Volume Control?

New to XMOS and XCore? Get started here.
MaximLiadov
XCore Addict
Posts: 130
Joined: Mon Apr 16, 2018 9:14 am

Post by MaximLiadov »

Dear Gordon,
That's not the only crazy thing I heard. Just try to do not fall down. I heard some "musical" USB cables with golden connectors affect on sound quality dramatically (sound stage is more wide, trebles more crystal), and even software media player (especially Aurvana for $70) with standard API and different OSes even being bit accurate sending some unknown fluids through a standard USB bus that affect much on sound quality. Modern science just cannot explain this, because engineering is helpless when it comes to music stuff. I am facing with this everyday. It's interesting what do you think, as you are closely standing behind physical part of USB interface and have proper equipment. Is that possible somehow? Did you see any difference on hardware level with "better" USB cables or with OS changing? Bit perfection really sucks and means nothing? Audiophilic approach rules? By your personal opinion.


User avatar
Wavelength
Experienced Member
Posts: 76
Joined: Mon Sep 24, 2018 9:56 pm

Post by Wavelength »

Maxim,

It's not the gold it's the configuration of the cables. With Full Speed you could basically make a cable out of zip cord. But the problem with High Speed is the changing of direction and the geometry of the cable can make all the differences in the world. I have a Tektronix compliance system and USB analyzer with HS differential probe that I can insert into my testing. I have tested probably 80 USB cables, some over $1K and they suck. The trick with HS USB is that the data lines cannot be shielded, but the problem with VBUS/GND interfering with the signal is then a problem. This has led to design of dual cables with the VBUS/GND in one and data pair in another. Some of these are then gathered together to look like one cable and others are not.

I can see errors on the TEK screen with HS running say a sine wave into a DAC at 1Khz. Changing the sample rate from 44.1K -> 384K you will see and increase of errors as the packets get larger.

Or if the cable has a lot of hangover (when the signal transmitted stays active and the receiver has to switch to transmit back, say the Feed Back packet IN request) then the HOST will error that packet received and that causes pops and clicks as the Async protocol will then fail.

Software: I did research for 4 years trying to figure out why the software effects the sound. I setup the following system:

BootCamp macOS MacBook Pro and Windows <--USB with TEK Analyzer-->DAC with breakout I2S

I would test with a flat file WAV is easier than AIFF at 24/88.2 and compare the data as seen on the TEK and the I2S in a perfect setup. Meaning the data was the same from File, USB and I2S without errors.

I did find a few applications that were not bit perfect, but in general they all were the same. BUT!!! None of them sounded the same. I repeated this test at the RMAF show in Denver (without all the testing equipment, but same system) and the results were the same.

So why do they sound different. I had a problem with my Mac in that one core was spinning. So I bought iStat Monitor plug in and found a device driver was just eating one of the cores at 100%. I removed it and all was good. I then tested a DAC I designed for another company running the suite of software through it, Bit Perfect, iTunes, Audirvana, Decibel, Roon, Pure Music.... and I noticed on iStat Monitor the amount of usage each program had. The correlation between the usage and sound was identical to their performance. Audirvana was the best, then Roon/Decibel and so forth. The switching noise associated with the processor usage was invading the stereo through both the mains and over cables and even from the monitors.

Everything makes a difference!

Thanks,
Gordon
Wavelength Audio, ltd.
MaximLiadov
XCore Addict
Posts: 130
Joined: Mon Apr 16, 2018 9:14 am

Post by MaximLiadov »

Gordon, thank you very much! So seems to be we have no such a testing tool (even very expensive one) that could prove or disprove our subjective expertise on 100%. Even if Tektronix shows the same numbers, you still hear some difference in music. Very interesting results with 80 USB cables. Seems to be USB interface is the only modern digital interface that could bring you errors or some difference in signal just with 0.5 meter cable and silly 480 Mbps speed and even could do not work at all with cables more than 1.8 meters. Maybe we need to switch to Ethernet (AVB or Dante or Ravenna)? Merging Technologies' Anubis looks very promising. It's pity that there is no any dedicated bus for audio. SPDIF bs, USB bs, not sure yet about 1Gb Ethernet.
User avatar
Wavelength
Experienced Member
Posts: 76
Joined: Mon Sep 24, 2018 9:56 pm

Post by Wavelength »

Maxim,
Let's face it Digital connectivity is flawed all around. Gigabit Ethernet is not full duplex and therefore it's a big problem, plus it does everything in nibbles (4 bits bi-directional). 10/100 was smart and was full duplex. SPDIF/AES is flawed, especially AES... really an XLR connector being 110ohms, really??? Common in what electrical world is that true. BNC is really the only one for SPDIF, RCA is total BS diameter over diameter. Dante is better than AVB if you can get them to answer any emails.

You can still do 24/192 on 10/100 Ethernet if you don't have a bunch of channels. Maybe force the Dante card into that and it may work better. Most of the big guns touring use Dante and have everything at the stage and just a control panel back at the sound stage.
Thanks,
Gordon
Wavelength Audio, ltd.
Post Reply