moon-audio article saying that for shorter than 10 feet, RCA cables have a "stronger signal" than balanced cables. True?

[theREALdotnet]

Signal strength (or attenuation) is a not a concern with analog audio interconnects. What matters are CMRR and EMI/RFI and, to a much lesser extent, frequency response (i.e. HF roll-off due to capacitance).

 

[xaxxon]

difference in SNR is just due to 6 dB level difference.

isn't that useful, though?

(what follows is newb)

Doesn't that mean that the buffer stage needs to do less work which then helps maintain low noise through the "normal" amp? On the recent ncorex stuff, wasn't that a point of contention? And the benchmark ahb2 was measured with no buffer stage "pre-amplification"?

 

[restorer-john]

Stronger signal means better s/n ratio.
And i not see how rca can have bettet s/n ratio than balanced, if there is not some poor electronic involved. Balanced was invented to have better s/n. And it works. Yes on short distances rca can be nearly the same as balanced. But if rca is better its poor electronic, not the balanced connection.

It doesn't matter how high you lift the signal in order to claim a better S/N ratio, if the noise itself is greater in the first place. The noisefloor is the defining number.

A device outputting 40V with a residual of 20uV can claim a S/N of 126dB.
Another device outputting 5V with a 2.5uV residual can claim the same 126dB S/N.

Which device has the lowest noise floor?

 

[sam_adams]

"The theory underlying balanced interfaces seems to be misunderstood . . ." - Bill Whitlock

Bill Whitlock - A New Balanced Audio Input Circuit For Maximum Common-Mode Rejection In Real-World Environments

Bill Whitlock - Design of High-Performance Balanced Audio Interfaces

 

[JSmith]

Balanced vs. Unbalanced Audio Cables: Everything You Need to Know

Are Balanced Audio Cables Better for Your Audio System? Cables matter. They’re an important part of the equation if you are trying to make your system as clean-sounding as possible. Not all cables are created equal and are oftentimes an oversight in improving the sound quality and clarity of...

www.moon-audio.com

JSmith

 

[Chrispy]

While it would be nice if all audio gear was balanced/differential and used suitable connectors.....it didn't happen. Not that big a deal. But Moon for information? Let alone boxcast?

 

[Pursuit]

Handwaving:

• With two drivers (+ and -), a differential signal (XLR) is twice a single-ended signal's amplitude so a gain of 6 dB.
• The two drivers each add noise, but uncorrelated with each other, so do not add linearly but RSS (total noise = sqrt(N+^2 + N-^2), a noise increase of 3 dB.
• Thus you have a net SNR improvement of 3 dB.

Some designs cut back on signal levels or attenuate the input of a differential signal so you may not always see an advantage. Balanced also offers common-mode noise rejection and the ability to isolate the shield from the signal path, providing greater rejection of external noise, and making it easier to break a ground loop.

But, a single-ended design can have a lower noise floor, since there is only one driver and in any event noise floor and signal amplitude can be designed differently for different circuits, balanced or single-ended, so you cannot really draw a general conclusion about one or the other having a lower noise floor. A lot of high dynamic range (120+ dB) RF gear is single-ended, though most serial data links (Ethernet, PCIe, SATA, SAS, USB, etc.) are differential.

HTH - Don

Hi, I am just thinking a best but worst scenario, that is, the two drivers are truly identical and they produce noises at the same time. Then, total noise = sqrt((N+ + N-)^2), so 6db of noise, and total floor is also doubled. Does it make sense?

 

[Cbdb2]

This noise is random. Take 2 identical devices, transistors or even resistors and there noise will be uncorelated (the 2 noise signals will be different at every instant in time) and thus not add to 6db.

 

[fpitas]

This noise is random. Take 2 identical devices, transistors or even resistors and there noise will be uncorelated (the 2 noise signals will be different at every instant in time) and thus not add to 6db.

Correct. Uncorrelated noise adds 3dB. One way to think of it is that you have twice the noise power.

 

[DonH56]

Hi, I am just thinking a best but worst scenario, that is, the two drivers are truly identical and they produce noises at the same time. Then, total noise = sqrt((N+ + N-)^2), so 6db of noise, and total floor is also doubled. Does it make sense?

Answer is what they said:

This noise is random. Take 2 identical devices, transistors or even resistors and there noise will be uncorelated (the 2 noise signals will be different at every instant in time) and thus not add to 6db.

Correct. Uncorrelated noise adds 3dB. One way to think of it is that you have twice the noise power.

Noise is by definition random so even if you have two "identical" amplifiers their noise output will be different. If you look in real time with enough bandwidth (using an oscilloscope or other instantaneous capture device) there will be times when both outputs randomly align and noise is doubled (6 dB in voltage), and other times they are exact opposites and cancel (so zero noise). You cannot predict when either event happens. That is why noise is usually calculated statistically, using probability functions and such, with SNR calculations using RSS/RMS or average values. There are cases when instantaneous noise matters but I have not really seen that in audio -- speakers and our ears integrate the noise anyway.

My 0.000001 cent (microcent) - Don

 

[Pursuit]

Thanks! How about floor? Are they positive and negative and get doubled or of the same direction and get cancelled? If the floors add up, I would say the balanced design is just chasing that 1-2 db marginal gain.

 

[DonH56]

Thanks! How about floor? Are they positive and negative and get doubled or of the same direction and get cancelled? If the floors add up, I would say the balanced design is just chasing that 1-2 db marginal gain.

They are random so the noise floor will increase by about 3 dB (sqrt(2) times). Read the posts above again. Signal level increases by 6 dB (2x), thus a net gain in SNR of 3 dB.

The reason to use balanced interconnects, or more accurately differential signals, isn't really about a 3 dB increase in SNR. It is about suppressing common-mode noise applied from other sources (e.g. external to the signal path) and providing a shield independent of the signal return path for better shielding (noise rejection) and the ability to break a ground loop. So about better isolation from other (usually external) noise, like EMI/RFI and ground loops, not its intrinsic (self noise) SNR improvement.

See if this helps: https://www.audiosciencereview.com/...d-balanced-unbalanced-and-all-that-jazz.1352/