mansr Posted July 13, 2016 Share Posted July 13, 2016 Ever wondered what USB data errors actually sound like? I decided to find out. First I built the worst USB cable the world has ever seen: The data wires are about a foot of twisted mess. For extra badness, a variable capacitor is connected between D- and ground. With the capacitor at minimum, this assembly just barely transfers data without error. I connected an iFi Nano DAC via this monstrosity and recorded its outputs (both analogue and S/PDIF) using a Tascam UH-7000 while playing a 1 kHz sine tone at 96 kHz sample rate. While this was running, I slowly increased the capacitance. The recordings can be downloaded here: https://www.amazon.com/clouddrive/share/nwWu5edHZ71VX38KFU6PictAZxj2ck9geI8quk75EUS Part of the captured S/PDIF data looks like this: Note that the waveform is piecewise correct but riddled with skips and drop-outs. It sounds awful. In conclusion, if there are data errors on a USB connection, it will be obvious. Link to comment
The Computer Audiophile Posted July 13, 2016 Share Posted July 13, 2016 Ever wondered what USB data errors actually sound like? I decided to find out. First I built the worst USB cable the world has ever seen: [ATTACH=CONFIG]27662[/ATTACH] The data wires are about a foot of twisted mess. For extra badness, a variable capacitor is connected between D- and ground. With the capacitor at minimum, this assembly just barely transfers data without error. I connected an iFi Nano DAC via this monstrosity and recorded its outputs (both analogue and S/PDIF) using a Tascam UH-7000 while playing a 1 kHz sine tone at 96 kHz sample rate. While this was running, I slowly increased the capacitance. The recordings can be downloaded here: https://www.amazon.com/clouddrive/share/nwWu5edHZ71VX38KFU6PictAZxj2ck9geI8quk75EUS Part of the captured S/PDIF data looks like this: [ATTACH=CONFIG]27667[/ATTACH] Note that the waveform is piecewise correct but riddled with skips and drop-outs. It sounds awful. In conclusion, if there are data errors on a USB connection, it will be obvious. Very interesting mansr. Thanks for posting. Can you poke any holes in your test that may weaken your conclusion? I ask because usually the people who do the test know its weaknesses and are good at providing all the information for consumption. Thanks! Founder of Audiophile Style | My Audio Systems Link to comment
mansr Posted July 13, 2016 Author Share Posted July 13, 2016 Very interesting mansr. Thanks for posting. Can you poke any holes in your test that may weaken your conclusion? I ask because usually the people who do the test know its weaknesses and are good at providing all the information for consumption. Well, I'll admit that I don't know exactly why this "cable" causes errors. I don't have the equipment to measure that. Maybe some other cause would result in an error distribution with a different sound, although it's unlikely anything could cause large numbers of undetected errors. Another DAC might also react differently, perhaps trying to interpolate data for corrupt packets. Link to comment
wgscott Posted July 13, 2016 Share Posted July 13, 2016 Who cares about the data? What does it do to the sound–stage? But seriously, In conclusion, if there are data errors on a USB connection, it will be obvious. If there are data errors of this magnitude, it will be obvious. What if there were errors that were measurable but much more subtle (or can you remove them by decreasing the capacitance, or making other reproducible improvements)? When I clicked on this thread, with great anticipation, I was expecting it would say "look at all the deficiencies, yet it measures and sounds perfectly fine." My expectation bias has been thwarted. Link to comment
The Computer Audiophile Posted July 13, 2016 Share Posted July 13, 2016 Well, I'll admit that I don't know exactly why this "cable" causes errors. I don't have the equipment to measure that. Maybe some other cause would result in an error distribution with a different sound, although it's unlikely anything could cause large numbers of undetected errors. Another DAC might also react differently, perhaps trying to interpolate data for corrupt packets. Thanks for the follow up. I've often thought of doing similar things with other components, such as finding the worst old CD player with tons of jitter and comparing it to the sound I have now with a computer based system. Just to hear the differences and to see how far the industry has come. Founder of Audiophile Style | My Audio Systems Link to comment
whoozwah Posted July 13, 2016 Share Posted July 13, 2016 I wonder how common it is to hear a cable that's built to spec have data errors like this? It seems you had to build a cable specifically to fail to record the data. If I am anything, I am a music lover and a pragmatist. Link to comment
mansr Posted July 13, 2016 Author Share Posted July 13, 2016 Who cares about the data? What does it do to the sound–stage? But seriously, If there are data errors of this magnitude, it will be obvious. What if there were errors that were measurable but much more subtle (or can you remove them by decreasing the capacitance, or making other reproducible improvements)? The first few seconds of the recording are fine. When errors start showing up, it's as an occasional skip which is clearly audible. The thing is, as soon as a single bit is wrong, an entire packet is discarded, and this causes a huge error in the recovered data stream. Link to comment
alfe Posted July 13, 2016 Share Posted July 13, 2016 Well, I'll admit that I don't know exactly why this "cable" causes errors. I did some awful stuff to test USB but this put you straight in the hall of fame Link to comment
