SunRa,
Your project is very interesting for all of us on this thread. You are venturing into uncharted territory. exaU2I has never been tested with AK4396.
The more you clean and simplify the signal path, the more the sound will open up in terms of details and realism. There is no end to it. At this point you are enjoying the benefits of excellent player, driver and interface. Once you finish the DAC you may feel compelled to perfect the amplifiers, the speakers and eventually the room
.Well I definitely feel the need of a speaker upgrade after this last week-end!
Now I've discovered an odd thing when playing with the U2I. I have two files at 352.8Khz 32 bits and I didn't even considered playing these in foobar without a sample rate converter. Now I accidentally played one in foobar and guess what, I was hearing music. Now how on earth is this possible?? AK4396 can play higher sampling rates?? How? I am all settled regarding the 32bits matter, I am sure the chip is ignoring the last bits. Still, what about the sampling rate?
Now here's another funny thing: I also have some 24 bit / 352.8 files. And these are playing also, only that it's also coming with a terrible hiss noise. The only difference I get in Foobar is that when playing the 32bit ones it says PCM (floating point) followed by bitrate and sampling rate.
Any ideas on what is happening?
Now I've discovered an odd thing when playing with the U2I. I have two files at 352.8Khz 32 bits and I didn't even considered playing these in foobar without a sample rate converter. Now I accidentally played one in foobar and guess what, I was hearing music. Now how on earth is this possible?? AK4396 can play higher sampling rates?? How? I am all settled regarding the 32bits matter, I am sure the chip is ignoring the last bits. Still, what about the sampling rate?
Now here's another funny thing: I also have some 24 bit / 352.8 files. And these are playing also, only that it's also coming with a terrible hiss noise. The only difference I get in Foobar is that when playing the 32bit ones it says PCM (floating point) followed by bitrate and sampling rate.
Any ideas on what is happening?
SunRa, we need to investigate.
The most suspicious issue is the floating point. The exaU2I ASIO driver supports only Integer formats. Can you provide screen-shots of Foobar reporting floating point? Is it reporting floating point source file or floating point output? I would like to see the Foobar Status Bar and the media properties pane during playback. Please verify that there are no DSP plugins loaded. Check also the settings window - Output Settings. When ASIO output is selected the drop-down for output format should be blank and disabled. The ASIO driver takes over the control of the output format.
You may be discovering something here. If your Foobar configuration is correct and there is no implied re-sampling than.... your DAC may be more capable than expected.
We also experienced noise and distortion in some modes during our initial tests with ES9018. I would like to play the same 24 bit / 352.8 files with my DAC. I should test with the exact same Foobar settings as you. This will allow us to narrow down the root cause of the noise problem.
Cheers,
exa065
The funny thing is that when building this DAC I researched the white papers released by Anagram regarding their technology used in Cambridge CD players. Basically they were able to upsample 44.1K to 384k, bypass the DAC's digital filter and play this through AD1955 and Wolfson parts.
Here's the screen. I also can confirm the same behavior with the JRiver Media Center. Oh and I have no DSP that can affect the sample rate of the file.
Here's the screen. I also can confirm the same behavior with the JRiver Media Center. Oh and I have no DSP that can affect the sample rate of the file.
Attachments
Last edited:
Your setup from the screen shot is correct and identical to mine. When I play Carmen in the original WAV format I also have the (floating point) notice the way you do. This is the source format. Normally I listen tho this piece converted to WavPack, which is Integer. That's why I never noticed the message.
I trust Foobar, J. River and exaU2I, so congratulations! You made history! You are playing 352.8 kHz.
Last edited:
Hi exa,
Well, the LED indicators seem to tell that's 352.8K I am playing alright. I'll take a closer look tomorrow on this. It would be great checking this with an oscilloscope, maybe I can find something within the DIY community here in Bucharest.
The 24 bits files causing the noise are from HERE the bottom half of the page. When playing these through cPlay (downsampled to 192) they are OK.
Can't tell why the 32 bit files are crystal clear and all of the 24 bit files are making that hiss in the background.
Anyway, this thing is getting really exciting...
Well, the LED indicators seem to tell that's 352.8K I am playing alright. I'll take a closer look tomorrow on this. It would be great checking this with an oscilloscope, maybe I can find something within the DIY community here in Bucharest.
