Very interesting USB->I2S adapter with native DSD

[Javier]

I just found this:

hifiduino.wordpress.com/2012/08/29/another-usb-interface-for-diy/

There is GB going on right now at DIYaudio for this device so it can be gotten at 39€ (+ transport +VAT) instead of its regular price 79€ (+ transport +VAT). I think this is exactly what I was looking for not only to feed my currently stopped Will's PK project but as great base for future higher end DACs (Buffalo III?) that can fully exploit both its 352.8/384KHz and DSD capabilities.
It comes with drivers for Windows and doesn't need any for Macs.

This is how it looks like:



The manufacturer's site:

amanero.com/

[Will]

Hi Javier (or is that Jackal ? )

A nice spot that, and a very interesting bit of kit.

Hopefully nobody will put the kibosh, like what happened with a similar device a while back.

[Javier]

yep, that'll be yours truly.

If finally it delivers what the specs promise it'll be one of the biggest audio bargains available.

[XTRProf]

I assumed this can only support stereo only, right? MC?

So far I saw Javier, Will and Davek from RG had ordered one at DIYAudio. I think I will join as I wanted one too that can do 384-32.

Thanks Javier for the link.

[Javier]

Chong, don't forget to add "exclude VAT" after Singapore if you don't want to be charged the Italian VAT (21%). As you are outside the EU you don't have to pay it.

[XTRProf]

Aug 30, 2012 21:45:18 GMT Javier said:

Chong, don't forget to add "exclude VAT" after Singapore if you don't want to be charged the Italian VAT (21%). As you are outside the EU you don't have to pay it.

Yes, I did by putting 0.00 at VAT. The 4th RG Musketeer there in the doc!

[cj]

Excellent heads up Javier

It looks rather nifty plus it's footprint will allow me to pop it straight into my work rig.

[Javier]

Next step is finding a nice DAC with a I2S/DSD input that doesn't cost as much as the BIII.
There are plenty of chips that could do the job with great specs: Cyrrus' CS4398, TI's DSD1794 or PCM1795, Asahi Kashei's AK4399, Wolfson's WM8740 and WM87411, etc. the problem is that most current cheap Chinese DAC's only come with SPDIF or, at most, with the TE7022 and none configured for DSD.

[XTRProf]

Aug 30, 2012 23:26:33 GMT Javier said:

the problem is that most current cheap Chinese DAC's only come with SPDIF or, at most, with the TE7022 and none configured for DSD.

Er, can mode, right?

[sandyk]

Aug 31, 2012 0:47:30 GMT XTRProf said:

Aug 30, 2012 23:26:33 GMT Javier said:

the problem is that most current cheap Chinese DAC's only come with SPDIF or, at most, with the TE7022 and none configured for DSD.

Er, can mode, right?

NO !!!

[sandyk]

Chong
IMO, it doesn't matter how many of these little gadgets, or genuine DSD capable DACs appear, the major Record Companies are highly unlikely in the forseeable future, to release or re-release their library in DSD format.They may sometime in the future, sniff a bit of profit by re-releasing stuff that many already have collected in various formats.DSD and SACD is likely to remain as it is, which is a niche product for classical and Jazz lovers, and a VERY small % of total sales.
Although, I really like a few releases from Cookie Marenco for example, there are very few of her releases that would interest me, and most likely most others. Don't pin your hopes on the majors introducing NEW material in DSD. It simply isn't likely to happen.
Alex

P.S.
However,while I do applaud the decision with Foobar to do something about SACD noise residual by providing different filter options, it seems to me that it is dumbing down the format.
I do however like the suggestion to move the noise spectrum to above 100kHZ. I doubt that it needs to go higher.

[FauDrei]

Aug 30, 2012 23:26:33 GMT Javier said:

...the problem is that most current cheap Chinese DAC's only come with SPDIF or, at most, with the TE7022 and none configured for DSD.

In a week or so I should be testing an entry level Chinese ES9018 with integrated asynchronous 384KHz/32bit USB to I2S interface.

