[gregorio]

A fair amount of bad info in this thread.

There’s some bad info, not as much as in your post though! Jitter/Clocking is a classic example of audiophile marketing picking up on an issue that was solved decades ago, then selling expensive solutions to that already solved non-issue and along the way, inventing all kinds of misinformation, outright lies and lies of omission falsely explaining how and when jitter and noise occurs, it’s consequences and the scale of it. This has been going on for so many years, with so much marketing misinformation/lies invented during that time, it’s hardly surprising audiophiles are completely confused/mislead about jitter and therefore post “bad info”.

Toslink is one of the worst for jitter at up to 20 times worse than decently implemented USB, yet all people want to talk about is it galvanically isolated.

Firstly, that’s “bad info” because “all people” (audiophiles) don’t only want to talk about galvanic isolation, there’s still quite a lot of talk about jitter, clocks, etc. Secondly, “up to 20 times worse” is just marketing BS. While it *might* theoretically be true, so what? The amount of jitter relative to a completely different protocol is irrelevant, what’s relevant is if the jitter exceeds the specifications of it’s own protocol. The data transfer rate of TOSLINK is two orders of magnitude lower than even USB 2 (let alone USB 3 or 4), so USB therefore requires a far lower jitter specification in order to work.

Yea but it still transits noise in the data stream and jitter.

So what? That’s why digital audio was invented and how it’s supposed to work! Audiophile marketing over the years has focused on “noise in the data stream and jitter” because that opens the door to a whole range of audiophile products which are not only completely useless but have profit margins so massive they’re hard to ignore for some/many manufacturers; “decrapifiers”/“cleaners”, audiophile digital cables, audiophile clocks, etc. The marketing BS is largely based on the lie that the digital data stream is effectively an analogue signal. There is always noise in the transmission of signals, the reason digital audio was invented is because below an extremely high level of transmission noise digital audio is completely immune to it. This is unlike analogue signals, where any amount of transmission noise is represented in the output signal and worse still, all the noise is cumulative. In any asynchronous protocol, jitter exists in the transmission but has zero effect and is effectively irrelevant. With synchronous protocols transmission jitter does have some effect but is reduced to orders of magnitude below audibility by the clock recovery circuitry always built into even cheap DACs.

Not the piece of crap clock in our PC trying to time sensitive audio stream via spdif toslink.

More marketing BS. Your PC has to have a clock signal in the several gigahertz range (otherwise it simply would never work), not the paltry several megahertz clock signal required by your DAC. The precision of the clocks (transmitter or receiver clocks) themselves is largely irrelevant, what’s relevant with digital audio is the precision of the clock signal at the point of conversion and that’s determined by the clock recovery circuitry.

This has all been done to death and put to bed decades ago, except in the audiophile world, because so many audiophile products rely on the myth that it’s a relevant/current problem which their products solve!

G

 

[redrol]

A fair amount of bad info in this thread. Toslink is one of the worst for jitter at up to 20 times worse than decently implemented USB, yet all people want to talk about is it galvanically isolated. Yea but it still transits noise in the data stream and jitter.

Can USB be better, not perfect, but yes of course, especially when the USB portion of the dac is well designed. And one has to remember the clocks in most dacs are async and much better with USB. Not the piece of crap clock in our PC trying to time sensitive audio stream via spdif toslink.

There are BUFFERS in audio devices. Jitter doesn't matter. There is zero noise with optical SPDIF. This is stupid.

 

[kenny518646]

There are BUFFERS in audio devices. Jitter doesn't matter. There is zero noise with optical SPDIF. This is stupid.

I'm not quite sure, but for USB audio I think I read some article saying there is no error correction.
If this is true, buffer doesn't matter if "jitter" mess up the signal before it goes in to buffer.
I know someone saying optical is better because ofthere is no power or ground noise, but same I am not sure about it and have no chance to test it.

 

[gregorio]

I'm not quite sure, but for USB audio I think I read some article saying there is no error correction.

Most likely it was an audiophile marketing article, because as I mentioned above, there is a requirement to invent false or already solved problems in order to justify the existence of certain audiophile products. In this case it’s true that USB audio does not have error correction but there’s a typical “lie of omission“ here, it doesn’t have error correction because it doesn’t need it. The USB protocol does have error detection though, which allows the receiving device to apply it’s own error correction if necessary, however it’s virtually never necessary.