esldude Posted July 13, 2016 Share Posted July 13, 2016 I feel a missed opportunity. As in there are no problems only opportunities. A new USB interface with custom system matching in form of a variable capacitor would have sold like hotcakes. Adjustable to your gear as one size never fits all. Btw, how is image height over this connection? PS you are an evil man for things like this. ? Sent from my Nexus 6P using Computer Audiophile mobile app And always keep in mind: Cognitive biases, like seeing optical illusions are a sign of a normally functioning brain. We all have them, it’s nothing to be ashamed about, but it is something that affects our objective evaluation of reality. Link to comment
Abtr Posted July 13, 2016 Share Posted July 13, 2016 Pretty useless experiment.. USB normally transmits bits perfectly, so it takes a "horrible" USB cable to explore the limit of that ability. The problem of USB audio is not bit transmission, it is electrical noise and timing.. Current audio system Link to comment
alfe Posted July 13, 2016 Share Posted July 13, 2016 Pretty useless experiment.. USB normally transmits bits perfectly, so it takes a "horrible" USB cable to explore the limit of that ability. The problem of USB audio is not bit transmission, it is electrical noise and timing.. Just need a reclocker. Link to comment
esldude Posted July 13, 2016 Share Posted July 13, 2016 Just need a reclocker. And an LPS. Sent from my Nexus 6P using Computer Audiophile mobile app And always keep in mind: Cognitive biases, like seeing optical illusions are a sign of a normally functioning brain. We all have them, it’s nothing to be ashamed about, but it is something that affects our objective evaluation of reality. Link to comment
Abtr Posted July 13, 2016 Share Posted July 13, 2016 And an LPS. Or galvanic isolation. Current audio system Link to comment
Paul R Posted July 13, 2016 Share Posted July 13, 2016 Well- it gives a good example of what a sort of square wave sounds like in music transferred by USB, now don't it? :) Anyone who considers protocol unimportant has never dealt with a cat DAC. Robert A. Heinlein Link to comment
kumakuma Posted July 13, 2016 Share Posted July 13, 2016 Or galvanic isolation. AND galvanic isolation Sometimes it's like someone took a knife, baby Edgy and dull and cut a six inch valley Through the middle of my skull Link to comment
esldude Posted July 13, 2016 Share Posted July 13, 2016 AND galvanic isolation LOL Sent from my Nexus 6P using Computer Audiophile mobile app And always keep in mind: Cognitive biases, like seeing optical illusions are a sign of a normally functioning brain. We all have them, it’s nothing to be ashamed about, but it is something that affects our objective evaluation of reality. Link to comment
wgscott Posted July 13, 2016 Share Posted July 13, 2016 as soon as a single bit is wrong, an entire packet is discarded, and this causes a huge error in the recovered data stream. OK, that was the logical piece I was missing. Assuming I now understand, you are saying the error is quantized, and the signature you see (and hear) is a single bit (smallest possible) error? Link to comment
YashN Posted July 13, 2016 Share Posted July 13, 2016 In conclusion, if there are data errors on a USB connection, it will be obvious. But you can't conclude that all SQ-related issues with USB are of that kind in audiophile-land, especially with DACs which are susceptible to a variety of noises. Dedicated Line DSD/DXD | Audirvana+ | iFi iDSD Nano | SET Tube Amp | Totem Mites Surround: VLC | M-Audio FastTrack Pro | Mac Opt | Panasonic SA-HE100 | Logitech Z623 DIY: SET Tube Amp | Low-Noise Linear Regulated Power Supply | USB, Power, Speaker Cables | Speaker Stands | Acoustic Panels Link to comment
mansr Posted July 13, 2016 Author Share Posted July 13, 2016 OK, that was the logical piece I was missing. Assuming I now understand, you are saying the error is quantized, and the signature you see (and hear) is a single bit (smallest possible) error? There's no way of telling how many bits were wrong in a dropped packet. In this test it isn't even possible to tell how many packets were dropped. I can test with a different signal that reveals more about the errors. Link to comment
Middy Posted July 13, 2016 Share Posted July 13, 2016 Does it come in other colours ..... But for your next trick mansr a good usb for the starter diy people.... 8^p Sent from my SM-G900F using Computer Audiophile mobile app Link to comment
mansr Posted July 14, 2016 Author Share Posted July 14, 2016 There's no way of telling how many bits were wrong in a dropped packet. In this test it isn't even possible to tell how many packets were dropped. I can test with a different signal that reveals more about the errors. To get a better picture of the errors, I used a test signal that simply increases by one with each sample. At 96 kHz this takes about 87 seconds. Dropped packets thus show up in the digital capture as a sample to sample difference greater than one. With the capacitor in the minimum setting, I got a total of 71 skips (the exact number is obviously random). Most of the skips were 12 samples long, some 60 samples, and a few were 13 or 61 samples. Increasing the capacitance at first made the skips more frequent. Then at some point, runs of zeros started appearing. This could perhaps be because errors are frequent enough to cause buffer underruns in the XMOS USB receiver. Link to comment
esldude Posted July 14, 2016 Share Posted July 14, 2016 Very nicely tested. And always keep in mind: Cognitive biases, like seeing optical illusions are a sign of a normally functioning brain. We all have them, it’s nothing to be ashamed about, but it is something that affects our objective evaluation of reality. Link to comment
Abtr Posted July 14, 2016 Share Posted July 14, 2016 Many scientific results are irrelevant.. Current audio system Link to comment
mansr Posted July 14, 2016 Author Share Posted July 14, 2016 Many scientific results are irrelevant.. Why are you commenting in this thread? Link to comment
Abtr Posted July 14, 2016 Share Posted July 14, 2016 Why are you commenting in this thread? Your're right, it's irrelevant. Current audio system Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now