Seems USB audio interfaces are slowly but surely moving forward. We'll see...

[XTRProf]

Aug 31, 2012 1:08:45 GMT sandyk said:

IMO, it doesn't matter how many of these little gadgets, or genuine DSD capable DACs appear, the major Record Companies are highly unlikely in the forseeable future, to release or re-release their library in DSD format.They may sometime in the future, sniff a bit of profit by re-releasing stuff that many already have collected in various formats.DSD and SACD is likely to remain as it is, which is a niche product for classical and Jazz lovers, and a VERY small % of total sales.
Although, I really like a few releases from Cookie Marenco for example, there are very few of her releases that would interest me, and most likely most others. Don't pin your hopes on the majors introducing NEW material in DSD. It simply isn't likely to happen.
Alex

P.S.
However,while I do applaud the decision with Foobar to do something about SACD noise residual by providing different filter options, it seems to me that it is dumbing down the format.
I do however like the suggestion to move the noise spectrum to above 100kHZ. I doubt that it needs to go higher.

Er, there are currently many SACDs that can be ripped already in the market than we have time to listen for. With MFSL, DG, East Coast, etc as well as many Japanese SHMSACD reissues, there will be a constant pool of DSD to listen to provided someone ripped and put it on the Internet.

In fact, original DSD format is not as good as PCM at 192-24. DSD should have the original format spec at DSD128 when they first started. Now people know and find that if they resample to 128 or higher, the noise will be pass 100Khz. Yup, I agree that 100khz should be enough as too high we will go into much RF noise. 100khz is already 5 harmonics what we human can theoretically hear at 20khz and should not create any more square wave issues.

[solderdude]

In a week or so I should be testing an entry level Chinese ES9018 with integrated asynchronous 384KHz/32bit USB to I2S interface.Seems USB audio interfaces are slowly but surely moving forward. We'll see...

The ENOB of ESS DAC is around 19 bits, perhaps with the most optimal PCB design it may even reach 20 bits.
This is true for most DAC's by the way.
You can view the 32 bits 'handling' is more in the line of it can 'accept' the format.

I am convinced there will never be a DAC reaching an ENOB of 24bit.
a matter of physics I am afraid.

[sandyk]

there will be a constant pool of DSD to listen to provided someone ripped and put it on the Internet.

Chong
Yes, ripped using PS3s and various types of software. Many are crappy in comparison with the SACD that they came from, just like most of those LPs that people have converted to 24/192 and uploaded.
Most people want newly released material, not regurgitated stuff from the 70s and 80s or earlier.

Alex

[Javier]

@alex, it would be great if you could come up with something positive instead of keeping "rainin on everybobody else's parade". This is exactly what I complained about last time. I know you find it very difficult to accepts other people's opinions or tastes but I kindly suggest you make an effort. It would have been so much better and constructive if instead of what you posted you suggested an external high quality 3.3V PS to power the board in discussion or something in that line.

I'm not "evangelizing" DSD or practising any type of DSD proselitism. I think it is interesting for many reasons and want it, that is all.

I happen to like "regurgitated" stuff from the 60s, 70s and even 80s and I think many here do too. In fact I like it much better than the new stuff most of the auidophile labels tend to release no matter how well recorded it may be.

As to rips being crappy, that is your personal opinion and may be some other's too but is far from being a universally accepted truth. As I'm not going to battle you or discuss how many people support you findings, I'd like to keep this thread on topic please, it'll save a lot of posts erasing laterand will keep everybody in a better mood.

Some Cookie Marenco:

www.youtube.com/watch?v=HFRG6IMLRV4

========================================

For me the potential "beauty" of DSD is not in reaching higher DNR (bit depth) or higher Freqs (sampling rate), it is its "purity" in the current world of Sigma Delta ADCs and DACs:



No decimations so no potentially "dammaging" filters needed (IIR, FIR, polywhatever, apodizing, etc.) which always imply a compromise between eliminating aliases and ringing, no oversampling too, just a simple analogue filter does the job.




