Error correction - chain with both USB and ethernet to a USB DAC

[asdf1000]

Hi all

 

Appreciate any advice / expertise on the below.

 

From Gordon Rankin:

 

"With Isosynchronous USB transmission, packets are sent without any error correction / resending. But guess what? This is the USB protocol used for audio frames. The bad news is they are not error free. The good news is these Isosynchronous frames are afforded the highest priority in the system.”

 

Source: https://darko.audio/2016/05/gordon-rankin-on-why-usb-audio-quality-varies/

 

So I get that.

 

Scenario (example only): if one had an isochronous ICRON 2301GE-LAN (for example) which takes USB audio, converters to ethernet for sending over the LAN, then converts back to USB for feeding the USB DAC...

 

... and assuming there ARE errors in the first USB cable transmission in the chain (between USB audio source and the ICRON sender unit), then since ethernet transmission follows in the chain (and ethernet transmission is sent WITH error correction/resending)....

 

.... does this mean any USB errors in the first USB section of the chain will be corrected / resent, since ethernet transmission follows?

 

In other words,  transmission WILL be error corrected / resent, right up to the ICRON receiver unit, regardless of what happens in the first USB cable section?

 

Cheers!

[hsmeets]

Short answer: no.

 

What you send into the Icron at one end emerges at the other (at best). So malformed USB will emerged malformed at the other end. GIGO :-)

 

The Icrons are mediaconverters, basically they take the USB electrical signal, adapt it for sending it longer distance over UTP cables (or optical fiber) and convert it back to USB electrical signal. The Icron's have no clue about what is in the USB data/signal. Basically they convert one type of LVDS to another and back again (LVDS=Low voltage differential signaling) this is changing electrical levels and impedance adjustments for the transmission line (=cable).

 

The Icrons are not protocol converters and are also a peer-to-peer connection, you can't mixed them with your LAN or connect via switch.

 

Why would you still use synchronous USB today for audio as a-synchronous USB audio solved that issue?

 

 

 

 

 

 

 

[asdf1000]

5 minutes ago, hsmeets said:

The Icrons are not protocol converters and also a peer-to-peer connection, you can't mixed them with your LAN or connect via switch.

 

Hi, thanks for the info. However the model I linked to is isochronous and so the LEX unit can be plugged into a router for example and the REX plugged into a switch on the same network, and they should talk to each other.

 

 

 

[hsmeets]

Ah, yes, that particular model can be mixed in your existing LAN via a switch (level 2). These indeed do protocol conversion by sending the USB signal as the payload in an ethernet frame.

 

But still will not correct for errors that are already in the USB data when it is received at the sender.

[asdf1000]

51 minutes ago, hsmeets said:

But still will not correct for errors that are already in the USB data when it is received at the sender.

 

 

Many thanks, this is the info I was after.

 

A followup question: USB errors are NOT always clearly audible right? You can have USB audio errors but still not hear dropouts?

 

[mansr]

7 hours ago, Em2016 said:

From Gordon Rankin:

 

"With Isosynchronous USB transmission, packets are sent without any error correction / resending. But guess what? This is the USB protocol used for audio frames. The bad news is they are not error free. The good news is these Isosynchronous frames are afforded the highest priority in the system.”

It's called isochronous. I wouldn't trust someone who can't even use the correct terminology.

[mansr]

3 hours ago, Em2016 said:

A followup question: USB errors are NOT always clearly audible right? You can have USB audio errors but still not hear dropouts?

Each USB frame contains 125 μs worth of audio data. A single lost frame is unlikely to be noticed, though depending on the receive and the surrounding data, there may be a minor tick. If errors start occurring several times per second, it will be noticed. Fortunately, you're unlikely to see even one error per week. It's a non-issue.

[asdf1000]

19 minutes ago, mansr said:

Fortunately, you're unlikely to see even one error per week. It's a non-issue.

 

The background of my query was the below post from John S. So I was thinking about this, when USB cables are used in combination with isochronous ethernet (or fiber) transmission (like the ICRON LAN ranger/rover and fiber units). Especially the 1st USB cable in the chain (at the sender end).

 

From the below, a good rule may be to keep the USB cables <1m, then errors should be a non-issue as you say.

 

Or just get Supra high speed certified USB cables, if you need longer lengths (certified for up to 15m !)

