-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Why is everyone using USB/XMOS. Why not PCIE SPDIF Out?
- Thread starter cheapsoundguy
- Start date
D
Deleted member 23982
Guest
Well the thing is, as it seems to me, that people here way to easly resort on measurements to "deny" stuff, which is (i think) a waaay more "dangerous" path instead of trying/listening yourself to fool your mind, also blind testing has some flaws in my opinion
tho, you are right there are people / forums who actually understand most of it, like that cables "really" matter, or even changing capacitors to another brand changes sound, as well as other opamps tho i really cant tell you what exactly is changing, less noise (beside everyone saying its "not audible) seems to be it most of the time, maybe you cant hear the noise "in itself" but its influence... atleast thats my guess
for me it was just eye opening to get studio monitors and tested my way up, and thats a thing i think does the most with "not hearing stuff", a bad setup which masks the "lesser" issues, like a weak link you have to fix first before "really" hearing more stuff (and also speakers who color the sound to begin with) , otherwise i cant explain it myself "why" MANY people hear exactly the things i do and many dont (and i dont think either is lying in what they truly hear it was hard to grasp this myself first too)
one example would be why many dacs sound so different (tho some say they dont...) , the "test-tones" are the same beside different noise floor (which -should- be inaudible)
but i just start another discussion about things "cant be possible" so just read my opinion and -please- make your own
i just searched myself again and there was this reddit post i found: which shows improved noise floor
also i readed many times that the jitterbug has no or nearly no effect, so i wouldnt say perhaps that "any" of this devices dont work because of the jitterbug measurement (would love to see measurements of the "other" devices here tho!)
beside this, i encountered a (new, i think) one which is this one: https://www.kitsunehifi.com/product/holo-audio-titanis-active-usb-processor/
which seems to be a very fairly priced reclocker with a bigger impact than the isilencer+ (from what i readed) and i will probably give it a go
- Joined
- Feb 23, 2016
- Messages
- 20,696
- Likes
- 37,432
That whole thing is so wrong in so many ways it would take 10 times the typing to set it straight. Brandolini's law invoked.Well the thing is, as it seems to me, that people here way to easly resort on measurements to "deny" stuff, which is (i think) a waaay more "dangerous" path instead of trying/listening yourself to fool your mind, also blind testing has some flaws in my opinion
tho, you are right there are people / forums who actually understand most of it, like that cables "really" matter, or even changing capacitors to another brand changes sound, as well as other opamps tho i really cant tell you what exactly is changing, less noise (beside everyone saying its "not audible) seems to be it most of the time, maybe you cant hear the noise "in itself" but its influence... atleast thats my guess
for me it was just eye opening to get studio monitors and tested my way up, and thats a thing i think does the most with "not hearing stuff", a bad setup which masks the "lesser" issues, like a weak link you have to fix first before "really" hearing more stuff (and also speakers who color the sound to begin with) , otherwise i cant explain it myself "why" MANY people hear exactly the things i do and many dont (and i dont think either is lying in what they truly hear it was hard to grasp this myself first too)
one example would be why many dacs sound so different (tho some say they dont...) , the "test-tones" are the same beside different noise floor (which -should- be inaudible)
but i just start another discussion about things "cant be possible" so just read my opinion and -please- make your own
i just searched myself again and there was this reddit post i found: which shows improved noise floor
also i readed many times that the jitterbug has no or nearly no effect, so i wouldnt say perhaps that "any" of this devices dont work because of the jitterbug measurement (would love to see measurements of the "other" devices here tho!)
beside this, i encountered a (new, i think) one which is this one: https://www.kitsunehifi.com/product/holo-audio-titanis-active-usb-processor/
which seems to be a very fairly priced reclocker with a bigger impact than the isilencer+ (from what i readed) and i will probably give it a go
Trell
Major Contributor
- Joined
- May 13, 2021
- Messages
- 2,752
- Likes
- 3,285
I would say that it's more like we reject anecdote when used to claim stuff.
