Why do different USB ports sound different?

[Audiofire]

Pete Brown from Microsoft talks about how the performance of different USB ports can vary.

He said that one USB port can have more latency than another, which is true. Funnily enough, he said in the same video that this doesn't matter for audio playback (only audio recording). Decent software players have adjustable buffer size.

 

[pablolie]

He said that one USB port can have more latency than another, which is true. Funnily enough, he said in the same video that this doesn't matter for audio playback (only audio recording). Decent software players have adjustable buffer size.

USB interfaces are not even remotely challenged by real time media these days. Zero chance latency is an issue unless it is one of those incompetent special audio implementations. Look up the speeds. USB is bit accurate and secure. Far more critical applications than audio leverage USB and Ethernet and what not.

It is tiring to hear this voodoo charlatenerie claiming interfaces that are completely trusted in mission critical applications supporting Gigabit+ speeds somehow are flawed for audio. The military uses them. Medical equipment. Banks. Come on. Major yawn here. Audio does not pose the remotest challenge to any of these interfaces.

 

[Audiofire]

Read speed of formats is more interesting, where HDDs and especially optical media can be a relevant issue.

 

[sam_adams]

charlatenerie

I commend you on your perspicacity.

 

[Audiofire]

cromulent forum

 

[pogo]

...Medical equipment...

In mission-critical applications, however, you will not find interfaces in simple form, but in redundant form, possibly even double redundant, which check each other. The interfaces are electrically isolated to the maximum.

 

[KSTR]

I have an Intel NUC10i5FNH mini-computer and a Topping MX3s DAC. Why does the DAC sound different when connected to different USB ports? Both USB ports are USB 3.1 gen2. This is not self-hypnosis, I even did blind testing to make sure. Which USB port should I choose? The one that I like the sound from more?

IME, different USB ports of the same computer sounding different is always an EMC scenario. You may have some form of "GND loop" noise or RF disturbance causing problems downstream (demodulation in the analog section of the DAC or any further downstream gear like amps).
USB ports on different locations on a motherboard can indeed behave differently with regard to the above, I've measured that myself (with NUCs, actually).

To check (and 100% fix) the issue, use an USB Isolator like Topping's own HS01 (or HS02, as HS01 has reports of premature death from overheating) or similar (make sure they can handle USB2.0 High Speed -- 480Mbit/s).
This is my go-to recommendation for any audio enthusiast for years, isolate your USB connection to the DAC and you're all set with 100% guarantee for success.

 

[Sokel]

NUCs are notoriously noisy,I wouldn't trust any conclusion drawn out using a device like this.

 

[audio_tony]

NUCs are notoriously noisy,I wouldn't trust any conclusion drawn out using a device like this.

I'm using an older NUC (Intel(R) Celeron(R) CPU N2820 @ 2.13GHz) for my player and don't have any noise issues at all.

I've not had access to newer versions to test though.

 

[Sokel]

I'm using an older NUC (Intel(R) Celeron(R) CPU N2820 @ 2.13GHz) for my player and don't have any noise issues at all.

I've not had access to newer versions to test though.

Don't know the specific but old ones around 2015 were bad,lots of test in diyAudio.
Some of them didn't even deliver the spec'd 500mA (tested with ESS9018 and Amanero I²S>USB at high rates where power needed if I recall right) but only did something like 300mA or so.
Noise was bad too.

 

[somebodyelse]

He said that one USB port can have more latency than another, which is true. Funnily enough, he said in the same video that this doesn't matter for audio playback (only audio recording). Decent software players have adjustable buffer size.

Notably the latency he mentions is NOT in the latency of the audio stream, but in the USB protocol. This would be in how long the 'I need some more samples' message takes to get from the DAC to the CPU to be handled, and the corresponding message with audio samples from the CPU back to the DAC. The only way this has an audible effect is if the DAC's audio buffer runs out of data because the next packet of samples didn't arrive in time. You will probably hear this as a click or pop.

 

[JeremyFife]

I have an Intel NUC10i5FNH mini-computer and a Topping MX3s DAC. Why does the DAC sound different when connected to different USB ports? Both USB ports are USB 3.1 gen2. This is not self-hypnosis, I even did blind testing to make sure. Which USB port should I choose? The one that I like the sound from more?

Use the USB port that you prefer the sound from. Listen to music. Stop worrying about electronics

 

[audio_tony]

Don't know the specific but old ones around 2015 were bad,lots of test in diyAudio.
Some of them didn't even deliver the spec'd 500mA (tested with ESS9018 and Amanero I²S>USB at high rates where power needed if I recall right) but only did something like 300mA or so.
Noise was bad too.

Interestingly I'm using my NUC with an Amanero at the moment. In fact I have the Amanero, a Topping D10 and CM6631A all connected to the same NUC and no issues at all!

