How to make a streamer (Pro-Ject Stream Box S2 Ultra) playing Qobuz & Tidal to sound better than CD

[raysablade]

(as the majority upgrades from a standard Pi appear to be focused on making the USB digital output sound decent)

Whilst I would never want to comment directly on the MDAC thread(s). I'm struck by the volume of time and effort devoted by John Westlake to his battle with USB. Why bother, when direct I2S connection is available and essentially free?

 

[vinyl_paul]

Interesting thread. I do love this stuff.

The bit(!) that I don't understand in all of this, is what actually happens in the DAC. Digital Analogue Converter. The DIGITAL (10101) information is converted to analogue. As long as the packets arrive at the DAC in the right order and in a reliable fashion, what else matters for the DAC? Doesn't the DAC have some buffering to ensure consistent digital feed?

Why do I care about the USB cable? Let alone the ethernet?
Anyone tried this with Token Ring? I've heard its brilliant.

Full disclosure: I am running a RPi with Volumio, outputting on USB to a Chord Hugo. I think the Hugo is doing all the clever stuff and the RPi is supplying digital packets on USB. But I am just a computer networking guy - started in the industry early 1990s when you had to tell people what the Internet was, and you could impress friends with talk of TCP/IP.

 

[AndyU]

Interesting thread. I do love this stuff.

The bit(!) that I don't understand in all of this, is what actually happens in the DAC. Digital Analogue Converter. The DIGITAL (10101) information is converted to analogue. As long as the packets arrive at the DAC in the right order and in a reliable fashion, what else matters for the DAC? Doesn't the DAC have some buffering to ensure consistent digital feed?

Why do I care about the USB cable? Let alone the ethernet?
Anyone tried this with Token Ring? I've heard its brilliant.

Full disclosure: I am running a RPi with Volumio, outputting on USB to a Chord Hugo. I think the Hugo is doing all the clever stuff and the RPi is supplying digital packets on USB. But I am just a computer networking guy - started in the industry early 1990s when you had to tell people what the Internet was, and you could impress friends with talk of TCP/IP.

What else might matter to the DAC is all the electrical noise coming from the source and affecting the DAC. Your DAC was designed by Rob Watts, and if you read his posts on Head-Fi you will find he gives reasons how noise via USB can affect the ground plane of the Hugo and one of the reasons he gives for the superiority of the Hugo2 over the Hugo is that it is galvanically isolated, thus minimising the effects of whatever noise there might be. Nonetheless he advocates running your laptop on batteries to remove any possible earth loops for the last ounce of performance even with galvanically isolated DACs.

If you think about it another way, laptops and PCs produce horrific amounts of noise - just hold a portable radio near one - so much noise that there are laws about limiting it, nobody designing a pc gives a toss about such noise because they, mostly quite rightly, think it’s all 0’s and 1’s, and then you connect a laptop to a DAC which is shooting for a dynamic range of 120dB or better hoping for no effect. Unlikely.

It is of course completely reasonable to say, as someone soon will, that Rob Watts is an incompetent designer and his DACs are crap. Here’s a few of his posts ..

https://www.head-fi.org/threads/chord-hugo.702787/page-578#post-10994758
https://www.head-fi.org/threads/chord-hugo.702787/page-578#post-10994758
https://www.head-fi.org/threads/cho...official-thread.869417/page-222#post-14839706
https://www.head-fi.org/threads/chord-hugo.702787/page-560#post-10924471
https://www.head-fi.org/threads/cho...official-thread.831345/page-934#post-14445774

 

[John Phillips]

The bit(!) that I don't understand in all of this, is what actually happens in the DAC. Digital Analogue Converter. The DIGITAL (10101) information is converted to analogue. As long as the packets arrive at the DAC in the right order and in a reliable fashion, what else matters for the DAC? Doesn't the DAC have some buffering to ensure consistent digital feed?

I think a wise man would not want to enquire too closely about how DACs work . I had to do so professionally for a short period in the late 1980s. It was complicated even then and more so now in ways I am sure I no longer understand.

The mathematics of making a perfect DAC have been well known for many decades. But the engineering needed to approach as close to perfection as can be achieved today gets complex. It's very much like the well known quote: "Laws are like sausages. Better not to see them being made".

And then there's the unpredictability of human preference which may not coincide with mathematical perfection anyway.

Full disclosure: I am running a RPi with Volumio, outputting on USB to a Chord Hugo. I think the Hugo is doing all the clever stuff and the RPi is supplying digital packets on USB. But I am just a computer networking guy - started in the industry early 1990s when you had to tell people what the Internet was, and you could impress friends with talk of TCP/IP.

You have some very good sausages there. I hope you continue to enjoy them. It might be wise to not learn reasons to dislike what you currently like.