The 24 bits files causing the noise are from HERE the bottom half of the page. When playing these through cPlay (downsampled to 192) they are OK.
Can't tell why the 32 bit files are crystal clear and all of the 24 bit files are making that hiss in the background.
Anyway, this thing is getting really exciting...
You just "drop" Flac files in the Pure Music "Flac player" window..
There are several optional configurations setting and you choose what suits you.
With the default setup Pure Music adds a dummy album/melody in iTunes / iTunes database with tagging from the FLAC file.
These "dummies" are just for integration of the FLAC files in iTunes database as the FLAC, AIFF, WAV, DXD, etc. are played directly by Pure Music.
I use full memory play and with DXD and AIFF/WAV files that are upsampled (on disk) to 352.8k/32bit you need either a fast local disk or a 1Gbit network with a fast file server and it still takes 5 to 10 second to load the 500MB to over 1GB files.
I will experiment with hybrid play where the file starts playing before the whole file are loaded into memory.
If the exa unit would show any differences in case those "audiophile" players
are used, we'd know that the device fails to decouple properly from the PC.
However. No worries.
Quote from homepage EXADEVICES:
I'm sure they verified what they're claiming. I mean... ...they call their unit an ultra-high-end device.
Hi SunRa,
I checked my DXD 352.8k/24bit files (23 albums) and all are PCM integer and plays without any kind of noises.
If you have a pointer to where the 32 bit files can be downloaded I would like to test those also.
I will also convert some DXD files to floating point format and check what happens then...
Have you checked the input data format your DAC chip defaults to?
You should be able to change between several formats...
As your DAC only uses 24 bit of the 32 bits that are clocked in each word clock frame it is possible that there are some misalignment like left vs. right justified etc..??
For the time beeing I uses a M2Tech USB -> I2S devices with OSX, and the exaU2I with Windows 7..When I evaluated (some time back) OSX with iTunes and other players, and Windows with J.River and Foobar2000 and used the M2Tech device I found the OSX based players to outperformed ALL the Windows based setups.
Now I can only compare OSX - iTunes/Pure Music using the M2Tech with Windows 7 - Foobar2000/J.River using the exaU2I.
At the moment the Windows 7 setups using exaU2I outperforms the OSX setups using the M2Tech by a large margin..
I am curious to the Fidelity results when the exaU2I can be used with OSX.
Hi RayTech,
Thanks for your input. The DAC is set to I2S compatibility mode, I don't believe there's a problem with that.
Any other lower sampling rate files at 24 bits play well, without noise. Only the 352.8 from the above link are having that noise. The noise is not evident when playing through cPlay and downsample the files to 192Khz (real-time). And yes, foobar tells me that these are integer files.
As you can see in the screenshot the 32bit files are floating point. And they also play well. Is it because of the bit rate? Is it because of integer/floating point? None of this makes sense because all the other integer, 24 bit files I have play without noise.
So it must be something in relation with all these three aspects: 352.8K sampling rate, bit depth and maybe floating point/integer.
I plan to diether the 352.8k 24 bit files to 32 bit files and check what happens. I also plan to convert the 32 bit files I have to 24 bit and also play with WavPack. I'll tell you more about my findings.
Thanks for your input. The DAC is set to I2S compatibility mode, I don't believe there's a problem with that.
Any other lower sampling rate files at 24 bits play well, without noise. Only the 352.8 from the above link are having that noise. The noise is not evident when playing through cPlay and downsample the files to 192Khz (real-time). And yes, foobar tells me that these are integer files.
As you can see in the screenshot the 32bit files are floating point. And they also play well. Is it because of the bit rate? Is it because of integer/floating point? None of this makes sense because all the other integer, 24 bit files I have play without noise.
So it must be something in relation with all these three aspects: 352.8K sampling rate, bit depth and maybe floating point/integer.
I plan to diether the 352.8k 24 bit files to 32 bit files and check what happens. I also plan to convert the 32 bit files I have to 24 bit and also play with WavPack. I'll tell you more about my findings.
When I evaluated (some time back) OSX with iTunes and other players, and Windows with J.River and Foobar2000 and used the M2Tech device I found the OSX based players to outperformed ALL the Windows based setups.
