thanks for sharing this, i have already started a thread on the xmos forum about such a piece of hardware https://www.xcore.com/viewtopic.php?f=47&t=5720
an engineer at Xcore forums says: It's a whole bunch of work. Host driver, hardware, embedded firmware. Many man years of work and $100Ks of investment I would think. And after all of that I'm not sure how transport of audio data over ethernet is any better/worse than USB, or any other bit perfect digital interface. I would never have imagined its so much work and money to make a PCI card with DSD output... this DAC doesnt use USB for DSD1024 (native), instead its i2s over a HDMI cable. what hardware can put i2s DSD1024 out? or maybe it will in the future? https://www.shenzhenaudio.com/matri...68khz-dsd1024-audio-dac.html?p=2#comment_html
The currently available XMOS platform can do DSD1024 (as claimed by diyinhk) with some programming effort. The "classic" Amanero can do DSD512..
All you need to do then is to convert this signal to LVDS and send it to the DAC. It's not difficult to do - there even exist a few low cost ready-made solutions. Of course if your XMOS or Amanero is inside your DAC's enclosure (which is usually the case), you don't need to do any I2S to LVDS conversions.
But in any case, we're always talking about USB devices.
All you need to do then is to convert this signal to LVDS and send it to the DAC. It's not difficult to do - there even exist a few low cost ready-made solutions. Of course if your XMOS or Amanero is inside your DAC's enclosure (which is usually the case), you don't need to do any I2S to LVDS conversions.
But in any case, we're always talking about USB devices.
there is a dac coming that uses HDMI cable for i2s native DSD1024, but what is the source of such an i2s signal? maybe a PCI card will soon follow?
https://www.shenzhenaudio.com/matri...68khz-dsd1024-audio-dac.html?p=2#comment_html
https://www.shenzhenaudio.com/matri...68khz-dsd1024-audio-dac.html?p=2#comment_html
It's easy to have an input that supports DSD1024 if the DAC chip supports it. You don't have to do anything "special" to support DSD1024 (vs. DSD64/128/256/etc.).
It does not mean that there exists such a compatible source for DSD (over HDMI cable).
I'm pretty sure that we've seen all that we'll ever see on the PCI front.. I don't expect any new I2S/DSD cards to come out. No reason any more, with the SBCs and USB boards that have become available.
It does not mean that there exists such a compatible source for DSD (over HDMI cable).
I'm pretty sure that we've seen all that we'll ever see on the PCI front.. I don't expect any new I2S/DSD cards to come out. No reason any more, with the SBCs and USB boards that have become available.
Yes, those SBCs look promising, but i have been looking for one that would do the job and havent yet found it. Please post links here if you know of any such SBC capable of i2s/DSD1024 prefferably with the latest skylake i7 support.
PS: reading some other forums, it is supported in theory, there is just no drivers (like the BBB botic) readily available for it. please post here if anyone has successfully done this
Sadly many believe that DSD is obsolete I hope it wont impact its evolution in the negative way:
Last edited:
I was talking about an SBC (like the RPi) with a USB to I2S/DSD board. But the upconverting to DSD1024 will need to be performed elsewhere, by a powerful computer.
An SBC is usually powered by a low power ARM processor. No need for an i7 since the heavy processing is done elsewhere. If you need to do the heavy processing locally, you need a powerful machine, not an SBC.
An SBC is usually powered by a low power ARM processor. No need for an i7 since the heavy processing is done elsewhere. If you need to do the heavy processing locally, you need a powerful machine, not an SBC.
I've used a BBB Botic at upto DSD512 but that's the max it'll be capable of as it works the CPU hard. I upsample to DSD256/512 using HQPlayer running on a powerful Xeon processor equipped workstation and stream the DSD output to the BBB Botic for rendering.
I observe though Presley that you still have no actual practical experience of a NoDAC solution and it begs the question of why you are looking for such an extreme implementation when you don't know whether NoDAC DSD playback is for you. As I've suggested before, build a NoDAC capable of DSD256, using a USB board, such as the JLSounds, and find out about how it presents music and learn more about how to implement a NoDAC so you can develop towards your ideal.