 

[monya]

Seems Westlake designed this streamer too....

https://stackaudio.co.uk/link/

 

[vinyl_paul]

@AndyU I get where you are coming from, and my problem is I don't understand the technology down at that level. My understanding is all higher up the stack!

For instance, suppose I have a digital music file on my NAS. I can generate a cryptographic fingerprint of that file (SHA256 anyone?). I can then move that file anywhere I like, using whatever, to another place and I can prove its the exact same file. Regardless of networks, power, noise etc etc. Every single bit will be the same.
In my simple mind I don't see why I can't get the file from the NAS (or streamer or RPi) into the DAC in the exactly the same way, and then let the DAC process it.

 

[AndyU]

@AndyU I get where you are coming from, and my problem is I don't understand the technology down at that level. My understanding is all higher up the stack!

For instance, suppose I have a digital music file on my NAS. I can generate a cryptographic fingerprint of that file (SHA256 anyone?). I can then move that file anywhere I like, using whatever, to another place and I can prove its the exact same file. Regardless of networks, power, noise etc etc. Every single bit will be the same.
In my simple mind I don't see why I can't get the file from the NAS (or streamer or RPi) into the DAC in the exactly the same way, and then let the DAC process it.

Ask Rob Watts that question on HeadFi, or read the links I gave you, or read his other posts. Compare the s/pdif input of your DAC with the USB input. DACs don't take a whole file and process it internally; that would be far too sensible. (Though streamers like the venerable squeezebox Touch have large buffers that can hold many seconds of data.) USB transmits data in 13 byte or so packets I think, and is electrical. A DAC is also electrical. No good reason to suppose a DAC is necessarily uninfluenced by the noise from the source. Galvanic isolation is a good start, and the need for it seems to be well understood in other fields like medical instrumentation. http://www.engineersjournal.ie/2015/06/16/isolating-medical-devices/

 

[adamdea]

@AndyU I get where you are coming from, and my problem is I don't understand the technology down at that level. My understanding is all higher up the stack!

For instance, suppose I have a digital music file on my NAS. I can generate a cryptographic fingerprint of that file (SHA256 anyone?). I can then move that file anywhere I like, using whatever, to another place and I can prove its the exact same file. Regardless of networks, power, noise etc etc. Every single bit will be the same.
In my simple mind I don't see why I can't get the file from the NAS (or streamer or RPi) into the DAC in the exactly the same way, and then let the DAC process it.

You are right of course. And provided that the dac has been competently designed, it won't make any difference to its ouput how the bits got there to within a degree of tolerence which exceeds anything you are likely to be able to hear, although in some cases you can *measure* super tiny inaudible differences
But there are four reasons why most folk involved in hifi don;t want to know
1) as a matter of fact you can hear all kinds of different things for all kinds of reasons unconnected with the sound waves reaching your ears
2) if your hobby is buying new boxes and marvelling at how cool they are, the truth is a bit deflating and is inconsistent with 1) if you assumethat all fluctuations in your listenign experience are caused by detection of tiny differences in the sound field.
3) manufacturers have to keep selling new stuff.
4) whatever anyone says, it is of course possible that inaccurate reproduction sounds better than accurate reproduction

At least since the benchmark dac1 (if not before) it has been possible to buy a dac with essentially com;lete jitter rejection on a toslink input. At that point you are basically getting the file reproduced as accurately as you can hear. Usb was a massive step backwards in galvanic isolation in the name of improving jitter rejection (which had already been solved). None of which made any difference if the dac was designed properly anyway. Designed properly used to require a fairly expensive product. Now you can get one for about £70

The good news is that if you want to keep buying new kit, someone will sell it to you. And they can serve it up with sciencey sounding stuff if you like too. Rob Watts makes good dacs. But he has to keep making new ones to make a living. For decades he has been on a trail of making dacs with filters which get longer with every iteration. They can go on getting longer forever. Each time the output will in a sense be getting more accruate. It's like subscribing to a book with all the digits of pi written out and a new supplement every month. The maths is real, the interest limited.

 

[JensenHealey]

I am sure all these Dacs deal with all the 'bits' perfectly. But to get the 120dB Dynamic Range you need very good design in all departments inside the box - even if all the 'bits' are accurate. Hence stuff around ground plane noise, multiple voltage regulators for different circuit sections and so on.

Now, of course, it remains the case that no music source available within the home needs, or can use 120dB of dynamic range. So some of this stuff is the equivalent of putting just another last layer of lacquer on an already well polished article.

 

[AndyU]

@AndyU I get where you are coming from, and my problem is I don't understand the technology down at that level. My understanding is all higher up the stack!

