Is Digital Audio A Mature Science?

[iaval]

11 minutes ago, marce said:

Well I've been working in electronics for over 3 decades and we have known about the noise electronic systems generate since before then, so its nothing new... Plenty of info out there on noise mitigation and filtering as well, Henry Ott, Ralph Morrison, Keith Armstrong amongst others have written about it....

Some noises are so impossible to filter out. Not necessarily audio-related noises. RFI from switching supplies, home appliances, computers, phones, gadgets, ... . There's no easy mitigation there, even bandpass filter sometimes only exaggerate the issue. 

 

Also ripple on DC rails is quite a challenge to filter out, with one of the better ways to deal with it is capacitative multiplier, which however runs at abysmal efficiency, negating main efficiency benefits of switching power supplies over linear ones. 

[iaval]

10 minutes ago, marce said:

No they are not though... plenty of high precision, low noise design in the world of electronics, plenty of designs that are not affected by noise, what is so special about CONSUMER audio...

When dealing with very sensitive signal detectors, RF amplifiers and sharply directional antennas, everything becomes a noise. Devices used by HAM radio operators and radio astronomers operate at the very edge of signal and noise, close to what Johnson-Nyquist noise allows for a set bandwidth. It's not impossible to hear someone's PSU from 30km away on 430MHz band. 

When you need to work at -137dBm levels, there's no easy solution to filter all the junk signal around (your neighbour's car in standby, their LED lights, refridgerator, ...), if it's in your range of frequencies of interest. That is, without going at least 200km from civilization.

Every design has its set of advantages and drawbacks. 

 

[iaval]

1 hour ago, marce said:

I don't get the relationship to consumer audio

It's not directly related , that's why I said:

2 hours ago, iaval said:

Not necessarily audio-related noises.

 

I reacted to your comment on noise mitigation, because if there is any effective isolation for RF, and/or a way to clean AC ripple out of DC switching supply without significantly worsening its efficiency rating, I'd be interested to know how is that done properly. 

Dave Jones dedicated one video on his EEVBlog to removing power source ripple using two transistors, both of which however drops the circuit voltage by 1V, so a 12V clean supply would have to be 14V and would lose 15% efficiency on top of efficiency lost by for example LM2596

[iaval]

16 minutes ago, mansr said:

Distortion in the expensive one, either intentional or a result of boneheaded design choices like discrete resistor ladders or vacuum tubes.

Also size, power consumption, and the need for drivers which installs kernel extension that'd cause constant BSODs

[iaval]

Sometimes I'm so glad that I've given up on the hobby and made a choice to focus on music rather than on sound. 

 

Harmony, counterpoint, music form, unity, variety, and no longer soundstage, veil, digital glare, boomy, micro details, etc. 

[iaval]

5 minutes ago, marce said:

just to say LIGO used quite a few op-amps, so there is no reason not to use them...

And broad selection of Neve consoles are using hundreds of opamps and switching relays with no issue whatsoever. 

[iaval]

A lot of popular and classical music from 80s up until late 00s was mixed using puny $1 TI NE5532A on Neve 8108. I guess we're screwed then. 

[iaval]

The biggest problem with SQ, at least from my perspective, is that a lot of people are chasing it as the #1 priority, and are willing to sacrifice a lot for it. I don't mean necessarily money. But convenience is a factor too, as well as:

• power consumption
• heat dissipation
• longevity
• compatibility
• maintenance required
• durability
• weight
• extra features

There's just no compromise with some. I'm more than willing to trade 5 times lower power consumption of TLE2062IP vs JRC4556AD over extra 6dB of noise, worse THD+N into loads below 32Ohm, worse linearity, etc. because 2062IP still gets the job done reasonably well at 4.5mA quiescent over 22mA drained by 4556AD. It's not a compromise I'd hear, therefore a tradeoff over what's very noticeable (battery life) is definitely worth it. 

[iaval]

2 hours ago, sandyk said:

Measurements often don't coincide with what many prefer, which is very obvious considering the preferences with Vinyl vs. Digital and Solid State vs. Vacuum tube.

It's the interpretation of the measurements, not the measurements themselves.