So many claims...so little evidence.
Dismissing controlled testing isn't going get traction here. It will definitely go over better on those other forums you mention where they understand cables and stuff better.
As of UAC3 - excellent presentation by the author of linux UAC3 support https://archive.fosdem.org/2019/schedule/event/linux_and_usb_audio_class_3/ https://archive.fosdem.org/2019/sch...Linux_and_USB_Audio_Class_3___FOSDEM_2019.pdf
tvrgeek
Major Contributor
All of these physical layers are implemented as unreliable layer two. So, all are susceptible to an error. So be it.
USB is bi-directional so the host can ask what is plugged in. I do not believe TOSLINK or Coax is.
My all-in-one Windows PC I use for a server has only USB and no internal slot.
How well any of these work has been demonstrated my Amirm. He has identified some equipment with identical performance on any, and some where one was better than the other, so it goes back to implementation, not technology.
Several companies, Schiit being one, have put a lot of effort into their USB interface.
From my career in the HPC industry, I love glass. Great BW, no ground loops, and pulling a SM fiber pair is a lot easier than an old Buss and Tag cable.
If I want to complain, it is the horrible mix and unreliability of small USB and worse HDMI connectors. Some engineers seriously need a dope-slap.
I read that page on the HOLO. Sounds like BS to me. Galvanic isolators are active devices. LTM2884 or AMDU40160 series quite common. They are transformers, but have drivers and receivers on either side as well as the DC-DC converter. Of course, any effect depends on if you have a problem or not. I ordered an LT board to use on my USB scope and I will test it on my music server, but it is the scope that I want to isolate for safety reasons. I have some cheap 40160 boards. I could measure the difference in a loopback around my Scarlett. But the differences were in the -130 dB range and so irrelevant.
USB is bi-directional so the host can ask what is plugged in. I do not believe TOSLINK or Coax is.
My all-in-one Windows PC I use for a server has only USB and no internal slot.
How well any of these work has been demonstrated my Amirm. He has identified some equipment with identical performance on any, and some where one was better than the other, so it goes back to implementation, not technology.
Several companies, Schiit being one, have put a lot of effort into their USB interface.
From my career in the HPC industry, I love glass. Great BW, no ground loops, and pulling a SM fiber pair is a lot easier than an old Buss and Tag cable.
If I want to complain, it is the horrible mix and unreliability of small USB and worse HDMI connectors. Some engineers seriously need a dope-slap.
Well the thing is, as it seems to me, that people here way to easly resort on measurements to "deny" stuff, which is (i think) a waaay more "dangerous" path instead of trying/listening yourself to fool your mind, also blind testing has some flaws in my opinion
tho, you are right there are people / forums who actually understand most of it, like that cables "really" matter, or even changing capacitors to another brand changes sound, as well as other opamps tho i really cant tell you what exactly is changing, less noise (beside everyone saying its "not audible) seems to be it most of the time, maybe you cant hear the noise "in itself" but its influence... atleast thats my guess
for me it was just eye opening to get studio monitors and tested my way up, and thats a thing i think does the most with "not hearing stuff", a bad setup which masks the "lesser" issues, like a weak link you have to fix first before "really" hearing more stuff (and also speakers who color the sound to begin with) , otherwise i cant explain it myself "why" MANY people hear exactly the things i do and many dont (and i dont think either is lying in what they truly hear it was hard to grasp this myself first too)
one example would be why many dacs sound so different (tho some say they dont...) , the "test-tones" are the same beside different noise floor (which -should- be inaudible)
but i just start another discussion about things "cant be possible" so just read my opinion and -please- make your own
i just searched myself again and there was this reddit post i found: which shows improved noise floor
also i readed many times that the jitterbug has no or nearly no effect, so i wouldnt say perhaps that "any" of this devices dont work because of the jitterbug measurement (would love to see measurements of the "other" devices here tho!)
