Battle of S/PDIF vs USB: which is better?

[Kal Rubinson]

Yeah. I have gotten to hate any audible noise in my system. For a while, I had a preamp that I loved for every reason but one: High residual noise. It caused me to jump through hoops in the rest of the system to rebalance gain structure and minimize the output noise which was, then, barely audible at the speakers. I've now dispensed with that preamp (and any other) and go from server to DAC by USB or LAN and from DAC directly to power amps. No (audible) noise whatsoever with my ear to the speakers.

 

[remlemasi]

I’m losing the S/PIDF out on my 2015 MBP as soon as I upgrade this year. Any recommendations for a USB DAC or should I just get a cheap bridge?

How much would I need to spend on a new USB DAC to sound better than a $40 Douk U2 USB bridge to my Benchmark DAC1 (2003 build date)?

Would a $150 Topping E30 be better already? I primarily listen to 16/44.1 lossless and once in a while hi-rez via Tidal MQA (no final hardware unfolding).

 

[Chrise36]

The Topping E30 is enough probably or the D10S which has spdif out will suit you better

 

[Veri]

Seconding the D10S for RCA + s/pdif out.

 

[Sal1950]

Yeah. I have gotten to hate any audible noise in my system. For a while, I had a preamp that I loved for every reason but one: High residual noise. It caused me to jump through hoops in the rest of the system to rebalance gain structure and minimize the output noise which was, then, barely audible at the speakers. I've now dispensed with that preamp (and any other) and go from server to DAC by USB or LAN and from DAC directly to power amps. No (audible) noise whatsoever with my ear to the speakers.

That's the way it was for me with my VTL monoblock tube amps and super high efficiency Klipsch LaScala speakers. When everything was perfect I could get the noise low enough that I couldn't hear it at the listening chair and a quiet room. But every now and then one of the 12AT7 input tubes would get noisy and it would drive me nuts till I replaced it. The high input sensitivity of the VTL's contributed to the issue (JA measured it as 720mV back in 1988.) Great amps in their time but a bit quirky that needed to be attended to.

 

[remlemasi]

The Topping E30 is enough probably or the D10S which has spdif out will suit you better

Seconding the D10S for RCA + s/pdif out.

I guess I could get the D10s, use S/PDIF out to the DAC1 and have an extra/backup DAC if I ever need it. But is the D10s analog out really better sounding (not just better performing) than the DAC1?

 

[Chrise36]

I guess I could get the D10s, use S/PDIF out to the DAC1 and have an extra/backup DAC if I ever need it. But is the D10s analog out really better sounding (not just better performing) than the DAC1?

That is an answer you must reply for your self. You might find some differences in the high frequencies if any.

 

[remlemasi]

That is an answer you must reply for your self. You might find some differences in the high frequencies if any.

Fair enough. Thanks.

 

[Offler]

My 2 cents to the debate, but I assume there is nothing new to add... I compare USB and SPDIF from the way how they are constructed and how they operate.

USB in general requires a Processor to operate, and that require memory, or at least cache which acts as a buffer. In the past, the processor was usually CPU of a computer, more modern devices are using processor in the USB interface and devices as well. Also from what I have learned about modern USB Sound cards and DACs, this has become very common.

SPDIF interface is in contrast very simple, and does not require processor (x86, x64 or ARM or other common architectures) and data are rendered from output and presented to the DAC in a very raw form.

Problem with older USB interfaces was that they relied way too much on computer CPUs. That could produce noticeable jitter or distortion simply because CPU was running low on resources. Now much of the work is offloaded to the processors within USB interface and within the device itself, and since computer CPUs have a lot more performance to offer, so the negative effects are long thing of the past.

In this regard, SPDIF wasnt developed that much since, but DACs were. From my understanding full support for 44.1, 88.2, 48, 96 and 192 KHz is a MUST for any DAC and sampling rate conversion should not occur. Also there should be large enough buffers in the DAC to process all the data without overflows.