Now I can only compare OSX - iTunes/Pure Music using the M2Tech with Windows 7 - Foobar2000/J.River using the exaU2I.
At the moment the Windows 7 setups using exaU2I outperforms the OSX setups using the M2Tech by a large margin..
I am curious to the Fidelity results when the exaU2I can be used with OSX.
On Windows it needs a bit more effort to squeeze better stuff out. It's a bit tricky.
Linux and OSX got a better base. It's a bit easier. Though the rules are the same.
"Less is more".
Some advise for setting up Windows 7:
1. Use Wasapi (foobar & JRiver) or Wasapi event style (JRiver) - The todays interface modes of choice ; Wasapi event style for async USB DACS in particular.
2. U need to select full file ram buffering modes (available in foobar & afaik JRiver)
3. And run Fidelizer
Cheers
Last edited:
I believe you at least partly misunderstood my posting
The Windows setups was NOT a problem..
It was the improvement with the exaU2I in the Windows setups compared to the M2Tech..
If the OSX setups improves by the same degree of magnitude with the exaU2I I will be delighted...
EDIT: I expect the Fidelizer will do a basic level of the job to tune the PC..
I have started by selecting all hardware parts and tweaked the hardware before the tweaking of the OS and then the player applications...
Last edited:
I believe you at least partly misunderstood my posting
No. I just keep flooding this thread with a certain amount of cynicism.
(and a little bit of information)
The Windows setup is always a problem.
No. I just keep flooding this thread with a certain amount of cynicism.
(and a little bit of information)
The Windows setup is always a problem.
I figured out the cynicism part easily enough...
With respect to the Windows setup issues I was a beta tester of Windows NT and a VMS (Windows NT = WNT is a shift left from VMS
) internals support engineer for years before that..I used Windows NT for over a year before I installed the first official Windows NT installation at the Norwegian Microsoft headquarters on the Microsoft product managers PC (the PC was also supplied..)
As a perfectly setup PC hardware and perfectly setup OS and player application are not even close to achieve perfect audio playback I requested exa065 / exaDevices to upgrade / change the ASIO drivers to rectify some of the issues that causes problems and thus will improve the Fidelity when corrected...
The ASIO drivers where upgraded and after some testing the new drivers will soon be ready for download by the exaU2I users.
As soon as I have the time available I will post these informations and some screenshots..
No. I just keep flooding this thread with a certain amount of cynicism.
(and a little bit of information)
The Windows setup is always a problem.
Let me see if I can summarize your last little bits of information:
- WASAPI is better than ASIO because it is the choice of audiophiles;
- However WASAPI needs Fidelizer to sound good;
- Windows setup is always a problem, including with WASAPI and Fidelizer.
- ASIO is the choice of the audio professionals;
- ASIO is so light-weight compared to WASAPI that it doesn't require squeezing the last bit of performance from Windows;
- For a change Windows with exaU2I ASIO is not a problem. Think of it - even some dye-hard MAC users have surrendered the comfort of their superior computing platform. They are prepared to use Windows just because of exaU2I until OS X drivers are developed.
You don't have a clue. You don't even understand what I'm writing about.
Your ASIO stuff probably needs Fidelizer even more then WASAPi. That's how I'd turn it.
I'm running Linux - most of the time. All those ASIO/WASAPI/USB2.0 problems just don't exist on a Linux platform.
Yep - there are other issues with Linux. I complained about that more than once. However, the basic structure is pretty well done.
The biggest issue is that especially audio eq. manufacturers don't support that platform.
BTW I used to run ASIO drivers. Long enough to stay away from those lightweights.
Professionals:
Taking a professional as reference. Hmmh.
1. How many technically well done recordings do we know?
2. How many of those professionals are crazy audiophiles?
I even have the feeling that since the audiophiles coming into play on
the PC side, certain shortcomings related to that pro-audio world became visible.
Proprietary drivers:
It's a pity that manufacturers, with potentially good products, force people to use a certain platform, just because they are not willing to comply to standards. They don't even realize that on the long run less cost is associated to the products if they'd comply to standards.
My advise is to stay away from (small) manufacturers with proprietary drivers. After 2-3 years your sound card might exist. That manufacturer might be gone by then - eaten up on the efforts to maintain proprietary software.
Cheers