• 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

Do you have any suggestions about this case?

As I am neither supporting staff of exaDevices nor that of Twisted Pear Audio, please regard my consideration as just a private speculation.

I think the frequency of oscillator device on the Buffalo II board has no direct relation with this issue. However, TPA might have changed ES9018 register setting for DPLL bandwidth parameter at the same timing when they upgraded previous 80 MHz oscillator to new 100 MHz one.
In my case, two new Buffalo II boards with 100 MHz oscillators shipped this year suffered the similar problems. After replacing the new MCU with old one that came with an old Buffalo II board with 80 MHz oscillator, the problem disappeared.

If you want to confirm this, one possible way is detaching the MCU from the board. This means you run the ES9018 DAC chip with full default settings. If you find no unlock problem then, the cause must be the MCU.
However, please remember that a sound volume will decrease in this case because of input distribution setting of the default set.

There may be two possible measures to solve the problem when my speculation is correct. One way is to obtain a MCU of old version programming and another way is to prepare your own MCU that can set appropriate register values on the DAC chip.

Bunpei
 
If you want to emulate a passive analog crossover made with capacitors and inductors, then Apple provides low-pass and high-pass AudioUnits for free - with the operating system. You just need an application that can host them.

Thank you Rsdio for the advice. I did almost the same, but through the Logic. I was planning to share it here, so here is the opportunity. Hopefully someone could use it, just like I learned from you and RayCtech

My situation, and believe anyone else's doing active on Mac is that although there is an exceptional application - Pure Music that hosts its own crossover, it doesn't cover everything. In another word, I could play my files from the hard disc, but I cannot simply slip CD into and play, or I cannot play streaming radio stations. Now, for the highest fidelity I could do no better than PM. Plays from memory, plays easily FLAC files, uses iPad to control, and uses iTunes for file management. Awesome, but sometimes I just want to stream my favorite Absolute radio station... or pop fast CD. So what than if my crossover is hosted by Pure Music?

I used Jack for rerouting to Logic that is hosting my plug ins. From Logic I go out to whatever my output is, for now RME card, soon replaced with I2S to USB card. I tried very popular SoundFlower, but that did not work for me. Instead when I tried Jack it work flawlessly, and it comes with good documentation.

Logic is an awesome app, and plug ins that comes with it are really good. As far as I am aware, there are no crossover plug ins for Mac, but as you suggested using low shelf, high shelf or bandass filters do a perfect job. As you could see, I was able to control the angle of the curve in dB just like with any crossover, and filters used are linear phase. The amount of additional filters that could be used if desired, such as room correction or similar is limited just by memory and processing power.

This way my crossover is independent of the music playing apps, and I could combine various players for various needs. In my listening tests I did not hear any difference comparing to regular crossover (the one implemented in Pure Music). I am happy camper so far.
 

Attachments

  • Screen-shot-2011-05-26-at-2.19.jpg
    Screen-shot-2011-05-26-at-2.19.jpg
    196.5 KB · Views: 483
I used Jack for rerouting to Logic that is hosting my plug ins. From Logic I go out to whatever my output is, for now RME card, soon replaced with I2S to USB card. I tried very popular SoundFlower, but that did not work for me. Instead when I tried Jack it work flawlessly, and it comes with good documentation.

Logic is an awesome app, and plug ins that comes with it are really good. As far as I am aware, there are no crossover plug ins for Mac, but as you suggested using low shelf, high shelf or bandass filters do a perfect job. As you could see, I was able to control the angle of the curve in dB just like with any crossover, and filters used are linear phase. The amount of additional filters that could be used if desired, such as room correction or similar is limited just by memory and processing power.
The shelving filters are not appropriate for crossover use, but I see that in Logic you are actually using the high-pass and low-pass that are appropriate.

For folks who have not purchased Logic, but who perhaps have installed the Apple developer tools, you might consider AULab. It is a free application from Apple that hosts AudioUnits and can work with multichannel hosting. You should be able to combine it with Jack, although I have never tried this combination. I bought a license for Logic Pro, but I often use AULab for real-time processing or analysis, so I thought I would share this option for folks who do not want to pay.
 
Just a little note of experience with my exaU2I and Buffalo II DAC.

I downloaded some music from High Resolution Music DOWNLOAD services .:. FLAC in free TEST BENCH, its in 24BIT/192kHz res. With my exaU2I connected to the Buffalo II DAC I tried to play this music using Foobar2000. The sound was glitching every other second and the glitches were like this during a couple of seconds: . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . .
if you understand what I mean, it went like in a loop before it sounded perfect in a couple of seconds before it started all over again...

I tried everything, playing files locally instead of from my network drive. I also tried to increase the buffer size, but nothing really worked...

Then I noted that someone talked about the firmware on the Buffalo II DAC, that it had changed somewhat during the shipments...