FauDrei: The Audio-GD has terrific specs and a very interesting price but the TE8802 they have chosen as USB receiver is PCM only which to me is a big limitation. If you have PCM material only, then it is indeed a great VFM DAC. I've read that they sometimes have build quality issues, has this been improved lately?

@chong: You could remove the TE7022 from the board and solder the I2S wires comming from the new USB device to its now empty I2S pads but I guess is a complicated mod and requieres both skil and knowledge. It'll be easier to find a board made for this.

[cj]

For my usage, at the moment, the DSD side is less important as I do not have any such recordings at the mo.

My short term interest is the hope that it should be noticeably better than my existing USB to I2S board, especially with a nice JLH supply waiting for it

Then, it is more future proof compared to what I have now too.

[sandyk]

@alex, it would be great if you could come up with something positive instead of keeping "rainin on everybobody else's parade". This is exactly what I complained about last time. I know you find it very difficult to accepts other people's opinions or tastes but I kindly suggest you make an effort.

Javier
Chong and many othere apparently do not have much idea of where this device can be used. Looking at the DIYAudio thread thread there also appears to be a few unanswered questions, such as it an Asynchronous USB device, and will there be Asio software drivers for Windows 7.(I could have missed those answers though.)
It would appear that Chong may have ordered one, when he will not be able to use it unless he orders a Twisted Pear Buffalo 3. That would be beyond the capabilities of many members to get working to a very high standard without a considerable additional investment in the power supply area.Also, I have yet to see a subjective report on how it performs, or the extra power supplies needed, in addition to those already used in Will's PK or elsewhere. The who;e power supply issue could become very complex and require several boxes unless there is a great deal of planning and guidance given.
I believe that it was a fair comment about many SACD rips being of a much lesser quality than the SACD they were derived from.Many appear to have a lot of ultrasonic noise due to all the additional processing.
I can understand though why people like yourself and Will are interested in this design.Perhaps you could have given a rundown on what it does, and the other equipment and software needed at the start of the thread so that people would know if it could be worthwhile for them ?
Alex

[sandyk]

My short term interest is the hope that it should be noticeably better than my existing USB to I2S board, especially with a nice JLH supply waiting for it

Chris
If you are going to marry it up with Will's PK. where several separate high quality 3.3V supplies , such as via a Salas, JLH or PH,would normally be in use, then it would be highly deirable to supply anything preceding it with the best possible power too, such as from a SOtM USB 3.0 PCI-E card, or a suitable DIY solution, NOT normal USB power, or if it doesn't need +5V USB power, then a means of disabling the incoming supply.
However, with a project like this which is fairly inexpensive, you could also risk missing the boat if you do not jump straight in.
Alex

[sandyk]

Javier
I will be shutting down for the night shortly, so feel free to move what you consider to be my "unhelpful" comments to Saucerful for now.
Alex

[solderdude]

I agree with Javier on the format issue (will stay away from other issues),

Although I have nothing to contribute about the USB device I would like to comment on the nice plot of the impulse response which is somewhat misleading.
Why ? I would need to explain those drawn pulse responses.

Firstly the ‘analog’ input of the 3uS pulse with (-6dB) could only come from a tone generator (one that can at least also generate a 1MHz sine-wave).
It is a signal that cannot occur in any music signal nor can it come from an analog source.
A vinyl cartridge for instance cannot handle those rise times (the needle would need to accelerate to speeds that are impossible) and would swing out on the top and bottom (like an analog filter) considerably.
An analog tape recorder would also not be capable of reproducing it either b.t.w.
The analog source would need to have a bandwidth of at least 500kHz to reproduce such a signal.
A 3us pulse is just half of a 6us square-wave where the negative part is not reproduced.
6us = 166,666 Hz (166.7 kHz) so the pulse seen in the graph would be so high that it would even be inaudible even by certain members.
I wonder if there are even commercially available transducers and microphones that would be capable of recording/reproducing (remember a 500kHz minimum bandwidth would be needed)
Even the digital filter swinging would be way above that what even die hard bandwidth lovers would ever be capable of hearing.

