USB to USB Streamer ¦ Replacing a computer in an audio system

[Fleuch]

The reasons for providing an “air gap” in an audio system between a computer and a DAC are many, particularly the noise on either the USB or Ethernet outputs. A laptop, tablet or smartphone provide options to control audio standard devices such as network players, network streamers, network bridges and other devices.

Many of the DACs listed on ASR have USB inputs and may be designed with direct connection to a computer as part of the brief.

Few network streamers are available to replace a computer that have a USB input for a hard disk drive, as the music source, and a USB output to the DAC.

This is important, as changing or upgrading the DAC is less expensive than changing a network player. From the specifications of most network players, the actual DAC chips often have lower specifications, or are an earlier release by the chip manufacturer, compared to stand-alone DACs. For example, Audiolab have used the ESS ES9018K2M chip in its latest range of audio products. Agreed, and for the avoidance of doubt, the performance of a DAC depends on much more than the DAC chip itself but it helps if mainstream manufacturers incorporate new and improved products in their designs, such as ESS ES9038Q2M or PRO. It gives the purchaser confidence that the manufacturer has a policy of continuous development..

There is not a wide selection of network streamers available that have a USB input and a USB output to connect to a USB DAC available at reasonable cost, where software, and the hardware required to control the streamer, further restricts choice.

A very incomplete survey suggests the following are some of the options currently available,

Volumio: Primo and Mini. From the information available, the devices have the Volumio software embedded and controlled through a web browser running on another device.

Pro-Ject: Head Box S2 Digital; Stream Box S2; Stream Box S2 Ultra. Only the Stream Box S2 Ultra has a USB output.

Auralic: Aries Mini: no longer available from Auralic but can still be found. The problem is the app only runs under iOS. Retailers have suggested a web browser can be used to control the Mini.

SOtM : the basic design is the SMS 200 Neo, with a range of “improved” options and upgrade plug-in modules available. An earlier version of the SMS200 has been reviewed on ASR (Many thanks for the hard work, Amir)

Stack Audio : The Link. Reviewers have noted the similarity of the design, influenced by John Westlake, to the Stream Box S2 Ultra. Also reviewers have noted an improvement in sound quality due to additional filtering and cooler operation.

Going upmarket

Lindemann : Limetree Bridge and Limetree Network. Reviewers have reported excellent sound quality when using both devices, but note the cost of either is high. Both devices use a Lindemann app.

Leema : Elements Streamer. There is no USB output and the cost is excessive.

Most of the designs appear to incorporate a SBC, in some cases a RaspberryPi Compute Module.

The only device tested by ASR is an earlier version of the SOtM SMS 200. There may be similar products available from SMSL, Topping and other manufacturers but are not easy to find.

Streaming is not important to me now but would be a welcome as a form of future proofing.

Discussion on the use of these devices and advice on how to obtain the best value for money would be most welcome.

As a post script, the Topping D90 DAC looks very attractive; again thanks to Amir for a review so soon after product release.

If you only do what you have always done, you will only get what you have always had (attributed to Albert Einstein)

 

[renaudrenaud]

I do not understand totally. For my point of view the Volumio thing I've seen is a card with a DAC and a Raspberry Pi inside. A Rasp is a computer, a Single Board Computer. The OS is Linux. Volumio adds software of their creation and a card "a la" Hifiberry. So, from my point of view, you have here a computer with a sound card.

just compare the RJ45 and the 4 USB ports. Does the Rasp looks like a "streamer" you have seen recently?

I do not know other products, but if you need a processor, some ram, an operating system and some apps, at the end you can call it a computer. Of course you can disguise it as an audio bla bla. But it is a computer.

Also, can you describe the noise from the USB you want avoid? Does you DAC not enough well engineered to avoid this kind of problem?

And noise from RJ45? You can solve it with the marvelous upton Regen as you can read here.


I know my solution is far as optimal but I solved this problematic with an Atomic Pi (30$) running Daphile (free, Linux distro) connected with USB cable to the DAC with correct engineering. There is no commercial bullshit in the system and the DAC is low cost and in the ASR blue quadrant. Cheaper then Volumio and more fun.