Last edited:
Your observation is correct and your words are wise, master Yoda 🙂 just kidding.
Well you know I am not going to use it only for music but for another purpose and here is the reason why I want o do such an extreme implementation, it will be an experiment.
But have no fear (because fear is the path to the dark side), i am ready to try the JLSounds~noDAC solution. Just tell me a few things i wanna know about it (i know I should read the whole thread but its very long and fragmented). If you like you can post a full presentation of the features, uses and design right here in one post (im sure others too would be interested in such a compact presentation).
1. how do i use the mute function exactly, i understand it makes noise between tracks but is it possible for the circuit to be auto muted (maybe with some additional signal via a COM port from the PC that the software would send on track switch, ever thought about it?) or do i have to actually press the mute button?
2. as I understand the higher the input frequency of the DSD signal, the better it performs. would the circuit be future proof, if later (lets say in a few years) removed from the JLSounds USB and hooked up directly on a i2s DSD1024 or DSD2048 signal from a SBC or PCI card? i know you havent tried but does the design allow it?
3. have you ever compared the USB noDAC to some other well built DAC (DSD and/or PCM) like the Subbu or some good DSD DAC?
Sorry, I don't have the time at present but the principle is simple, the info is here and I'm happy to try and field any questions, as I know, will others. Remember, this is DIY.
Mute is automatic, it uses the mute signal from the JLSounds board, but you can choose to have a manual switch too if you wish. No need for anything like your com port suggestion. The mute has an option for a variable time delay to tweak the muting if required.
Yes, for the SE version it is future proof in that the LP filter isn't DSD rate related. Whether the much higher frequency of DS1024 creates problems with the PCB tracks being sub-optimal at the higher rate remains to be seen. That said, the project is cheap so probably better to build another version incorporating any lesson learned.
I have compared my noDac to my Buffalo 3SE DAC (with the Buffalo running at DSD256 and DSD512 (and 384KHz PCM)). The B3SE sounds excellent. I think the noDAC (at DSD256) is a little better (just my opinion mind, YMMV) , if a little more troublesome regarding muting and noise.
Ray
If you are not on top of these question in general I think you will have a hard time reaching your goals as stated in post #1 - good luck anyways.
//
thats wonderfull news! on which page of the noDAC thread can i find a list of the necessary components to be put on the SE version and where exactly on the PCB are they to be soldered to? sory, but somehow i cant find it
The PCBs are essentially self-documenting, check out the pictures;
http://www.diyaudio.com/forums/digital-line-level/273474-best-dac-no-dac-105.html#post4649662
There are also schematics I posted.
Ray
best connection for noDAC (DSD only)
Hi everybody,
Great tread! I like the idea of an noDAC (an chip-less DSD only DAC) very much
OfcourseI would like to help you guys wherever I can.
Some thoughts of my about this matter:
- In my earlier reviews I wrote that that the I2S Pink Faun card was the way to go, but
that was before I heard the Amenero USB module (that works with USB)
https://amanero.com/
The problem is that I2S is an protocol developed in 1986 for CD redbook specs only.
also this protocol is designed for internal short distances and not for long connections
with cables.
Then a second problem is that I2S is developed for PCM only.
When I heard the Amenero module my prejudges about USB were taken away
and I had to admit that for high resolution playback USB is much better than I2S
Another thing:
The shortest path is often the best, and for the no-DAC design you could consider:
- the no-DAC should be "USB UAC2" compatible, so that the operating system sees the DAC as an external sound device.
For more info read my last segment of this review:
Audio Dandy - How to enjoy High Resolution Audio
This way the operating system doesn't do anything with the sound signal, and just
transport it directly to the DAC.
The last thing you want is that the signal first goes to ALSA sound stack and then through proprietary drivers to your DAC...
This path is way too long, and not preferred I think.
What dou guys think about the above?
Wishing you all the best!