So this is what I did, note that my friend Oblivion tried this before me...:
I took out the firmware chip from the Buff II DAC, then powered it up all over again and started to play the music with 24BIT/192kHz resolution. Guess what happened!! I sounded perfect with no glithing at all!! I haven't got a clue why they release such a firmware on the DAC without properly testing, but this solved anything for me. Removing the firmware did the trick. And it still sounds fantastic!

Anyone else with trouble like this?
 
I haven't got a clue why they release such a firmware on the DAC without properly testing, but this solved anything for me. Removing the firmware did the trick. And it still sounds fantastic!

Anyone else with trouble like this?

Just about two months ago. I posted the issue on their support forum.
DPLL Bandwidth setting on B II - Buffalo DAC - Twisted Pear Audio Support
However, I have received no answers from them so far.

Any DPLL Bandwidth parameter other than "default best" is very hard for any transports of I2S output to catch up because the DPLL frequency might be 64 times higher for I2S than that of S/PDIF.
 
yeah well everything is set to defaults this way isnt it?

AR2, great to see you are having success with logic, you can also get convolver for logic for room correction, its a bit of a processor hog, but does the job very well. for logic you can also get hardware acceleration for the plugins if you wish to use higher resolution, or more channels. i would recommend waves also, which has a superb eq. you would only need the basic set, not diamond, or whatever the top level is these days.

you can get an ipad app for logic control also, which would make it easier to sit in the listening position when doing any adjustments
 
Last edited:
hang on AR2, which rme have you got? depending which, you can use totalmix to route an output back to the input without wires and access the audio that way without needing jack at all. just control click the channel strip of the output (see below) that has the data you want playing so that it turns red, then in logic you set the outputs of your crossover to the channels after that. you lose 2 channels, but you dont need to use another application. then as you are probably doing, you hit record and pause to process any audio playing through the plugins and assign your channels. make sense? the output from the bounce will be seen at the corresponding inputs, so to monitor a bounce from channels 1 and 2 as here, from memory they will be available on input channels 1 and 2. if thats wrong just play till you get them, or send me a pm and i'll double check, its been a while since i have used it in this way.
 

Attachments

  • Picture 440.png
    Picture 440.png
    121.9 KB · Views: 319
Last edited:
Supra was unable to make any progress with his Buffalo II. Removing the MCU in his case produced only silence.

Then is might simply be that the SPDIF switch are in the wrong position...

Or that it is a dual mono setup with two Buffalo´s.
If you take out the MCU / Firmware on a Buffalo setup for mono mode there WILL be silence due to the outputs then will be in phase - the MCU / firmware runs half of the outputs in anti-phase in mono mode due to the way the inputs are connected :confused:

If there are silence with the MCU / firmware out on a Buffalo then it must be some kind of error in the connections...
Wrong lineup of I2S signals, GND connected wrong etc..

I observed Bunpei posted while I was writing.....
 
Thanks it everyone for their suggestions-I connected it up again and this time it works perfectly, no dropout loops at higher 96k playback.
Thanks again to everyone for advice and especially to exa065 for his back up and help to get the exaU2I working with the buff 2. Have to say I'm really pissed off with the Twisted Pear Audio administrators - they see these sites and their only input is to claim there's not a problem, when they know this has been a problem for many users for some time.
Even a private message was ignored.
Ok I got that off my chest, the good news is that this sounds pretty darn good, I have a pretty accurate system that is also very transparent and I tell you what folks, this is what the master tapes really sound like.
Theres no coloration or augmentation , this is what it is.
What are the implications of using the Buff2 without the MUC chip in I2S mode?
Obviously spdif is out, but is there a function for the MUC in I2S?
Any other audio company would get replacements to customers but I won't hold my breath here.
Thanks again, very relieved to get some tunes.
 
Even a private message was ignored.

Hey, I just got your PM this morning and replied. I also just saw this.

Russ just posted about it here as well: DPLL Bandwidth setting on B II - Buffalo DAC - Twisted Pear Audio Support

So, the uC on the Buffalo is set to use the optimal DPLL setting, which in our testing with both the XMOS eval and our XMOS solution in development, has provided the best and lowest jitter performance. We have not used this USB interface, so I hesitate to offer an explanation.

We have had no issues with high-res material and the current firmware on the Buffalo, or we would have changed it. I would stress that most people using the Buffalo for high-res are not having any issues.

If you are having success w/o the firmware chip in place, then the default DPLL settings are what you want, as it can deal with higher jitter from the source (lower jitter rejection). Running w/o the firmware, however, means you have incorrect DAC mapping for the I2S signals, so that is not good either.

We will make a "High Bandwadth" version of the firmware available to those who want it with the less optimal default DPLL. That's totally not a problem. I will add a link to the web site for people to order it (will post an URL when I have it set up).
 
Last edited:
...
Obviously spdif is out, ...

SPDiF should be on with the default condition. Make sure the comparator switch in the BII board is in the right position

...Any other audio company would get replacements to customers but I won't hold my breath here.....