 

[Matias]

Also consider the Sonore Rendu series, starting with the microRendu.

 

[Fleuch]

Many thanks for the replies.

There will will always be EMI, or what might be called "pick up" on the USB output of a laptop or a PC, simply because of the many processes executed in the machine. A well-designed audio aware device will minimise the impact of the noise on the data stream.

Volumio is a Linux-based audiophile operating system similar to Daphile but the enthusiasts who have developed the software also have developed hardware streamers based on the software:

Volumio Mini 86: https://volumio.org/product/volumio-mini86/

Volumio Primo: https://volumio.org/product/volumio-primo/

The Stack Audio Link uses Volumio as its operating system:

Stack Audio Link: https://stackaudio.co.uk/link/

All of the streamers listed in the original post are "ready to run out of the box" and are "ready made" solutions to the problem of streaming from a USB hard drive to a USB DAC.

For me the simplicity of this approach is important, rather than the equally valid approach using Raspberry Pi boards. As you may guess, I could put together a DIY system based on the RPI and an appropriate operating system, or use a similar SBC such as the Asus Tinkerboard, but would prefer to pay the premium and purchase a finished product.

My other thought was to avoid if possible the other devices that might be connected in the signal path, such as reclockers and galvanic isolators, not to mention the possibility of linear power supplies for each device - again to keep things as simple as possible.

There are differencess in the various software music players, as demonstrated by another recent thread. In the same way there must be differences in the various hardware streamers. Rather than rely on retailers for information, I was hoping there might be those who read these pages who might be able to offer advice based on their "lived experience".

The opinion in the shop is that the sound quality from the Lindemann packages is superior. At the price it really should be, but it would be great to find something that is nearly as good but at less cost.

Many thanks for signposting to Sonore (https://www.sonore.us/microRendu.html). On first reading the it would seem the microRendu requires an Ethernet input and provides a USB-A output, but not sure about the software used to control it. As I do not want to use a NAS drive as a server there might be a problem.

 

[FrantzM]

To paraphrase Shakespeare:

A Computer by any other name...

Getting the digits from a file to a DAC requires some kind of computer; as for the USB "noise" read Amirm DAC reviews.. After 200+ DAC reviews , it is not a real problem.

 

[DDF]

To paraphrase Shakespeare:

A Computer by any other name...

Getting the digits from a file to a DAC requires some kind of computer; as for the USB "noise" read Amirm DAC reviews.. After 200+ DAC reviews , it is not a real problem.

It’s not a problem in Amir’s test system but the ground loop using usb to Dac can and for some of us does cause audible noise in the speakers from background processes on the computer. Be careful to not apply one test condition and set up as a ubiquitous behaviour

 

[Fleuch]

It is important not confuse sources of noise. Even though the modern world is training each one of us to think "digitally" the transport medium for the voltage levels representing the 0 and 1 states is, without a doubt analogue, where, as a simple example, the 0 state is represented by 0volts and the 1 state is represented by 5volts. Although the bit stream is said to be "digital" it is essentially an analogue electrical signal with two set levels, rather than the infinite number of levels in the analogue output from an amplifier.

The time required to change from one state to the other is referred to as the rise time and may be influenced by the electrical environment, such as the cables used. A good digital design engineer must also be a good analogue design engineer to understand how the various parts of a digital system can influence the shape and timing of a "digital" pulse, such as the rise time and ringing. Ringing will be one component of "jitter". Expensive external reclockers and master clocks are recommended to minimise jitter, as well as linear power supplies; SMPS devices are electrically "noisy" and can inject noise into the system, whether on the power line or by EM interference. Carefully designed filters are included in the better DAC circuits in an attempt to remove all forms of unwanted noise (for example https://www.project-audio.com/en/product/pre-box-s2-digital/).