Hi everybody,
Great tread! I like the idea of an noDAC (an chip-less DSD only DAC) very much
OfcourseI would like to help you guys wherever I can.
Some thoughts of my about this matter:
- In my earlier reviews I wrote that that the I2S Pink Faun card was the way to go, but
that was before I heard the Amenero USB module (that works with USB)
https://amanero.com/
The problem is that I2S is an protocol developed in 1986 for CD redbook specs only.
also this protocol is designed for internal short distances and not for long connections
with cables.
Then a second problem is that I2S is developed for PCM only.
When I heard the Amenero module my prejudges about USB were taken away
and I had to admit that for high resolution playback USB is much better than I2S
Another thing:
The shortest path is often the best, and for the no-DAC design you could consider:
- the no-DAC should be "USB UAC2" compatible, so that the operating system sees the DAC as an external sound device.
For more info read my last segment of this review:
Audio Dandy - How to enjoy High Resolution Audio
This way the operating system doesn't do anything with the sound signal, and just
transport it directly to the DAC.
The last thing you want is that the signal first goes to ALSA sound stack and then through proprietary drivers to your DAC...
This path is way too long, and not preferred I think.
What dou guys think about the above?
Wishing you all the best!
This noDAC is a very interesting option to me. However my speaker design (LinkwitzLabs LX-Mini +subs) requires 6 channels DSP for XO. The DSP solution I'm using doesn't support DSD.
Still, I'm interested in this topic of Isolating (or attenuating) the Computer Noise
from the USB signal.
I've attached a photo I found on the internet of the frequency of various clocks in an I2S setup. -Now I know that many PC's with Intel, Mac, or AMD processors we are running in modern day range from 1.8 Ghz to nearly 4 Ghz or maybe more. What would that look like on the photo I attached? Much faster obviously, but also a much lower amplitude in comparison with the signal, because it's noise. If the frequency of the switching-noise varies enough from the frequency of the (music)clock-signal, wouldn't it be benign? Or, what could be done to filter the switchnoise, and or amplify the clock signal. Obviously the ladder question is more relevant to noDAC from my limited understand of reading a little in that thread.
So PC World says that "The now-aging USB 2.0 standard can theoretically transfer data at a very high 480 megabits per second (mbps), or 60 megabytes per second (MBps). That's impressive, but not as much as the newer USB 3.0, which can handle up to 5gbps (640MBps)—over ten times as fast as the 2.0 maximum."
If a second or so of latency can be tolerated, wouldn't one simply make sure that the music clock rate is different than and not devisible by the USB clock rate, cut the power lines in the USB cable, and stack up USB data, in a self powered (with very clean PS) FIFO buffer. Next convert the USB info to I2S/DSD and feed that to noDAC?
I think you'll find you have some catching up to do, suggest you check out the NoDAC thread;
http://www.diyaudio.com/forums/digital-line-level/273474-best-dac-no-dac.html
I think you'll find that there are several working solutions doing isolating/reclocking that have been around for quite a while. Search for Acko SO3, Twisted Pear Cronus and IanCanada's FIFO.
Thank you kindly for your suggestions about Acko. I was not aware of them.
I know my ideas shared above are easier said than done, and I'm not a designer. I wonder though, in the example you listed of IanCanada's FIFO, doesn't the First In First Out buffer work on the PCM digital music data, before it sends to the DAC? I need to read to learn more about that, because I know little of what I'm talking about. However if my assumption is correct, then I want to point out that my idea is to FIFO/buffer the USB Data before converting it to I2S, rather than (or in addition to) FIFO/buffering the I2S data before sending it to DAC.
Keep up great work with noDAC and noUSB projects.. Thank you for all of your postings, they help us laymen to begin to grasp some concepts about digital audio.
Why?
Just isolate/reclock once, immediately before the I2S/DSD data is handed over to the DAC.
Incidentally, to add to my previous list, the JLSounds I2S over USB board has onboard isolation/reclocking and is excellent value for money.
- Status
- Not open for further replies.