[GoldenSound]

It looks like it was the Chord Asio 1.05 driver that is totaling Roon and it's only fixed with a restart. All is good now and Hqplayer actually sounds pretty good. I wouldn't say it bests the M-scaler just yet though, but @GoldenOne 's suggestion of using 18 bits as a setting has helped.
Any suggestions for USB streamer/DDC or filter? I am also curios, do USB reclockers even make sense? So many people swear by the Innuos Phoenix, but I just can't find the reason why it would help.

Best DDC/USB treatment option will be situational as some DACs will perform better in different setups to others.

Whether it's better to use USB or a synchronous connection (AES/SPDIF/I2S) depends on the quality of the DDC vs the quality of the USB implementation/internal clocking of your DAC.

USUALLY, if your DAC has I2S input that's the best way to go. Grab a high performance DDC like an Audio-GD DI20HE or Singxer SU6.
The DAVE does not though so something like the SRC-DX is probably the best option to make sure you can still do the full 768khz.

As to USB reclockers/filters, there are two aspects.
Firstly, 'reclocking' on USB basically just means retransmission/a repeater. This is handy if you're doing longer runs but will not result in a quality improvement/jitter reduction as USB does not provide a clock signal in the same way that I2S/AES/SPDIF do.
Reclocking SPDIF/I2S provides a demonstrable change in performance at the output of the DAC because that clock signal coming from the DDC is controlling the DAC itself. And so a lower jitter DDC will provide a lower jitter output at the DAC analog outs.
Using a high performance clock for the USB controller chip will not provide a benefit to timing at the DAC as there is no clock signal. In fact, USB audio isn't even sent in a constant stream anyway. It's sent in chunks at irregular intervals as instructed by the DAC itself, put into a buffer, and then the DAC uses its OWN clock to convert from that buffer.
As long as that buffer isn't either empty or full, it doesn't matter what the timing of extra stuff arriving into the buffer is. If you check your mailbox at mid-day every day, it doesn't matter if the postman puts letters in there at 10am or 11am.
Your DAC's jitter performance is dependent on its own clocks, NOT the clocks in the USB transmitters.

The second aspect is noise isolation or filtering.
Most of the USB 'reclockers' will have some form of filtering in them to reduce noise on the 5v/ground lines. This can help when you're connecting your DAC to a powerful PC especially, as a big PC is just about the noisiest device you can have in an audio chain. Sometimes this will have an effect that is directly audible, such as hearing GPU whine through the output of your DAC.
And sometimes it's indirectly-audible stuff. For example noise causing the clocks in your DAC to perform worse and increase jitter. Meaning you won't get directly audible noise but the quality of playback may be poorer.

If you want to completely eliminate noise, get something like an ifi iGalvanic or Intona 7055-C. These provide full galvanic isolation, meaning there is literally no way at all for noise to get from the source to the DAC. You could set off a defibrillator on one end and your DAC wouldn't see it.

 

[adrianm]

The second aspect is noise isolation or filtering.

I agree, this is probably the part of the Phoenix that actually does something.

The DAVE does not though so something like the SRC-DX is probably the best option to make sure you can still do the full 768khz.

Would this yield any benefit over USB? There would still be the the noise isolation aspect plus the clocking aspect. Since none of those ddc's do 768, it looks i would need an isolator in front of it anyway if i was planning on using the pc. I could build a raspberry pi, but that's another rabbit hole to go down...
Did you get a chance to test the SRC_DX for yourself?

And sometimes it's indirectly-audible stuff. For example noise causing the clocks in your DAC to perform worse and increase jitter. Meaning you won't get directly audible noise but the quality of playback may be poorer.

This is definitely audible even with galvanic isolation of the USB port.

 

[Triode User]

Did you get a chance to test the SRC_DX for yourself?

Patience, I am sending one to him but Covid decided to come and visit our household!

 

[GoldenSound]

Would this yield any benefit over USB? There would still be the the noise isolation aspect plus the clocking aspect. Since none of those ddc's do 768, it looks i would need an isolator in front of it anyway if i was planning on using the pc. I could build a raspberry pi, but that's another rabbit hole to go down...
Did you get a chance to test the SRC_DX for yourself?

There are quite a few DDCs that do 768khz.
Denafrips Gaia, KTE SU2, SRC-DX, Gustard U18 etc
But those require I2S to do 768khz.

Chord doesn't have I2S input so you have to use their proprietary dual-BNC for 768khz.
I have an SRC-DX on the way so will do some testing once it arrives.

This is definitely audible even with galvanic isolation of the USB port.

Noise won't affect a fully galvanically isolated device cause that's exactly what the galvanic isolation does, blocks any noise.
The confusion comes in because some manufacturers say their DACs are 'galvanically isolated' when they aren't, at least not fully.

For example Chord DACs cut the gnd line but don't isolate the data lines (or power afaik)
Schiit uses transformer isolation which will block DC and prevent ground loops but will still pass HF noise
Denafrips and Holo (and most DDCs) use full optical or silicon dioxide isolators so that no noise at all can pass.