A good illustration of a "real world" digital pulse is provided by mitchco in the posting at https://www.audiosciencereview.com/...link-by-stack-audio-and-signal-detoxing.7578/. Signal detoxing is really another way of talking about removing superfluous noise from the "digital" signal. The USB output port of any computer will effectively inject noise and timing uncertainty (aka jitter) into the output stream to a USB DAC. Timing. or clocking, and pulse shape is what makes or breaks a digital system, loose the precision synchronisation between the clock signal and the digital data stream has a real time impact on the quality of output from the DAC and the sound quality of the music system. Jitter can be thought of as a measure of uncertainty of where the digital signal is in the time domain and the actual voltage level of the data signal when the system clock signal is active, compared to where it should be (as in the original source). In simple terms any pick-up that will affect an analogue signal will also affect a digital signal, but in a different way.

All of the devices listed previously are used to control the music files streamed to a USB DAC and some, through good audio / analogue design and other strategies, are able to "detox" the signal and improve the synchronisation of the system clock with the digital data stream, thereby reducing jitter.

The SMSL SU-8 and various Topping DAC designs have balanced outputs which should remove much of the noise between the DAC and the amplifier, but still leaves the digital stream to the DAC input to be considered. It also requires an amplifier with balanced inputs.

All of my digital music files are held on external USB hard drives and the current DAC has a USB input, which explains why I would like to add a control device that at the same time will detox the data stream. Perhaps part of this explains why contributors have posted something along the lines of "yeah, but its not all about the DAC chip" ! This is not about the noise measured within a DAC under test but about reducing the noise and signal timing inaccuracy in the data stream input to the DAC.

 

[chasefrench]

It is important not confuse sources of noise. Even though the modern world is training each one of us to think "digitally" the transport medium for the voltage levels representing the 0 and 1 states is, without a doubt analogue, where, as a simple example, the 0 state is represented by 0volts and the 1 state is represented by 5volts. Although the bit stream is said to be "digital" it is essentially an analogue electrical signal with two set levels, rather than the infinite number of levels in the analogue output from an amplifier.

The time required to change from one state to the other is referred to as the rise time and may be influenced by the electrical environment, such as the cables used. A good digital design engineer must also be a good analogue design engineer to understand how the various parts of a digital system can influence the shape and timing of a "digital" pulse, such as the rise time and ringing. Ringing will be one component of "jitter". Expensive external reclockers and master clocks are recommended to minimise jitter, as well as linear power supplies; SMPS devices are electrically "noisy" and can inject noise into the system, whether on the power line or by EM interference. Carefully designed filters are included in the better DAC circuits in an attempt to remove all forms of unwanted noise (for example https://www.project-audio.com/en/product/pre-box-s2-digital/).

A good illustration of a "real world" digital pulse is provided by mitchco in the posting at https://www.audiosciencereview.com/...link-by-stack-audio-and-signal-detoxing.7578/. Signal detoxing is really another way of talking about removing superfluous noise from the "digital" signal. The USB output port of any computer will effectively inject noise and timing uncertainty (aka jitter) into the output stream to a USB DAC. Timing. or clocking, and pulse shape is what makes or breaks a digital system, loose the precision synchronisation between the clock signal and the digital data stream has a real time impact on the quality of output from the DAC and the sound quality of the music system. Jitter can be thought of as a measure of uncertainty of where the digital signal is in the time domain and the actual voltage level of the data signal when the system clock signal is active, compared to where it should be (as in the original source). In simple terms any pick-up that will affect an analogue signal will also affect a digital signal, but in a different way.

All of the devices listed previously are used to control the music files streamed to a USB DAC and some, through good audio / analogue design and other strategies, are able to "detox" the signal and improve the synchronisation of the system clock with the digital data stream, thereby reducing jitter.

The SMSL SU-8 and various Topping DAC designs have balanced outputs which should remove much of the noise between the DAC and the amplifier, but still leaves the digital stream to the DAC input to be considered. It also requires an amplifier with balanced inputs.

All of my digital music files are held on external USB hard drives and the current DAC has a USB input, which explains why I would like to add a control device that at the same time will detox the data stream. Perhaps part of this explains why contributors have posted something along the lines of "yeah, but its not all about the DAC chip" ! This is not about the noise measured within a DAC under test but about reducing the noise and signal timing inaccuracy in the data stream input to the DAC.