beside this, i encountered a (new, i think) one which is this one: https://www.kitsunehifi.com/product/holo-audio-titanis-active-usb-processor/
which seems to be a very fairly priced reclocker with a bigger impact than the isilencer+ (from what i readed) and i will probably give it a go
I read that page on the HOLO. Sounds like BS to me. Galvanic isolators are active devices. LTM2884 or AMDU40160 series quite common. They are transformers, but have drivers and receivers on either side as well as the DC-DC converter. Of course, any effect depends on if you have a problem or not. I ordered an LT board to use on my USB scope and I will test it on my music server, but it is the scope that I want to isolate for safety reasons. I have some cheap 40160 boards. I could measure the difference in a loopback around my Scarlett. But the differences were in the -130 dB range and so irrelevant.
I haven't seen any AVRs that have lossless multichannel USB input capability. (They may exist, I just haven 't looked hard for them). S/PDIF is limited to two channels of lossless audio. So it's HDMI for me.
radix
Major Contributor
- Joined
- Aug 1, 2021
- Messages
- 1,397
- Likes
- 1,334
AES-derived protocols (aes, spdif coax or toslink) are all unidirectional protocols. That's why when you play something at an unsupported rate, it sounds like crap instead of re-negotiating. aes also has a ton of extensions, so not every transmitter and receiver are compatible over all bits depths and data rates. AFAIK, no digital audio interface uses retransmission -- it needs to be real-time. Streaming protocols are different and often use retransmission or sometimes FEC.
USB can use several modes to transfer data to a DAC. Synchronous is maybe not used anymore. It was used in cheaper products. The DAC takes the timing from USB and its not great. Isochronous lets the DAC use its internal clock, but if packets arrive out of time the DAC still needs to play some tricks or live with some jitter. Isochronous with asynchronous feedback lets the DAC adjust the USB rate to match their internal clock and this gives the best sound. I'm not sure, but I suspect most all good DACs are using asynchronous feedback now.
I'm not a USB digital audio implementor, so that description likely skims over a lot of the details.
So, if you have a DAC using one of the older transfer methods, a de-jittering buffer on the line could help.
Trell
Major Contributor
- Joined
- May 13, 2021
- Messages
- 2,752
- Likes
- 3,285
There is also ADAT that can have 8 channels 24/48 but I don't know of any receiver that has that support. RME supports both ADAT and S/PDIF on their optical input/output.
- Joined
- Feb 23, 2016
- Messages
- 20,696
- Likes
- 37,432
Schitt had to work on their usb interface because until unmasked here their usb isolation was unusually schitty.All of these physical layers are implemented as unreliable layer two. So, all are susceptible to an error. So be it.
USB is bi-directional so the host can ask what is plugged in. I do not believe TOSLINK or Coax is.
My all-in-one Windows PC I use for a server has only USB and no internal slot.
How well any of these work has been demonstrated my Amirm. He has identified some equipment with identical performance on any, and some where one was better than the other, so it goes back to implementation, not technology.
Several companies, Schiit being one, have put a lot of effort into their USB interface.
From my career in the HPC industry, I love glass. Great BW, no ground loops, and pulling a SM fiber pair is a lot easier than an old Buss and Tag cable.
If I want to complain, it is the horrible mix and unreliability of small USB and worse HDMI connectors. Some engineers seriously need a dope-slap.
I read that page on the HOLO. Sounds like BS to me. Galvanic isolators are active devices. LTM2884 or AMDU40160 series quite common. They are transformers, but have drivers and receivers on either side as well as the DC-DC converter. Of course, any effect depends on if you have a problem or not. I ordered an LT board to use on my USB scope and I will test it on my music server, but it is the scope that I want to isolate for safety reasons. I have some cheap 40160 boards. I could measure the difference in a loopback around my Scarlett. But the differences were in the -130 dB range and so irrelevant.