Reasons for me to use specifically Toslink are lower CPU utilization (according to LatencyMon, the driver appears to run smoother) and filtering out all EM noise from the PC (in my case there is a lot of it even in 20hz-20khz spectrum).

According what I have read about DACs (post 2000) i am quite surprised that sample rate conversion can still a problem, but I am not surprised that USB became a choice for people who use computers as an audio source.

 

[Rene]

All the arguments here are tedious.
SPDIF is a flawed interface, always allowing for the generation of clock jitter (especially via Toslink).. However, in its defense, it is simple to implement.
USB allows for low jitter clocking physiclly near the DAC. It has become relatively simple to implement and is ubiquitous.
Ethernet appears to be even better but at present is complex and costly.

 

[ReaderZ]

Problem with older USB interfaces was that they relied way too much on computer CPUs. That could produce noticeable jitter or distortion simply because CPU was running low on resources. Now much of the work is offloaded to the processors within USB interface and within the device itself, and since computer CPUs have a lot more performance to offer, so the negative effects are long thing of the past.

Also they were using waaaay too much CPU resource, back in early 2000s a USB DAC would make demanding games unplayable.

 

[Rene]

Also they were using waaaay too much CPU resource, back in early 2000s a USB DAC would make demanding games unplayable.

Yes, but this is a discussion of USB for audio.

 

[ra990]

As pointed out in the testing, this debate depends entirely on which DAC you are using and if it's prone to SPDIF jitter or not. The Chord DACs, for example, are immune to jitter and so USB doesn't bring in any benefits other than sample rates higher than 192k. With these types of DACs, I prefer SPDIF to keep the signal as clean as possible going into the DAC.

 

[Rene]

As pointed out in the testing, this debate depends entirely on which DAC you are using and if it's prone to SPDIF jitter or not. The Chord DACs, for example, are immune to jitter and so USB doesn't bring in any benefits other than sample rates higher than 192k. With these types of DACs, I prefer SPDIF to keep the signal as clean as possible going into the DAC.

Agreed. Most ESS dac chips also reclock the SPDIF signal internally to reduce jitter.

 

[ReaderZ]

Yes, but this is a discussion of USB for audio.

Gaming audio is audio too. DAC is not only for music, it is for all sounds.

 

[wwenze]

My 2 cents to the debate, but I assume there is nothing new to add... I compare USB and SPDIF from the way how they are constructed and how they operate.

USB in general requires a Processor to operate, and that require memory, or at least cache which acts as a buffer. In the past, the processor was usually CPU of a computer, more modern devices are using processor in the USB interface and devices as well. Also from what I have learned about modern USB Sound cards and DACs, this has become very common.

SPDIF interface is in contrast very simple, and does not require processor (x86, x64 or ARM or other common architectures) and data are rendered from output and presented to the DAC in a very raw form.

Problem with older USB interfaces was that they relied way too much on computer CPUs. That could produce noticeable jitter or distortion simply because CPU was running low on resources. Now much of the work is offloaded to the processors within USB interface and within the device itself, and since computer CPUs have a lot more performance to offer, so the negative effects are long thing of the past.

In this regard, SPDIF wasnt developed that much since, but DACs were. From my understanding full support for 44.1, 88.2, 48, 96 and 192 KHz is a MUST for any DAC and sampling rate conversion should not occur. Also there should be large enough buffers in the DAC to process all the data without overflows.

Reasons for me to use specifically Toslink are lower CPU utilization (according to LatencyMon, the driver appears to run smoother) and filtering out all EM noise from the PC (in my case there is a lot of it even in 20hz-20khz spectrum).

According what I have read about DACs (post 2000) i am quite surprised that sample rate conversion can still a problem, but I am not surprised that USB became a choice for people who use computers as an audio source.

To playback sound, you need a sound/audio controller that shows up in your Windows device manager. Your sound card, basically.

So USB DAC is essentially a USB sound card. Meanwhile onboard sound is usually Intel HDA located in your Southbridge / ICH, connected to your CPU via the DMI bus (or AMD's equivalent). While internal sound cards would be connected via PCI-E or whatever.