I would like to see Amir's response to this as it directly contradicts his findings on DACs and USB cables

I accept that digital signals are represented as a binary analogue voltage wave, but noise within this signal will only matter if it causes the DAC to misinterpret the byte. A good DAC, as the measurments show, will not allow noise generated from a computer to pass down the analogue signal after conversion.

 

[chasefrench]

As this previous thread discusses a similar point, it might be worth reading.

https://www.audiosciencereview.com/...puter-activity-can-impact-dac-performance.22/

On a well engineered DAC, computer noise is a non-issue

 

[Fleuch]

The thread referred to in #9 of this thread seems to confirm what has previously been discussed.

It is not about USB cables but noise that can be injected into the signal path. In that thread (https://www.audiosciencereview.com/...puter-activity-can-impact-dac-performance.22/), at #15 Amir writes "Get a good, isolated USB interface and then it shouldn't make a difference what is going on in the PC. Certainly I don't worry about it on my good interfaces." This makes it abundantly clear that the USB interface [to the DAC] should be isolated from the computer. It is also worth looking again at #32 in the same thread, which emphatically confirms the problem of noisy computer USB output ports.

Perhaps not using the same words, but this, and the posts following, really confirm the need to isolate the input to a DAC from a USB output on a computer. Pick up on cables is a contributing factor but it is sources that have the potential to inject unwanted signal noise into the system that is the main problem and how to manage and minimise that noise.

Agreed that that a well-engineered DAC should not inject noise into the system.

A poorly engineered USB output port on a computer is more than capabe of injecting signal line noise, unfortunately. Even with a well-engineered DAC the old acronym GIGO is applicable, garbage (or unwanted signal noise in), garbage (or poor sound quality) out. However the discussion is about overall system noise, not solely about internal noise generated in a DAC.

This is supported by #6 in this thread. It is difficult to understand how a good DAC will prevent noise present at the input being passed to the output. If the DAC interprets the input bit pattern correctly, with noise corrupting the input bit pattern, the output will represent that incorrect bit pattern when compared to the original data source. The DAC output will not be a true representation of the source input and the "distorted" output may be apparent in the amplifier output.

When Amir writes "Certainly I don't worry about it on my good interfaces" this implies that computer USB output ports are not connected directly to the DAC test rig. If a computer is used in the test rig this confirms there is galvanic isolation between the computer and the test rig. To emphasise what has already been written in this thread, Amir tests for the internal noise generated by the DAC. This does not include unwanted USB power line noise, pulse shape distortion, ground loop issues or EM interference. Careful placement of components in a DAC, or any other device, is required to minimise internal noise generation and goes some way to explaining why the air gap separation between sensitive components, particularly power supllies, on the circuit board(s) is a high priority at the design stage. In the same way careful placement of cables in a music system may reduce any noise present, which is why it is advisable to keep all signal interconnects as short as possible.

The Darko video report on the Volumio Primo (https://darko.audio/2019/06/a-short-film-about-the-volumio-primo/) is most complementary but it would be good to hear from others who have experience of using it, or any other similar devices.

 

[creativepart]

I used the SoTM SMS 200 for 9-months or so before moving to the SoTM SMS 200 Ultra. I like the SQ improvement over directly connecting my iMac to my Chord Qutest DAC via USB. Your mileage may vary.

There are other choices other than network streamers - a Schiit Eitr converts USB to S/PDIF. You can use a device like this from your computer USB to the Eitr and then connect your DAC via S/PDIF as another way to isolate USB noise and power from your DAC. Amir has tested an older version of this device, as well.

My guess is, that at ASR the folks here will be decidedly on the NO side of this discussion. The consensus here seems to be summed up in one sentence: "A cheap computer, a usb cable and a Topping Dac and your HiFi world is all set."

 

[majingotan]

My guess is, that at ASR the folks here will be decidedly on the NO side of this discussion. The consensus here seems to be summed up in one sentence: "A cheap computer, a usb cable and a Topping Dac and your HiFi world is all set."

Not everyone has problems with USB noise BTW hence the omittance of a dedicated streamer which I would say the majority of users here including me. A streamer doesn't just benefit from the lack of USB noise (as it omits USB in the signal path), but it allows a smartphone or whatever portable device you have as a remote to access your library on the NAS or streaming services.