For instance, suppose I have a digital music file on my NAS. I can generate a cryptographic fingerprint of that file (SHA256 anyone?). I can then move that file anywhere I like, using whatever, to another place and I can prove its the exact same file. Regardless of networks, power, noise etc etc. Every single bit will be the same.
In my simple mind I don't see why I can't get the file from the NAS (or streamer or RPi) into the DAC in the exactly the same way, and then let the DAC process it.

and here is an explanation by John Westlake of why your Raspberry Pi might not be as perfect as you hope it is ..

https://www.avforums.com/threads/wh...ve-700euro-for-an-r-pi.2175090/#post-26431682

 

[John Phillips]

@AndyUFor instance, suppose I have a digital music file on my NAS. I can generate a cryptographic fingerprint of that file (SHA256 anyone?). I can then move that file anywhere I like, using whatever, to another place and I can prove its the exact same file. Regardless of networks, power, noise etc etc. Every single bit will be the same.
In my simple mind I don't see why I can't get the file from the NAS (or streamer or RPi) into the DAC in the exactly the same way, and then let the DAC process it.

I think you use a USB interface to your DAC. Assuming this uses the modern "asynchronous" transfer (I am sure Chord does) the USB interface requests data from the source and fills a buffer with it. The DAC converts data from the USB buffer; and the USB interface makes further requests for data as its buffer approaches being empty.

In your case it's a buffered stream rather than a file transfer, so a file-wide fingerprint to check data integrity may not be useful. But USB transport over short distances is sufficiently reliable in practice as long as the cable and interfaces meet the basic USB specifications. Listen long enough and there will eventually be a bit error; and the DAC will probably generate a click of some kind. But I have never noticed this in practice from my USB-interface DAC.

For realistic purposes you can assume every bit will be the same.

 

[AndyU]

You are right of course. And provided that the dac has been competently designed, it won't make any difference to its ouput how the bits got there to within a degree of tolerence which exceeds anything you are likely to be able to hear, although in some cases you can *measure* super tiny inaudible differences
But there are four reasons why most folk involved in hifi don;t want to know
1) as a matter of fact you can hear all kinds of different things for all kinds of reasons unconnected with the sound waves reaching your ears
2) if your hobby is buying new boxes and marvelling at how cool they are, the truth is a bit deflating and is inconsistent with 1) if you assumethat all fluctuations in your listenign experience are caused by detection of tiny differences in the sound field.
3) manufacturers have to keep selling new stuff.
4) whatever anyone says, it is of course possible that inaccurate reproduction sounds better than accurate reproduction

At least since the benchmark dac1 (if not before) it has been possible to buy a dac with essentially com;lete jitter rejection on a toslink input. At that point you are basically getting the file reproduced as accurately as you can hear. Usb was a massive step backwards in galvanic isolation in the name of improving jitter rejection (which had already been solved). None of which made any difference if the dac was designed properly anyway. Designed properly used to require a fairly expensive product. Now you can get one for about £70

The good news is that if you want to keep buying new kit, someone will sell it to you. And they can serve it up with sciencey sounding stuff if you like too. Rob Watts makes good dacs. But he has to keep making new ones to make a living. For decades he has been on a trail of making dacs with filters which get longer with every iteration. They can go on getting longer forever. Each time the output will in a sense be getting more accruate. It's like subscribing to a book with all the digits of pi written out and a new supplement every month. The maths is real, the interest limited.

Which £70 DAC have you got?

 

[adamdea]

Which £70 DAC have you got?

I've been buying dacs for a long time. The price of well engineered ones seems to keep dropping. As it is the most expensive one I have is a benchmark dac2 hgc which cost me £1200 (presumably because the 3 had just come out). I only use it in my office for listening to headphones becasue I use a dsp antimode in my main system. I do have a tiny cambridge audio dongle which cost £40 and a cirrus audio card for pi which was fairly similar.

If I were buying one now, I would be very tempted to get one of these when it comes out (estimated $70)
https://www.audiosciencereview.com/...ents-of-e1da-9038s-bal-portable-dac-amp.8424/
Not sure whether it has been tested with a crap computer input though- so not clear how great the isolation is. I''m honestly not sure though whether anything more than a topping d10 is required.
Pushing the boat out to £370 on amazon
https://www.audiosciencereview.com/...ew-and-measurements-of-sabaj-d5-dac-amp.8337/

It is just amazing how many fanatastic dacs seem to be coming out for peanuts. I think this is attributable to the fact that the 9038 chip and the equivalent AKM model are now fairly well understood and it seems that ordinary engineers can get the most out of them. The bar is now set really high for a high end dac manufacturer to be able to get better performance.