He's not entirely wrong about USB being an iffy interface for audio. It's not great for a lot of things, actually, but given enough time and money it has been beaten into submission. Probably the least troublesome USB audio interfaces are from RME, and they have their own custom implementation in an FPGA and their own custom drivers. If it were so easy and great, there wouldn't be so few choices for quality USB to I2S bridge ICs in 2022 and there wouldn't be forums plagued with people that have clicks, pops, etc. The extra layer of host controller driver seems to make a difference here in Windows especially.
I realize the problems with USB audio devices are frequently other issues with the host system, but the fact is that the onboard Intel or Realtek codecs rarely have any issue, while USB devices do.
It works fine in 98% of cases now, but the whole thing is a bit kludgy when you read the UAC and UAC2 specs.
audio2design
Major Contributor
- Joined
- Nov 29, 2020
- Messages
- 1,769
- Likes
- 1,830
"Measurements"! Everyone has to drink.
I gotta say, this guy isn't crazy. I've had the same experience and with a D10 no less. He may have even got his power cord modification from my post on the original thread.
With Xmos right implemented you gain :
1. Asyncronuos = less jitter , better than AES or spdif.
2. Protection against 8 kHz USB packet noise. This seems to be very important and a big advantage.
3. A good 24 bit digital volume regulation can be implemented with dithering ( Converting 16 bit source material to xmos 32 bit - to 24 bit with dithering. )
Xmos is used widely nowadays in A/D-D/A recording interfaces like Audient and Motu.
This chip is a good one.
tvrgeek
Major Contributor
So the world is better! I got a hint from John Sebar (JDS) that they did spend a lot of time on their USB. Is that why the Atom sounds better?
I was investigating isolators, picked up the LT demo board. On the scope, it does filter a huge amount of trash. As I have not build a super clean 8V supply to feed it, I have not tested to see if it has an effect on the output of a USB powered DAC. ( like my Focusrite) I can see why the better DACS are all line powered. I can also see why the LT transformers are not SOP. They are very expensive. Retail singles $40, so I bet $10 to an OEM. That would bust the budget for about half of them out there. Then add well done surge suppression and, well, so much for affordability. I would not want to be in the volume consumer product design business.
I see one camp saying proper implementation of the XMOS chip works, and the other camp going to FPLAs to reduce dependency on one source. Again, it is most likely the execution, not the technology.
With a good ( not necessary expensive ) USB bridge you can make your laptop sound better than a cd-player
I use a ”sky song” 90 dollar bridge with separate clocks for 44,1 kHz and 48 kHz . And yes - driving the clocks LT1963 regulator with a battery brings a slightly better sound. Compared with an average spdif output from a regular CD player to a dac, this option is often better sounding. This was not the case 10 years ago.
I use a ”sky song” 90 dollar bridge with separate clocks for 44,1 kHz and 48 kHz . And yes - driving the clocks LT1963 regulator with a battery brings a slightly better sound. Compared with an average spdif output from a regular CD player to a dac, this option is often better sounding. This was not the case 10 years ago.
Last edited:
Better as opposed to what, and under what circumstances? Given the same source (i.e. a CD player's digital out vs a rip of that cd played by a PC then output via the bridge, all at same sampling rate and bit depth) this is only possible if the digital data produced would be different. Or something is very wrong with the clock. So where does the difference occur?
I won't even bother asking about your controls. When controls are absent, this is the sort of stupid stuff people believe.With a good ( not necessary expensive ) USB bridge you can make your laptop sound better than a cd-player
I use a ”sky song” 90 dollar bridge with separate clocks for 44,1 kHz and 48 kHz . And yes - driving the clocks LT1963 regulator with a battery brings a slightly better sound. Compared with an average spdif output from a regular CD player to a dac, this option is often better sounding. This was not the case 10 years ago.
tvrgeek
Major Contributor
Best I can tell is this is some box to convert USB to PCM coax?
Why not just get a half decent DAC?
Why not just get a half decent DAC?
Similar threads