There's another method but more common in the professional side called Dante interface. It uses the Ethernet (you can add that EtheREGEN in the chain too if you absolutely want peace of mind FWIW) to a transport that converts the Ethernet to AES/EBU or SPDIF then goes to the DAC.

 

[PierreV]

My guess is, that at ASR the folks here will be decidedly on the NO side of this discussion. The consensus here seems to be summed up in one sentence: "A cheap computer, a usb cable and a Topping Dac and your HiFi world is all set."

I believed one can still occasionally be unlucky on USB on some computers or with some peripherals. I've been building my own PCs and most of my business PC/servers since the 8086 days and have seen crappy stuff now and then. Ground loops may still be a pain with USB powered headsets for examples and some pre-built PCs or laptops are still intrinsically sub-optimal for some reason. As a preventive measure, I still tend to purchase motherboards with optical out, but it doesn't really seem to matter anymore at least if you don't use very low end components. Latest build is Gigabyte X570 Aorus Master + Ryzen 3900x + decent power supply. Even when the video cards are pushing it, I have not been able to notice any issue.

That being said, yes isolators are cheap but if people can isolate with a small add-on, can't we accept that the designers of a good DAC are also able to take care of the issue (and being immune to eventual trash is probably equivalent to being a good DAC anyway, given the number of options that perform beyond audibility in the other aspects).

It only gets ridiculous when people spend more on "cleaning" a signal (without objective measurements in most cases) than they would on a decent PC/Soundcard or motherboard. Or maybe naive when people insert a non-PC in the chain that turns out to be a PC anyway.

 

[chasefrench]

The thread referred to in #9 of this thread seems to confirm what has previously been discussed.

It is not about USB cables but noise that can be injected into the signal path. In that thread (https://www.audiosciencereview.com/...puter-activity-can-impact-dac-performance.22/), at #15 Amir writes "Get a good, isolated USB interface and then it shouldn't make a difference what is going on in the PC. Certainly I don't worry about it on my good interfaces." This makes it abundantly clear that the USB interface [to the DAC] should be isolated from the computer. It is also worth looking again at #32 in the same thread, which emphatically confirms the problem of noisy computer USB output ports.

Perhaps not using the same words, but this, and the posts following, really confirm the need to isolate the input to a DAC from a USB output on a computer. Pick up on cables is a contributing factor but it is sources that have the potential to inject unwanted signal noise into the system that is the main problem and how to manage and minimise that noise.

Agreed that that a well-engineered DAC should not inject noise into the system.

A poorly engineered USB output port on a computer is more than capabe of injecting signal line noise, unfortunately. Even with a well-engineered DAC the old acronym GIGO is applicable, garbage (or unwanted signal noise in), garbage (or poor sound quality) out. However the discussion is about overall system noise, not solely about internal noise generated in a DAC.

This is supported by #6 in this thread. It is difficult to understand how a good DAC will prevent noise present at the input being passed to the output. If the DAC interprets the input bit pattern correctly, with noise corrupting the input bit pattern, the output will represent that incorrect bit pattern when compared to the original data source. The DAC output will not be a true representation of the source input and the "distorted" output may be apparent in the amplifier output.

When Amir writes "Certainly I don't worry about it on my good interfaces" this implies that computer USB output ports are not connected directly to the DAC test rig. If a computer is used in the test rig this confirms there is galvanic isolation between the computer and the test rig. To emphasise what has already been written in this thread, Amir tests for the internal noise generated by the DAC. This does not include unwanted USB power line noise, pulse shape distortion, ground loop issues or EM interference. Careful placement of components in a DAC, or any other device, is required to minimise internal noise generation and goes some way to explaining why the air gap separation between sensitive components, particularly power supllies, on the circuit board(s) is a high priority at the design stage. In the same way careful placement of cables in a music system may reduce any noise present, which is why it is advisable to keep all signal interconnects as short as possible.

The Darko video report on the Volumio Primo (https://darko.audio/2019/06/a-short-film-about-the-volumio-primo/) is most complementary but it would be good to hear from others who have experience of using it, or any other similar devices.