So however nice the plots may be … the practical value of this theoretical signal is well uhm… theoretical.

Of-course it doesn’t invalidate the discussion/arguments and the nice converter.
Just thought to put the pulse response in perspective and be informative about it.

[Javier]

Thanks for your effort Alex, I really appreciate it. That was much better and far more constructive. No need to move anything.

No ASIO drivers are announced as of yet, but the device works with WASAPI which is as good and supported by most players, even "audiophile" ones like XXHighEnd, JPlay or HQPlayer by Signalyst, unfortunately not by cPlay which seems to be abandoned (no changes since Nov 2010).

No mention of working mode but it'll really surprise me if the manufacturer decided to use bulk mode instead of asynchronous, more so since he is using quite nice clocks on the board.

Some characteristics of the Amanero device mentioned in the Hifiduino blog (keep in mind this can be purchased for less than 60€ all included):

- Clean, very minimalistic design with no other feature than an I2S output and ready to be “plugged” into any DAC.
- The local regulators are the now-popular ultra low noise ADP-150 and ADP-151 from Analog Devices.
- The clock is taken from the two external local oscillators (and not generated by the FPGA) therefore keeping added jitter to the absolute minimum.
- The external oscillators are powered by the ADP-151 regulators through an LC filter which further decreases power noise (not an exotic implementation, but good practice).
- Supports up to 384 kHz and 352.8 kHz sample rates.
- Supports DSD


If there is any real interest, it is very easy to find how this thing is powered and what it does with USB's 5V supply, a USB schematic is provided by the designer to help modders:

amanero.com/USB_B.pdf

No reports of how it works as no units have been delivered yet but some are already on their way to early buyers so hopefully we'll have that information shortly. The GB is still in "pre" state so time enough to drop out if reports are not good, still, considered the price the loss won't be huge.

This device can be used with any DAC that accepts I2S which only leaves out the oldest R2R chips as all current production chips accept it perfectly, I'd say it is in fact the industry standard, so no need to get a BIII to enjoy, even though Chong has said more than once that he intends to in the future. Connecting it to an existing DAC can indeed be tricky as tapping into tiny pads will be necessary, some in this forum have soldered smd resistors between IC's legs (check Will's PK thread) and could do this kind of job. Where to tap can easily be found in the DAC chip spec sheet or one could ask in DIYform. If this story has a happy ending and we buy and receive the adapters a new thread on their implementation would be very interesting.

What I2S is and how it works goes way beyond this thread (en.wikipedia.org/wiki/I2s) but let’s say that most standalone CD/DVD or BD player’s laser heads output the signal in this format to the DAC circuit. In order to get this I2S signal - which was designed for very short runs (max. 10cm) - out of the player and sent to an external device both SPDIF and Toslink were designed, but that was quite a while back when technology had limitations it doesn't have today and costs for something better were too high. Both these systems take the 3 or 4 I2S signals, combine them into one (multiplexing) and send them to the receiver at the other side which has to disassemble the mix and try to recover the original signals (demultiplexing), which is an unneeded step now thanks to this USB->I2S adapters and avoid the mux-demux process with its clock guesswork (KISS?).

Chris is already feeding his PK with a USB->I2S device, though a less performing one based on the Tenor TE7022 IC which is limited to 44.1, 48 and 96KHz (@16 & 24bit) but with the advantage of being UAC1 so no drivers needed (no ASIO support either). The Amanero board is UAC2 compliant allowing for much higher sampling rates, bit depths and even DSD. As Windows doesn't have UAC2 internal drivers (Linux and MAC do) a 3rd party driver is needed. Shaun also has a USB->I2S device (WaveIO) connected to his BIII. I don't recall them complaining on finding it difficult to install and get them working though I may be wrong.