If this is true, buffer doesn't matter if "jitter" mess up the signal before it goes in to buffer.

USB, the same as all digital protocols effectively includes a jitter specification. As long as jitter does not significantly exceed that specified amount it cannot “mess up the signal”, so you get the correct/identical data in the buffer, where jitter ceases to exist. If jitter does significantly exceed the specified amount for the protocol, then obviously it’s not compliant with the protocol, it’s effectively not a USB signal. Obviously, something has to go seriously wrong for this to happen and the result would be incorrect/non-identical data in the buffer, which the receiving device would recognise due to the error detection built into the USB protocol. What the device then does about it, is up to the device but in pretty much all cases of a significant number of bit errors what you’ll get is a “drop out”.

I know someone saying optical is better because ofthere is no power or ground noise, but same I am not sure about it and have no chance to test it.

There’s really no need to test it, it was already tested by numerous independent teams of engineers years before the protocol was even ratified and released commercially. Sure, there are both advantages and disadvantages to optical but that doesn’t affect the outcome (perfect data transfer). If it did, the protocol would never have been ratified in the first place and even if it had, it wouldn’t have succeeded/lasted for long as a commercial application. The only thing those advantages/disadvantages affect is the specifications of the protocol, for example the max cable length, max data throughput, etc., and therefore whether it is appropriate in a specific scenario/application that would exceed the specifications of an alternate protocol.

G

 

[kenny518646]

The USB protocol does have error detection though, which allows the receiving device to apply it’s own error correction if necessary, however it’s virtually never necessary.

Just done some research, seems like isochronous mode for USB won't have error detection (and correction), but i agree it is not necessary to have those corrections if the cable is a normal cable. Just buy a normal shielded 10 or 20 dollars USB cable should be enough.

As long as jitter does not significantly exceed that specified amount it cannot “mess up the signal”, so you get the correct/identical data in the buffer

Yes, so having a shielded cable is important if there might be interference in the environment (but i don't think this is really a problem for most people..).

Sure, there are both advantages and disadvantages to optical but that doesn’t affect the outcome (perfect data transfer).

I am not an EE major, so if I am wrong please let me know. The digital will not be affected as long as the signal is correct(as last point). But the ground noise or noise in the 5V power (this happened in some but not all DAC cuz not all DAC use 5V from USB) may affect the analog signal that has been converted by DAC.

 

[gregorio]

Just done some research, seems like isochronous mode for USB won't have error detection (and correction),

As far as I recall off the top of my head, even isochronous USB mode included error detection. Not that it matters much now, USB asynchronous mode has been the standard for quite a few years and the BER (bit error rate) of older synchronous mode devices was almost infinitesimal anyway.

but i agree it is not necessary to have those corrections if the cable is a normal cable.

I presume that by “normal cable” you mean certified USB cable, in which case “yes”, obviously that’s the whole point of having a ratified protocol in the first place.

Yes, so having a shielded cable is important if there might be interference in the environment (but i don't think this is really a problem for most people..).

Shielding is essential from the point of view that it’s demanded by the USB protocol and therefore it can’t be a USB cable without shielding. I again don’t recall off the top of my head exactly what the shielding specifications are but they don’t need to be particularly demanding because the USB protocol specifies a differential signal. So unless you do something silly, it’s not a concern. The data rates for the most recent USB specifications are very high though and more effective shielding is probably specified than for USB 2.

The digital will not be affected as long as the signal is correct(as last point).

I wouldn’t phrase it as “correct”, the whole point of digital audio/data in the first place is that it doesn’t have to be “correct”. It can actually be very badly “messed up” without any affect at all, as long as it’s not messed up so badly it no longer complies with the specifications of the protocol.

But the ground noise or noise in the 5V power (this happened in some but not all DAC cuz not all DAC use 5V from USB) may affect the analog signal that has been converted by DAC.

Of course, once we’re in the analogue realm we’re back to and stuck with the rules/laws of analogue signals. Namely, any noise/distortion/interference of the analogue signal will affect the output and if it’s of sufficient magnitude will affect the sound (and therefore potentially be audible). However, you would obviously expect the manufacturer of a DAC to know the difference between digital and analogue signals! And therefore expect them to make sure noise/distortion/interference doesn’t leak into the analogue side of the DAC to any appreciable extent (in the usage scenario for which it was designed). So, a consumer DAC used appropriately is essentially defective if it allows noise/distortion to leak from the 5v power or ground into the analogue output signal to the extent of being audible (at any reasonable consumer listening level).

