Because you can't see the noise separate from the music?
-
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!
Is Digital Audio Transmission Analog? [video]
- Thread starter amirm
- Start date
tuga
Major Contributor
Isn't there a clever piece of software that can do a null-test of the waveform (not the data)?
Perhaps - I'm not sure of it's capabilities.
However it is not necessary in this case (probably in any case). If you can see the noise with no music playing, it will be there with the music when it plays. Similarly - if it is not there, then it's not there with or without music.
OP
- Thread Starter
- #564
How many times did you repeat it? You need to run at least 10 times and get 9 answers right.
The easiest test here is to play content on a streamer and connect and disconnect the Ethernet cable. Due to buffering, playback should continue uninterrupted. When I have done this, there is absolutely no difference when Ethernet cable is plugged and unplugged. I have even shown this in measurements:
I'd say if you hold your ear up close to your tweeter you will perform a pretty meaningful null test...
tuga
Major Contributor
What if the noise is only present with music?
In that case I'd disable any power saving feature to establish what the "active" state is like while not playing music.
And if you stop playing music and whatever noise is still present, you know the system is flawed somewhere, and you get more methodical as you eliminate components one by one to establish the source of noise.
It isn't. The noise arrives as soon as the source is connected. If it is in the output then it is in the output. Music playing won't suddenly make it appear. You *might* need to play a silent file in case the output of the dac shuts down with no input signal I guess.
However if you are thinking of something like intermodulation distortion - that will be *below* the level of any modulating frequencies - so if you can't see the noise, you also won't see any IMD products at a lower level.
tuga
Major Contributor
But let's imagine that noise is only present when the data is sent, and is absent when idle?
In some cases, the output is shorted when nothing is playing. That way, you will not measure noise in an active system.
Hence my suggestion of a silent file.
But in any case - this is known. OP has measured common mode noise on the ground plane coming in from the Ethernet - he knows (or can easily discover) if this is present only when there is data on the cable, or if it is simply common mode noise coming from the activity of the upstream device regardless of whether data is being transferred.
It is also easy to have a probe on the ground plane to check if the noise is present or not, then check the analogue output of the system to see if that noise gets through.
KSTR
Major Contributor
Well, there are many ways incoming RF noise (notably CM mode) can be distributed within a device and there are many ways how the RF ingress can manifest itself in the final analog output signal. For example. at RF we will see tank circuits from wiring/routing L and C, and additional capacitance at semiconductor junctions which often varies with the momentary voltage across the transistor/FET. Now, even with steady-state constant-level RF which would demodulate to (almost) pure DC only, that DC may vary still with actual audio voltage because resonant frequencies of tank circuits will wander (as C changes), applying varying RF gain before demodulation which lead to varying DC following the audio signal, sitting on top of it. If that relationship is nonlinear, distortion is introduced by the voltage transfer function that modulates itself with incoming signal "bias". Thus, no signal --> no artifacts.
Therefore, I think it is dangerous to blindly assume all possible artifacts will show up fully with zero signal.
I do agree, though, that most of the disturbance should be already be visible/audible with zero signal when we try to look at the disturbance difference/delta between two cables or whatever (ferrites, CM chokes, isolators, etc). The assumption of steady-state constant level RF ingress most likely does not hold outside an EMC lab.
KSTR
Major Contributor
Yep, zero signal must be always dithered silence. Some DACs need several bits toggling to fully unmute and I've found dithering to 19 or 20 bits reliably avoids muting/dimming.
tuga
Major Contributor
But if noise were to modulate the signal, wouldn't it be best to use white noise (or something) as a test signal, something with a high amplitude level? (my knowledge is limited and I understand that this may be a silly question)
The problem with any wide band test signal (music or white noise), is you cannot (easily) distinguish the impact of the noise from the signal. If the noise is going to modulate the signal, this can be checked with a single frequency sine wave.
On that note, timed pulses would be more audible, no?
at one or two second intervals.
KSTR
Major Contributor
Depends on how you want to analyze (or listen to) the test signal. Noise or very noise-like music signal (Death Metal, anyone?) is better suited to subtractive analysis that listening to the direct signal. If two cables "produce" (give rise to) different RF-induced artifacts, especially when the artifacts from one cable are allegedly significantly stronger, these can be isolated via subtraction of the recorded waveforms. Still it is harder or even impossible to identify correlated distortions/artifacts when the base signal does not have a clean structure (repetitive and/or with self-similarities) and is noise-like itself.
For direct assessment, sine signal is usually best both for listening (use bass/mid frequencies) and for FFT spectrum/THD measurements.
So use a tone.
We can, quite easily, find the differences in the RF induced artifacts which lay almost always high above audible spectrum. And we can also find the artifacts inside the audible spectrum if we speak about link level cables due to differences in signal return impedances.
Attachments
Just a silly question: is this thread now about how ultrasonic noise on an Ethernet connection may impact music payload bit integrity and hence sound quality?
Similar threads