The real issues is that the ESS 9018 DAC has a 7-step tweakable DPLL setting below the lowest setting available in the older 9008 DAC, and that setting applies to both I2S and SPDIF with SPDIF possibly having a larger BW on the same setting as Mr Bunpei said. I have no doubt that the shipping firmware for the DAC has been tweaked for best performace rather than best customer satisfaction because this is DIY

If you are going "all out" with the exa device, then the best solution is to tweak the setting that can match your configuration with separate settings for SPDIF and I2S.
 
Member
Joined 2009
Paid Member
If you are going "all out" with the exa device, then the best solution is to tweak the setting that can match your configuration with separate settings for SPDIF and I2S.
The "best" solution would logically be to improve the jitter performance of the exa device....
Maybe this is the way RayCtech power supply modifications improves the audio quality of the device.
I seem to recall that glt was able to get a better lock with a stringent DPLL setting on Buffalo after upgrading the PSU on a USB/I2S transport.
Just a guess - would be interesting to investigate.
Cheers,
Nic

P.S. Any news regarding OSX drivers :)
 
hang on AR2, which rme have you got? depending which, you can use totalmix to route an output back to the input without wires and access the audio that way without needing jack at all. just control click the channel strip of the output (see below) that has the data you want playing so that it turns red, then in logic you set the outputs of your crossover to the channels after that. you lose 2 channels, but you dont need to use another application. then as you are probably doing, you hit record and pause to process any audio playing through the plugins and assign your channels. make sense? the output from the bounce will be seen at the corresponding inputs, so to monitor a bounce from channels 1 and 2 as here, from memory they will be available on input channels 1 and 2. if thats wrong just play till you get them, or send me a pm and i'll double check, its been a while since i have used it in this way.

Quisp, thank you for the treasure of info you posted here. really cool stuff.
Many questions, but here are some answers. Until just recently I was using Fireface 800, but I gave it to my son. Now I use Fireface 400. Your advice on rerouting RME is really cool, but my intention is not to use RME interface but instead USB - I2S converter. I am not sure I could use both at the same time? If that is not the case than you solution is certainly way to go. Wave - I have a full collection, quite honestly there are way to many plug ins there, haha I get lost there finding what I need. I wanted to use native Logic plug ins to simplify life, but certainly I will try wave as well. I might have a few :headbash:questions, but I will send it to you directly, in order not to pollute this thread, you are correct on that one.

I just collected all the parts for my future set up. Now I am up to machining and doing case work to house all. Power supplies are elaborate and will be housed separately. Still trying to conceptualize to whole set up... Will keep you updated.

Thank you very much for such a comprehensive and valuable info. I hope there are other people who would try similar set ups.
 
The "best" solution would logically be to improve the jitter performance of the exa device....
Maybe this is the way RayCtech power supply modifications improves the audio quality of the device.
I seem to recall that glt was able to get a better lock with a stringent DPLL setting on Buffalo after upgrading the PSU on a USB/I2S transport.
Just a guess - would be interesting to investigate.
Cheers,
Nic

P.S. Any news regarding OSX drivers :)

The M2hightec Evo and Teradak usb-spdif devices with I2S outputs also have problem locking onto buff, so I don't think it is any problem with exadevices, anyway I'm grateful that TPA have come to the party and will provide appropriate firmware.
 
The "best" solution would logically be to improve the jitter performance of the exa device....
Maybe this is the way RayCtech power supply modifications improves the audio quality of the device.
I seem to recall that glt was able to get a better lock with a stringent DPLL setting on Buffalo after upgrading the PSU on a USB/I2S transport.
Just a guess - would be interesting to investigate.
Cheers,
Nic

P.S. Any news regarding OSX drivers :)

Hi Nic,
Supra has experienced the same issues with 3 different I2S boards. Bunpei has reported the same behavior with his SD Card Player. It appears many I2S devices on the market require flexibility with the Buffalo DAC settings. We all welcome the support from TPA on this issue.

As I have mentioned before, the jitter levels of the exaU2I board are well within the jitter cancellation capability of the ES9018 chip. Evidence of this is the customer satisfaction level and the measurements taken on the analogue output of exaU2I-ES9019 couple. It is easy to pick on any device by taking one specification out of the context of the overall result. There is integrity to the exaU2I that is more important tan one specific measurement. For sure the device is not perfect, however it is the first and only solution on the market that can offer 32/384 quad for less than $12000.

Mac drivers are under development. We actually enjoy the Mac development environment. However I will miss the Foobar sophistication and freedom when I move on the Mac side :)

-exa065
 
We will make a "High Bandwadth" version of the firmware available to those who want it with the less optimal default DPLL. That's totally not a problem. I will add a link to the web site for people to order it (will post an URL when I have it set up).

Hi Brian,
Thank you for offering a solution for using third-party I2S devices. A Buffalo DAC-exaU2I combo offers performance that is hard to match with consumer or professional gear.

-exa065