I've linked Amir's testing methodology below:

https://www.audiosciencereview.com/forum/index.php?threads/understanding-audio-measurements.2351/

If you scroll down to the section on PC testing, he implies that he tests the with a direct computer to dac usb interface on a normal computer running stock windows without modification.

Hopefully we can agree that if the best dacs like the new Topping D90 are able to produce their scores without any form of modification or special equipment, only using a standard usb cable direct from a PC, then computer noise is not a problem and they must have the required isolation circuitry already built in.

Keen to hear from the forum if this is not the consensus understanding...

 

[Willem]

I use the optical output of a Chromecast Audio and it is sonically perfect, very convenient, and costs almost nothing.

 

[DDF]

I use the optical output of a Chromecast Audio and it is sonically perfect, very convenient, and costs almost nothing.

I do as well but significant chromecast drawbacks I dont like are Spotify transcodes from ogg to aac and gapless isnt't supported, Tidal gapless requires Roon or Bubble (battery hog in this mode), and neither tidal nor spotify support a decent eq option when casted. So I rarely use it for serious listening

 

[DDF]

I believed one can still occasionally be unlucky on USB on some computers

I have 2 Dell e6420s, good business class laptops in their day, both are unusable to any usb dac ive put on them without switching to battery or ungrounded supplies. I hope this is a rarity but surprised to see a major company's mid range pc's so problematic, and the dacs are faultless.

 

[majingotan]

It’s a Dell. Not that surprising. I use Microsoft Surface Pro 2017 fan less model which has zero electrical/coil whine noise issues on USB plugged in on its docking station. I used to have a Lenovo ideapad Y470 back then and that too had no USB electrical noise issues except for the noisy fans cooling that 45W 2nd Gen i7 chip

 

[chasefrench]

My experience is similar to PierreV, overclocked 6 /12c server chip and an Asus ROG board, massive fans and heatsink. Not a hint of audible noise on a benchmark dac 1 usb (average by Amir's standards today) and Audeze LCD3.

However, DDF's Dell experience supports Fleuch's argument that "bad" computers will increase electrical noise. It will be interesting to see if a terrible computer is negated by a top tier DAC.

 

[DDF]

My experience is similar to PierreV, overclocked 6 /12c server chip and an Asus ROG board, massive fans and heatsink. Not a hint of audible noise on a benchmark dac 1 usb (average by Amir's standards today) and Audeze LCD3.

However, DDF's Dell experience supports Fleuch's argument that "bad" computers will increase electrical noise. It will be interesting to see if a terrible computer is negated by a top tier DAC.

cc @PierreV

Computers are reviewed with synthetic benchmarks of computing performance (PCMark etc). The Dell did well. Dacs are reviewed with synthetic benchmarks of audio performance (SINAD etc). No one is testing interoperation between the two in a meaningful way (i.e. using a "bad" computer). For the consumer, there is no way to suss out a "bad" computer for USB audio, its a buy and try situation.

Two of the dacs tested with the Dell were a $2.5k highly reviewed item (name withheld to protect the guilty) and an ARCAM irdac. Both produced audible noise at the listening seat. An Olimex USB isolator cured both of the noise, but the expensive DAC then suffered "clicks" every minute or so. Root cause would require an engineering investigation of the eye diagram: these low cost isolators use the same chip with relatively high jitter, so pure conjecture would lean towards timing slips. This illustrates yet another USB audio interop risk that's not tested.

A Galvanic isolator would fault-proof many set ups** but the manufacturers won't lift a finger until feeling some commercial pressure, similar to what @amirm is doing for SINAD and power output testing of dacs and headphone amps. Its cheap and easy (I added it 25 years ago to my CD player's SPDIF).

As a simplification to derisk purchasing decisions, ideally all dac reviews would comment on whether galvanic isolation was used on USB or not. I think very few DACs do.



** Many forum members are doubtful of these interoperation issues because they haven't experienced them. Lucky for them, but an internet search uncovers many that have, even with highly spec'ed and well reviewed/benchmarked PCs