It's likely that this could contribute to why people prefer the SRC-DX over Chord USB

 

[adrianm]

Patience, I am sending one to him but Covid decided to come and visit our household!

Sorry to hear that. Hope you're well.

 

[sajunky]

This is definitely audible even with galvanic isolation of the USB port

Correct. Audio GD DACs have isolator on the USB port, it is still leaky, there are benefits of adding DDC. In this case DI-20 also have USB port isolated but adding ground loop redirector like ifi iDefender with external power supply helps. In this configuration @GoldenOne was testing internal jitter of DI-20 to eliminate external influence.

However I would avoid USB regenerators, as it is not needed on short cables. Use right tools. Clock regeneration does nothing to the SQ, unless signal is weak, it usually breaks datastream completely for a quarter of second. It is not possible that such event get unnoticed.

 

[adrianm]

For example Chord DACs cut the gnd line but don't isolate the data lines (or power afaik)
Schiit uses transformer isolation which will block DC and prevent ground loops but will still pass HF noise
Denafrips and Holo (and most DDCs) use full optical or silicon dioxide isolators so that no noise at all can pass.

It's likely that this could contribute to why people prefer the SRC-DX over Chord USB

Audiowise are claiming that bypassing the USB clock lowers the overall noise and that has an impact on sound quality. They also claim they are filtering the noise, but since people are still using Wave cables with them, or the Opto-dx, not that well probably. There is still probably less than non isolated usb power lines out of a gaming pc.
It sounds like an Intona in front of the SRC-DX might be the way to go unless you have any other suggestion. I was really hoping to get rid of the cables and boxes on my desk.

 

[sajunky]

Noise won't affect a fully galvanically isolated device cause that's exactly what the galvanic isolation does, blocks any noise.
The confusion comes in because some manufacturers say their DACs are 'galvanically isolated' when they aren't, at least not fully.

For example Chord DACs cut the gnd line but don't isolate the data lines (or power afaik)
Schiit uses transformer isolation which will block DC and prevent ground loops but will still pass HF noise
Denafrips and Holo (and most DDCs) use full optical or silicon dioxide isolators so that no noise at all can pass.

Silicon and optical isolators give better protection, but introduce lot of jitter. Asynchronous USB transfers are frequently broken on isolators. It must be a bidirectional isolator (with a feedback signal), it means in practice leaving USB receiver on a dirty side. It is how isolator on DI-20 is made. It doesn't break asynchronous USB transfers.

Even with bidirectional isolators (where theoretically is zero jitter), there are benefits of adding ground loop redirector, see my previous post.

 

[GoldenSound]

Silicon and optical isolators give better protection, but introduce lot of jitter. Asynchronous USB transfers are frequently broken on isolators. It must be a bidirectional isolator (with a feedback signal), it means in practice leaving USB receiver on a dirty side. It is how isolator on DI-20 is made. It doesn't break asynchronous USB transfers.

Even with bidirectional isolators (where theoretically is zero jitter), there are benefits of adding ground loop redirector, see my previous post.

A USB isolator must be bidirectional inherently. USB is a bidirectional protocol. And whether it is/isn't bidirectional would be unrelated to jitter.
It's true that the isolator can add some jitter to the signal being sent to the other side, but that signal is not the one that feeds the DAC. The USB data is received by a device on the other side, then either re-transmitted or converted to SPDIF/AES/I2S and transmitted

You can see this in both DDCs like the DI20HE:

And in USB isolators like an intona 7055-C or ifi iGalvanic:

All DDCs (or DACs with built in isolation) have another device on the other side to receive that signal then retransmit (and/or convert if needed).
As mentioned, USB does not carry a Main clock/Word Clock like SPDIF/AES/I2S do, and the data isn't even sent in a constant stream

 

[GoldenSound]

In this case DI-20 also have USB port isolated but adding ground loop redirector like ifi iDefender with external power supply helps.

It'd only help if the DI20 isn't properly isolated (which to my knowledge it is).
If it is properly isolated then a tool to cut or replace the 5v/gnd line wouldn't do anything

 

[sajunky]

A USB isolator must be bidirectional inherently. USB is a bidirectional protocol. And whether it is/isn't bidirectional would be unrelated to jitter.
It's true that the isolator can add some jitter to the signal being sent to the other side, but that signal is not the one that feeds the DAC. The USB data is received by a device on the other side, then either re-transmitted or converted to SPDIF/AES/I2S and transmitted

You can see this in both DDCs like the DI20HE:


And in USB isolators like an intona 7055-C or ifi iGalvanic:



All DDCs (or DACs with built in isolation) have another device on the other side to receive that signal then retransmit (and/or convert if needed).
As mentioned, USB does not carry a Main clock/Word Clock like SPDIF/AES/I2S do, and the data isn't even sent in a constant stream

I don't know what you are trying to argue. You always do in the way where focus is shifted forcing your way of thinking which not always follows discussion.