G

 

[Roseval]

even isochronous USB mode included error detection.

Correct.
Isochronous Transfers are used for time sensitive information, such as an audio or video stream

• Guaranteed access to USB bandwidth.
• Bounded latency.
• Stream Pipe - Unidirectional
• Error detection via CRC, but no retry or guarantee of delivery.
• Full & high speed modes only

The frame contains a CRC allowing the receiver to do error detection.
Packages of data are send to the DAC. The amount is such that the DAC can just maintain the sample rate. Because of the CRC the DAC can detect errors but it is of no use to ask for a retry as next package of data is already on its way. That is the price of low latency.
https://www.beyondlogic.org/usbnutshell/usb4.shtml#Isochronous

 

[kenny518646]

it can’t be a USB cable without shielding

There are pretty much USB cable that doesn't have shielding :0

As far as I recall off the top of my head, even isochronous USB mode included error detection

okay, so probably what I read is still not accurate enough (although lots of them saying there is no correction tho).
but as the one above said, there is not guarantee of delivery.

Of course, once we’re in the analogue realm we’re back to and stuck with the rules/laws of analogue signals. Namely, any noise/distortion/interference of the analogue signal will affect the output and if it’s of sufficient magnitude will affect the sound (and therefore potentially be audible). However, you would obviously expect the manufacturer of a DAC to know the difference between digital and analogue signals! And therefore expect them to make sure noise/distortion/interference doesn’t leak into the analogue side of the DAC to any appreciable extent (in the usage scenario for which it was designed). So, a consumer DAC used appropriately is essentially defective if it allows noise/distortion to leak from the 5v power or ground into the analogue output signal to the extent of being audible (at any reasonable consumer listening level).

So I believe this is why some people always saying usb is not good or optical is better. No matter what I still use USB for my dac, it is convenient for me

 

[gregorio]

There are pretty much USB cable that doesn't have shielding :0

I’ve never seen or heard of one and again, as far as I recall it must have at least some shielding to be certified as a USB cable. Obviously the differential signaling reduces the shielding requirement significantly though.

okay, so probably what I read is still not accurate enough (although lots of them saying there is no correction tho).

Not accurate enough by a long way apparently. The USB-IF (usb.org) develops and administers the USB protocols, so the information there is accurate/definitive. Again, there’s no error correction built-in to USB isochronous mode (unlike RedBook CD for example) but that doesn’t mean there is no error correction, a device can implement it’s own error correction.

but as the one above said, there is not guarantee of delivery.

If there’s no delivery, there’s no data and therefore no sound.

So I believe this is why some people always saying usb is not good or optical is better.

Not sure I understand, “some people always saying usb is not good” because DAC manufacturers don’t know the difference between digital and analogue signals and therefore make defective DACs? There’s no doubt that the power delivery specs of USB make it more problematic but unlike the implications/statements of audiophile marketing, this doesn’t mean that such difficulties can’t be or haven’t been overcome. Even very cheap DACs more than adequately deal with this difficulty, so why should it be such an issue for audiophile DACs? The answer is that’s is not, but it is a marketing opportunity, as I mentioned previously.

G

 

[Voidrunner]

"a consumer DAC used appropriately is essentially defective if it allows noise/distortion to leak from the 5v power or ground into the analogue output signal to the extent of being audible (at any reasonable consumer listening level)."

I see people saying this an awful lot, but there's no shortage of consumer DACs, from quite well-regarded manufacturers, that still occasionally suffer from ground-loop issues, GPU noise/coil whine, etc. Which is usually resolved by inserting a USB isolator into the chain. I experienced those issues myself with both an SMSL M500 MkIII and a Fiio K7, and for both of them, a USB isolator completely got rid of the problem. Pretty good odds they're not "defective" and I'm willing to bet that almost any brand of DAC that you could think of, you'd be able to find examples of people having problems with ground loops/GPU noise in pretty short order.

 

[gregorio]

I see people saying this an awful lot, but there's no shortage of consumer DACs, from quite well-regarded manufacturers, that still occasionally suffer from ground-loop issues, GPU noise/coil whine, etc.

Sure but is that because there’s actually something defective with the DAC or because of some external problem the DAC was not designed to cope with? For example, if there‘s some external ground issue or there’s audible EMI from putting some relatively powerful EM transmitter in close proximity to a consumer DAC not designed for it, EG. Effectively user error.