As to SACD rips, I’m not sure if you mean files captured from the analogue output of SACD players, digital downloads from the likes of Cookie’s site or digital dumps (ISO or DSF/DDF files) from a PS3. The two latter involve no processing, the PS3 dumps are a digital copy just like copying a file from one HDD to another and their content is true untouched DSD. Analogue rips do indeed include processing and in fact they are not DSD files, they can be indeed compared to vinyl rips, they are an “impression” of the original, not a clone.

[sandyk]

Aug 31, 2012 12:37:24 GMT solderdude said:

I agree with Javier on the format issue (will stay away from other issues),

Although I have nothing to contribute about the USB device I would like to comment on the nice plot of the impulse response which is somewhat misleading.
Why ? I would need to explain those drawn pulse responses.

Firstly the ‘analog’ input of the 3uS pulse with (-6dB) could only come from a tone generator (one that can at least also generate a 1MHz sine-wave).
It is a signal that cannot occur in any music signal nor can it come from an analog source.
A vinyl cartridge for instance cannot handle those rise times (the needle would need to accelerate to speeds that are impossible) and would swing out on the top and bottom (like an analog filter) considerably.
An analog tape recorder would also not be capable of reproducing it either b.t.w.
The analog source would need to have a bandwidth of at least 500kHz to reproduce such a signal.
A 3us pulse is just half of a 6us square-wave where the negative part is not reproduced.
6us = 166,666 Hz (166.7 kHz) so the pulse seen in the graph would be so high that it would even be inaudible even by certain members.
I wonder if there are even commercially available transducers and microphones that would be capable of recording/reproducing (remember a 500kHz minimum bandwidth would be needed)
Even the digital filter swinging would be way above that what even die hard bandwidth lovers would ever be capable of hearing.

So however nice the plots may be … the practical value of this theoretical signal is well uhm… theoretical.

Of-course it doesn’t invalidate the discussion/arguments and the nice converter.
Just thought to put the pulse response in perspective and be informative about it.

At the risk of sidetracking the thread again, I will say this.
Originally Sony had a nice range of "Super Tweeters"with response to 70kHZ , specially made to go with SACD. However, the ultrasonic rtubbish from SACD caused quite a few reports of amplfiers overheating and destructing when playing at high levels. Even Martin Colloms had this happen to him. The result was further filtering in most SACD/DVD-A capable players to help prevent this. Due to being implemented in the analogue area of the player, the result was then often a compromised DVD-A perfornance as well.Even when using wideband amplification, assuming that recent developments see the artifacts pushed to over 100kHZ, the sad fact is that many loudspeakers roll off quite quickly above 20kHZ. Mine start to roll off markedly around 19kHZ. Many headphones also rolloff rather quickly in this area, although a few have a usable response to 40kHZ. Also, recent implementations to Foobar are achknowledging this out of band rubbish problem by giving different filter options. Unless this out of band rubbish is moved to past 100kHZ, I would agree with Chong that properly implemented 24/192 has greater audio quality potential, assuming also that SACD recordings were DSD all the way. Many are not.
Alex

[cj]

Aug 31, 2012 12:03:32 GMT sandyk said:

My short term interest is the hope that it should be noticeably better than my existing USB to I2S board, especially with a nice JLH supply waiting for it

Chris
If you are going to marry it up with Will's PK. where several separate high quality 3.3V supplies , such as via a Salas, JLH or PH,would normally be in use, then it would be highly deirable to supply anything preceding it with the best possible power too, such as from a SOtM USB 3.0 PCI-E card, or a suitable DIY solution, NOT normal USB power, or if it doesn't need +5V USB power, then a means of disabling the incoming supply.
However, with a project like this which is fairly inexpensive, you could also risk missing the boat if you do not jump straight in.
Alex

Exactly Alex.

As I intimated previously a 5v JLH cutting in at the USB socket end of the board would be the first port of call.
It would also seem that at least two 3.3v (JLH, Salas or such like) supplies should also be straight forward on this board

Attachments:

[Javier]

Sure, but whether you like it or not your Oppo, SC DAC and X-DAC all convert to DSD as the last step before analogue. They all use Sigma Delta chips.

EDIT:

IIRC neither of your DACs supports 24/192 so why should you care?