichannel i2s doable in mac? i'm not afraid of taking the iron to the cards tender bits.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.