In this case you bring completely different class of devices that do isolation on the USB to USB link. It is a challenge, as it involve not only bidirectional data transmission, but also DC signalling which is a part of USB protocol. These DC signals must be passed over a barrier forth and back during the negotiation, handshaking or error reporting. This is usually not done properly, which makes a bridge not completely transparent. Isolators are more leaky, the Intona by example use a mix of a silicon and a transformer. The weakest part makes a leak. See a part on the PCB marked DA2303, it is transformer for RS-485 data transmission which you do consider inferior. However I don't want to talk about these devices, I try to avoid them in first place.

A point you have completely missed is that USB port isolation in DACs / DDCs do not need to be placed on the USB wires. The output of the USB receiver is I2S, why do not put isolator on I2S wires? It is much simpler but effective solution. It is how USB port isolation is made these days. I mentioned it in the original post (it was not understood correctly), it require to place USB receiver on the dirty side of the barrier.

Judging from the wrong response as explained above, there is a need for a further comment. In DI-20 there is an isolator on the I2S lines indeed (not USB wires), so we deal with data and clock lines, please don't spread a confusion. Similar to other brands. In a typical solution it is a forward clock, which means asynchronous data stream is broken on the isolator, clock is passing through jittered. In many devices it is not a problem, as everything is reclocked internally. Audio GD do not share this concept of reclocking, they try to avoid whenever is possible, so a clock line direction is reversed, maintaining asynchronous data delivery. However this solution is only implemented in DI-20 and R8/R7 2021 DAC version and later. The base model R2R-11mk2 use a feed-forward clock as many others.

 

[GoldenSound]

I don't know what you are trying to argue. You always do in the way where focus is shifted forcing your way of thinking which not always follows discussion.

In this case you bring completely different class of devices that do isolation on the USB to USB link. It is a challenge, as it involve not only bidirectional data transmission, but also DC signalling which is a part of USB protocol. These DC signals must be passed over a barrier forth and back during the negotiation, handshaking or error reporting. This is usually not done properly, which makes a bridge not completely transparent. Isolators are more leaky, the Intona by example use a mix of a silicon and a transformer. The weakest part makes a leak. See a part on the PCB marked DA2303, it is transformer for RS-485 data transmission which you do consider inferior. However I don't want to talk about these devices, I try to avoid them in first place.

A point you have completely missed is that USB port isolation in DACs / DDCs do not need to be placed on the USB wires. The output of the USB receiver is I2S, why do not put isolator on I2S wires? It is much simpler but effective solution. It is how USB port isolation is made these days. I mentioned it in the original post (it was not understood correctly), it require to place USB receiver on the dirty side of the barrier.

Judging from the wrong response as explained above, there is a need for a further comment. In DI-20 there is an isolator on the I2S lines indeed (not USB wires), so we deal with data and clock lines, please don't spread a confusion. Similar to other brands. In a typical solution it is a forward clock, which means asynchronous data stream is broken on the isolator, clock is passing through jittered. In many devices it is not a problem, as everything is reclocked internally. Audio GD do not share this concept of reclocking, they try to avoid whenever is possible, so a clock line direction is reversed, maintaining asynchronous data delivery. However this solution is only implemented in DI-20 and R8/R7 2021 DAC version and later. The base model R2R-11mk2 use a feed-forward clock as many others.

Why are your responses always so aggressive? No one is attacking you man.....

 

[sajunky]

Why are your responses always so aggressive? No one is attacking you man.....

I am tired explaining things which have no connection to my post, like in this case I had to oppose claims you made on the example of devices like Intona which work on different principles than isolators inside DDC / DACs. You didn't say anything wrong about these devices, but presented arguments didn't apply to the situation.

A fact that you brought these devices indicate that you was not aware how USB port isolators work inside DDC/DAC, which means you have learned a bit, but it is not reflected in your last response. It was always like that. What do you expect from a person you never give a credit?

 

[GoldenSound]

The internet is a frustrating place...

If you want to talk about how people are supposedly attacking you feel free.

Meanwhile the rest of us will continue with the topic we were actually discussing

 

[GoldenSound]

Audiowise are claiming that bypassing the USB clock lowers the overall noise and that has an impact on sound quality. They also claim they are filtering the noise, but since people are still using Wave cables with them, or the Opto-dx, not that well probably. There is still probably less than non isolated usb power lines out of a gaming pc.
It sounds like an Intona in front of the SRC-DX might be the way to go unless you have any other suggestion. I was really hoping to get rid of the cables and boxes on my desk.

I'm not sure what isolation or filtering the src-dx has internally. Though it's of course possible that it in itself would be generating some sort of noise due to circuitry that comes AFTER whatever filters out noise from the source.

This is one of the reasons I do noise tests on DDCs. Cause the noise on the output can indeed vary a fair bit depending on the device.

Assuming the src-dx doesn't have full galvanic isolation than an intona before it could definitely be of benefit though