I also did some measurements of it here -> https://gtkc.net/usb-noise-measurements

 

[audio_tony]

Notably the latency he mentions is NOT in the latency of the audio stream, but in the USB protocol. This would be in how long the 'I need some more samples' message takes to get from the DAC to the CPU to be handled, and the corresponding message with audio samples from the CPU back to the DAC. The only way this has an audible effect is if the DAC's audio buffer runs out of data because the next packet of samples didn't arrive in time. You will probably hear this as a click or pop.

And in any modern PC (anything less than 15 years old?) this would likely never be an issue.

Microsoft still seem to have some issues with the Windows USB implementation. I use a Windows 10 PC for audio testing, and as such I can have multiple USB devices connected to it (USB oscilloscope x2, audio interface - sometimes more than one - mouse, keyboard etc.) and it's not uncommon to have a device just 'drop off' - unplugging it, and plugging it back in seems to solve the issue.

For what it's worth, the machine is dual boot with Linux and I have never have this issue with Linux.
I guess it could be driver related too, however when you think back to the teething issues Microsoft had with Windows 95 OSR2 (the first time Windows supported USB) and then even Windows 98 continued to have issues.
They just seem to struggle to get it right in the early days for some reason.

 

[somebodyelse]

Some of the unwritten rules of "USB ports don't make a difference" are:

1. You have a competent DAC - one that's designed to handle the expected imperfections of USB without degrading performance.
   • The Modi 2 is an example of a DAC that doesn't meet this standard. Tests here show this to be the exception rather than the rule. Fix this by getting a competent DAC.
2. Your USB ports are within spec and don't have hardware/firmware/OS bugs.
   • @Sokel suggests some NUCs couldn't supply the specified current
   • AMD had an issue a few years back that caused crackling on USB audio, among other things
   • Pi issue #2215 may affect the type C port usually used for power on the Pi 4, but not the type A ports.
3. You don't have a ground related noise issue - ground loop hum/buzz, 'USB noise' with GPU activity, mouse movement etc.
   • Different ports may have different ground routing, leading to differences in ground noise
   • Fix this with a standards compliant USB isolator. High speed ones have recently become relatively inexpensive.

Archimago's measurements show essentially identical performance with the USB ports on a Pi 3, Odroid C2 and MS Surface 3. Differences are at the level of run to run variation.

 

[somebodyelse]

And in any modern PC (anything less than 15 years old?) this would likely never be an issue.

Probably, as demonstrated by USB hubs not usually causing problems with audio. We're into 'highly unlikely but conceivable' territory here. If you already have an interrupt latency issue of just the right size (some manufacturer's poor/buggy driver?) the extra latency might just be the last straw, but even then it's a minor contributor not the root cause of the problem. In the exceedingly unlikely event that this is the case, switching ports is easier than getting a manufacturer to fix their driver.

 

[egellings]

You would have said "noise", or you would have described the noise, as hiss, hum, buzz, whine, clicking, etc.

Sometimes people don't have the language to describe what they're hearing but with electronics it's usually noise, distortion or frequency response, or the left channel is quieter than the right, etc.

You could have noise, but that would not cause the music itself to sound different, unless the noise was severe enough to interfere with the proper operation of the device. It that were the case, I think you would stop listening pretty quickly.

 

[pablolie]

You *can* screw up USB with extremely busy, very bandwdith hungry devices. But think about it... USB 3.0 supports 5Gbps per spec, let's say because specs are always under ideal circumstances you can load a USB port with 1Gbps (admitthat's a VERY conservative scenario). That's the equivalent of about ~2k uncompressed 16/44 red book streams concurrently. it'll take a LOT of wrong-doing for USB to not easily support audio streams. I seem to recall USB2.0 had something like a 100microsecond polling for devices, unlikely to ever reamotely cause latency or jitter issues even with HD audio. But it is *not* impossible to screw it up if one adds faulty/misbehaving devices.

As to noise, I'd never attribute it to a digital transport medium. It will be something that's impacted on the analog conversion end.

 

[LongKick64]

And some here claim that there can be no difference between different USB ports. Here is proof that there is a difference! With measurements. https://archimago.blogspot.com/2018/05/measurements-computer-usb-5v-power.html#more

 

[Audiofire]

And some here claim that there can be no difference between different USB ports. Here is proof that there is a difference! With measurements. https://archimago.blogspot.com/2018/05/measurements-computer-usb-5v-power.html#more

Decent audio interfaces filter the power to have smooth DC supply in the device itself. External power supplies can easily give smoother DC, but your anecdotes how the audio output is affected are no proof. The Topping MX3s DAC has an external power supply (not from USB).