 

[AndyU]

I've been buying dacs for a long time. The price of well engineered ones seems to keep dropping. As it is the most expensive one I have is a benchmark dac2 hgc which cost me £1200 (presumably because the 3 had just come out). I only use it in my office for listening to headphones becasue I use a dsp antimode in my main system. I do have a tiny cambridge audio dongle which cost £40 and a cirrus audio card for pi which was fairly similar.

If I were buying one now, I would be very tempted to get one of these when it comes out (estimated $70)
https://www.audiosciencereview.com/...ents-of-e1da-9038s-bal-portable-dac-amp.8424/
Not sure whether it has been tested with a crap computer input though- so not clear how great the isolation is. I''m honestly not sure though whether anything more than a topping d10 is required.
Pushing the boat out to £370 on amazon
https://www.audiosciencereview.com/...ew-and-measurements-of-sabaj-d5-dac-amp.8337/

It is just amazing how many fanatastic dacs seem to be coming out for peanuts. I think this is attributable to the fact that the 9038 chip and the equivalent AKM model are now fairly well understood and it seems that ordinary engineers can get the most out of them. The bar is now set really high for a high end dac manufacturer to be able to get better performance.

Since your DAC2 is worth around £800 why don't you sell it, buy a £70 Topping and trouser £700?

 

[adamdea]

Since your DAC2 is worth around £800 why don't you sell it, buy a £70 Topping and trouser £700?

Why not indeed? Probably for the same reason I have 4 pairs of headphones and a sugden amp I never use etc. I don't even need to buy the topping; I could just sell the benchmark and use one of any number of other dacs I have (perhaps the focusrite forte). The benchmark is useful in having analogue inputs though; if I ever go over to doing room correction on a pc I might swap it in place of the drc antimode in the main rig. I tend to buy things out of curiosity and I only sell them when I'm in a clearing out mode. Last time that happened was 5 years ago when I moved house.

 

[MasterTape]

@AndyU I get where you are coming from, and my problem is I don't understand the technology down at that level. My understanding is all higher up the stack!

I think you owe it to yourself to try a DigiOne SPDIF HAT now that you're already so close in terms of setup. It's £90 and you stick it onto your Pi and can still run Volumio (not sure if some re-install is needed). Then you get to try how a source with a fair bit of galvanic isolation and a near reference jitter spec sounds compared to what you're used to. I think it's a fabulous upgrade to anything USB that I've fed the Hugo with. It had a pretty crappy USB implementation compared to the later Chords, and given how Mr Watts' tune changed from "the noise issue is inherent to USB" in the Hugo era to "USB sounds the same as SPDIF" with the later galvanically isolated implementations, I think he'd agree with that assessment.

 

[Timo_F]

I think you owe it to yourself to try a DigiOne SPDIF HAT now that you're already so close in terms of setup. It's £90 and you stick it onto your Pi and can still run Volumio (not sure if some re-install is needed). Then you get to try how a source with a fair bit of galvanic isolation and a near reference jitter spec sounds compared to what you're used to. I think it's a fabulous upgrade to anything USB that I've fed the Hugo with. It had a pretty crappy USB implementation compared to the later Chords, and given how Mr Watts' tune changed from "the noise issue is inherent to USB" in the Hugo era to "USB sounds the same as SPDIF" with the later galvanically isolated implementations, I think he'd agree with that assessment.

Do you include Chord Hugo 2 in “later Chords”? I have a Hugo 2, and wonder about the difference in the Hugo’s USB and SPDIF implementation... In your experience or to your knowledge, is there much between the two?

Cheers!

 

[mjw]

A question for those in the know about the Cisco switch and its ilk - does an AirPort Extreme incorporate such a device? I see a row of Ethernet sockets on the rear panel and it got me wondering. (Before anyone suggests, I have no devices to connect and report back. Mine is an academic question).

 

[lordsummit]

I think you owe it to yourself to try a DigiOne SPDIF HAT now that you're already so close in terms of setup. It's £90 and you stick it onto your Pi and can still run Volumio (not sure if some re-install is needed). Then you get to try how a source with a fair bit of galvanic isolation and a near reference jitter spec sounds compared to what you're used to. I think it's a fabulous upgrade to anything USB that I've fed the Hugo with. It had a pretty crappy USB implementation compared to the later Chords, and given how Mr Watts' tune changed from "the noise issue is inherent to USB" in the Hugo era to "USB sounds the same as SPDIF" with the later galvanically isolated implementations, I think he'd agree with that assessment.

I’m very interested in this idea, I use a pi and usb out into a Mojo, would adding a hat and using optical out make a significant difference?

 

[webster]

I’m very interested in this idea, I use a pi and usb out into a Mojo, would adding a hat and using optical out make a significant difference?

I think that the Digione is Coax/BNC output only.