After the audio data reaches the audio controller, the audio device then chooses how to send it out. It can be 3.5mm, or SPDIF, or whatever. I2S solutions also exist. So SPDIF is already at the last stage of "dumb" transmission.

One key point worth noting: The audio controller generates the sampling rate. (VIA Envy owners probably had a blast playing with the highly-controllable setting including 2 onboard crystals plus external input. Which can be changed on the fly for some interesting effects.) Hence the motivation for USB over SPDIF: SPDIF's clock is in the signal and requires clock recovery which causes jitter. Sound card won't have this issue but internal = bad. So USB sound card becomes the logical choice.

So now, why would USB sound card have worse performance than PCI(-E) in terms of CPU utilization and reliability? One of the reason mentioned is internal interfaces (And firewire and TB, I guess) have DMA, while DMA is removed from USB due to security or whatever reason. DMA allows the peripherals to directly access memory, and without it they have to go through the CPU.

https://en.wikipedia.org/wiki/Direct_memory_access#PCI
PCI[edit]
A PCI architecture has no central DMA controller, unlike ISA. Instead, any PCI device can request control of the bus ("become the bus master") and request to read from and write to system memory.

We use USB DACs because they become the external master of the analog output clock. Yet by doing so we made it a "slave" (I guess) of the USB bus. Ironic.

 

[Offler]

Also they were using waaaay too much CPU resource, back in early 2000s a USB DAC would make demanding games unplayable.

Yup. Thats why we now have CPUs within USB bus and USB device, so we can enjoy Specter vulnerability even more

Nowadays its almost impossible to explain to 'snake-oil people' in PC gaming that having significantly higher polling rate than refresh rate of your display has little to no effect, as much of the data is simply discarded...

Yes, but this is a discussion of USB for audio.

Using older USB sound cards was very tedious as they were depending on computer CPU or later on Core#0 Thread#0, and you could end up with artifacts related to this... Its thing of a past, but you can run into NUC systems with ATOM processor and having trouble with certain USB devices.

As pointed out in the testing, this debate depends entirely on which DAC you are using and if it's prone to SPDIF jitter or not. The Chord DACs, for example, are immune to jitter and so USB doesn't bring in any benefits other than sample rates higher than 192k. With these types of DACs, I prefer SPDIF to keep the signal as clean as possible going into the DAC.

This.

 

[ReaderZ]

Yup. Thats why we now have CPUs within USB bus and USB device, so we can enjoy Specter vulnerability even more

Nowadays its almost impossible to explain to 'snake-oil people' in PC gaming that having significantly higher polling rate than refresh rate of your display has little to no effect, as much of the data is simply discarded...

From what I understand the benefit is still very real compare to say, 768khz high res audio.

 

[jockel77]

In post #242 I asked about differences between USB to SPDIF converters. Meanwhile I've bought a Topping D10 (not the newer one D10s) for USB to SPDIF conversion feeding my AK4396 based modified Non OS DAC.

Now, I want to build up a second stereo system in another room. Therfore I'd use another AK4396 DAC.
In this room, the disctance between the PC as source an the DAC itself has cable distance of 10 meters
Software is foobar, using USB out via CM6631A USB-SPDIF converter:

MEASUREMENTS: [UPDATE] Adaptive AUNE X1, Asynchronous "Breeze Audio" CM6631A USB, and Jitter...

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

My questions is what is the best way concerning cable connections?
1. 10m active USB cable - USB-SPDIF converter nearby the stereo-system feeding the DAC
2. Setting the USB-SPDIF converter nearby my PC and connecting via 10m coaxial SPDIF cable:

10m Digital Phono Audiokabel, Cromo Line - LINDY DE

Koaxial Audiokabel : Digital Koaxial Stecker an Stecker

www.lindy.de

 

[tvrgeek]

Curious here in 2021, a few integrated amps like Arcam have only S/PDIF. Of course my music server has only USB. I can see only one choice in budget equipment, but the Arcam is not. In all cases, the snake oil salesmen will find a way to ascribe magic to whatever cable connects your devices.