 

 

 

[pkane2001]

19 minutes ago, mansr said:

Each USB frame contains 125 μs worth of audio data. A single lost frame is unlikely to be noticed, though depending on the receive and the surrounding data, there may be a minor tick. If errors start occurring several times per second, it will be noticed. Fortunately, you're unlikely to see even one error per week. It's a non-issue.

 

Actually, each frame is sent every 125μs and contains a variable number of audio samples, controlled by the receiver. When an error occurs, the whole frame is discarded. This results in multiple samples being dropped from the playback all at once, which would certainly result in audible drop-outs should errors occur more than very occasionally. Errors are not an issue unless you have a really long USB cable, or one intentionally designed to distort digital signal.

[mansr]

6 minutes ago, Em2016 said:

The background of my query was the below post from John S. So I was thinking about this, when USB cables are used in combination with isochronous ethernet (or fiber) transmission (like the ICRON LAN ranger/rover and fiber units). Especially the 1st USB cable in the chain (at the sender end).

 

 

USB errors are still rare enough that they don't matter in practice. Gordon Rankin and John Swenson like to pretend it's a serious problem, possibly to boost their sales of various "remedies." Don't trust them.

[plissken]

41 minutes ago, pkane2001 said:

 

Actually, each frame is sent every 125μs and contains a variable number of audio samples, controlled by the receiver. When an error occurs, the whole frame is discarded. This results in multiple samples being dropped from the playback all at once, which would certainly result in audible drop-outs should errors occur more than very occasionally. Errors are not an issue unless you have a really long USB cable, or one intentionally designed to distort digital signal.

 

Even so, how much data do you think is in that 125μs period? Mansr's point still stands.

 

I would be shocked if you saw one error in 10 billion on a dedicated system driving a USB DAC only. 

 

It's all FUD tactics (Fear Uncertainty Doubt).  I simply use a Belden 18" USB cable. 

[mansr]

10 minutes ago, pkane2001 said:

Actually, each frame is sent every 125μs and contains a variable number of audio samples, controlled by the receiver.

Yes, the number of samples is variable, but it doesn't vary much. If the clocks were exact, each frame would have 6 samples at 48 kHz, and so on. In practice, a frame occasionally needs to have one sample more or less to compensate for clock drift. Either way, the data contained in a frame corresponds to a duration close to 125 μs.

 

10 minutes ago, pkane2001 said:

When an error occurs, the whole frame is discarded. This results in multiple samples being dropped from the playback all at once, which would certainly result in audible drop-outs should errors occur more than very occasionally.

A lost frame can be ignored, interpolated, or zero-padded. An isolated occurrence of either will at worst result in a tick. A steady stream of errors results in buzzing or distorted sound. If the DAC simply ignores bad frames (i.e. doesn't pad or interpolate), too many errors in a short time can cause a buffer underrun which will be audible as a longer drop-out.

 

10 minutes ago, pkane2001 said:

Errors are not an issue unless you have a really long USB cable, or one intentionally designed to distort digital signal.

Right. You have to try hard to get a cable that has just enough errors to be a problem while still working at all.

[pkane2001]

1 hour ago, plissken said:

 

Even so, how much data do you think is in that 125μs period? Mansr's point still stands.

 

125μs worth of data on average, perhaps? Don’t think I disagreed with Mansr’s point in any way.

[mansr]

11 minutes ago, pkane2001 said:

125μs worth of data on average, perhaps?

Yes, and the deviation is at most one sample. It's in the spec.

[Required]

Hi
I am not from the Audio Stuff but I have an similar Problem. Its noise over an Electric Cable. In my case I will receive Signals from an "Radio" Receiver who cover the hole Amateur Radio Band. The Problem with existing Technology is the are creating noise like an Raspberry Pi2 and that reduce the received Packed. ? I searched a lot online and I found an IT Distributor who offer Products from "icron". So I saw that here some people talk about there Products.

 

So my Question is can someone who own a icron USB -> Fiber Set Test something for me?

Thanks

[beerandmusic]

On 2/3/2018 at 8:34 AM, mansr said:

 

A lost frame can be ignored, interpolated, or zero-padded. An isolated occurrence of either will at worst result in a tick. A steady stream of errors results in buzzing or distorted sound.

 

In regards to a stream of errors distorting sound, did you see/hear this first hand, or just read about it?

It appears here that you are acknowledging usb transmission data errors can cause distortion?

 

[Ralf11]

why have a beer when you can have an Error Confection ?

[Required]

I have the problem my Icron USB Fiber extender lost the connection over night.

During the Day it works. Now its gets during the Day worse to.