I experienced those issues myself with both an SMSL M500 MkIII and a Fiio K7, and for both of them, a USB isolator completely got rid of the problem.

I’ve not used either of those, so I don’t know if they are generally defective, if they were just isolated instances of being defective or they weren’t defective and there was some other factor such as user error. I’ve used numerous DACs over the years, too many to count and never encountered a defective one, I have on a couple of occasions experienced audible interference but in both cases that was effectively user error. The problem was solved by eliminating the user error, no USB isolator required. If a USB DAC needs an additional USB isolator to operate correctly, then why wasn’t one built into the DAC the first place or a least the manufacturer clearly stating it needed an additional external USB isolator to work optimally?

G

 

[Voidrunner]

You're underestimating the complexity of the issue. You can take one DAC that passes through GPU noise and hook it up to another system, and get no noise at all.
Every part along the chain can have an influence. Swap to a different GPU, noise might go away. Use a different motherboard, amplifier, etc etc, there may be no issues at all. Or vice versa, you may have no issues to begin with, but you swap out equipment somewhere along the chain, and suddenly you do.

So no, the DACs are not inherently defective, nor is it just "user error", or the DAC operating "incorrectly".
It's a complex issue, and if it IS a problem for your system the easiest way to mitigate it is to do one of two things. Use a USB isolator, or use an optical connection instead.

 

[gregorio]

Every part along the chain can have an influence.

No, that’s the whole point of digital audio in the first place. If your assertion were true the internet wouldn’t work, and neither would USB in many circumstances.

Swap to a different GPU, noise might go away. Use a different motherboard, amplifier, etc etc, there may be no issues at all. Or vice versa, you may have no issues to begin with, but you swap out equipment somewhere along the chain, and suddenly you do.

Is a USB DAC designed to be connected to a computer with a GPU, with consumer motherboards and amplifiers?

So no, the DACs are not inherently defective, nor is it just "user error", or the DAC operating "incorrectly".

If the answer to the above question is “yes”, and yet it passes noise through to it’s analogue section from the consumer computer, then obviously it is defective.

It's a complex issue, and if it IS a problem for your system the easiest way to mitigate it is to do one of two things. Use a USB isolator, or use an optical connection instead.

It’s not a complex issue, it’s a simple issue, is it a consumer DAC or not? If it is but it doesn’t operate optimally with consumer equipment then either it’s being used incorrectly or it’s defective. The solution if there is such an issue is to eliminate the user error or replace the DAC if it’s defective.

G

 

[Voidrunner]

"No, that’s the whole point of digital audio in the first place. If your assertion were true the internet wouldn’t work, and neither would USB in many circumstances."

It has nothing whatsoever to do with the digital signal. No idea where you even got that from. Noise can absolutely be transmitted through a shared ground without affecting the digital signal.

It's not possible for manufacturers to test every conceivable combination of hardware. That's not a reasonable ask.
Take every single USB DAC that you've ever used, spend 5 minutes googling each model, and I will essentially guarantee you will find someone that has run into a noise issue.
You are using your own experience as the model for EVERYBODY's experience, without acknowledging that there could possibly be a different one.

 

[gregorio]

It has nothing whatsoever to do with the digital signal. No idea where you even got that from.

I got it from the fact that USB is a digital signal protocol, isn’t it?

Noise can absolutely be transmitted through a shared ground without affecting the digital signal.

Sure it can, which is why a designer of audio equipment has to take precautions when connecting with a shared ground.

It's not possible for manufacturers to test every conceivable combination of hardware. That's not a reasonable ask.

They don’t have to test every conceivable combination of hardware, it’s a consumer DAC, not a commercial/professional DAC. A competent manufacturer of a consumer DAC wouldn’t have to test “every conceivable combination of hardware”, just exceed the specifications required to avoid contamination from typical consumer hardware by a significant enough margin so that contamination is avoided with even unusually noisy consumer equipment. It’s hardly a recent issue, USB DACs have had to deal with connection to noisy computers/motherboards/GPUs since USB DACs were invented with USB supplied power and even cheap ones managed to avoid the issues you’re apparently plagued with. If a current DAC can’t do what a cheap ~20 year old DAC could do, either there’s some user error or the DAC has been insufficiently designed/tested, IE. It’s defective. It’s not a difficult concept to grasp.

G