• 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!

Sound - an interesting perspetive

pozz

Слава Україні
Forum Donor
Editor
Joined
May 21, 2019
Messages
4,036
Likes
6,827
It's a good piece. Thanks for posting.
 

Pluto

Addicted to Fun and Learning
Forum Donor
Joined
Sep 2, 2018
Messages
990
Likes
1,631
Location
Harrow, UK
Superb piece. Should be made compulsory.
 

JustAnandaDourEyedDude

Addicted to Fun and Learning
Joined
Apr 29, 2020
Messages
518
Likes
819
Location
USA
It's not a good idea to try and count the moving little dots.
 

JeffS7444

Major Contributor
Forum Donor
Joined
Jul 21, 2019
Messages
2,347
Likes
3,509
Very thought provoking, but should I feel a bit cheated that the soundtrack for the old movie "Krakatoa: East of Java" was lacking in dynamic range?

Because this is an audio forum, I couldn't resist checking out what would theoretically be possible to record in uncompressed form, and it seems that at least as far as the data format is concerned, pretty much anything possible on the planet could be captured with 32-bit audio with loads of dynamic range to spare:
https://www.sounddevices.com/32-bit-float-files-explained/
 

pozz

Слава Україні
Forum Donor
Editor
Joined
May 21, 2019
Messages
4,036
Likes
6,827
Very thought provoking, but should I feel a bit cheated that the soundtrack for the old movie "Krakatoa: East of Java" was lacking in dynamic range?

Because this is an audio forum, I couldn't resist checking out what would theoretically be possible to record in uncompressed form, and it seems that at least as far as the data format is concerned, pretty much anything possible on the planet could be captured with 32-bit audio with loads of dynamic range to spare:
https://www.sounddevices.com/32-bit-float-files-explained/
Real-world limitation is around 21bits because of analog noise. No getting around that for now.

If any of the members have a good understanding of superconductors or quantum computing I'd like to understand what kind of performance numbers we would see if the technologies were applied to audio electronics.
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Real-world limitation is around 21bits because of analog noise. No getting around that for now.

If any of the members have a good understanding of superconductors or quantum computing I'd like to understand what kind of performance numbers we would see if the technologies were applied to audio electronics.

While I have in the past worked with superconducting stuff for data converters, microwave filters, and computing, I am no expert. There are a lot of practical complications to deal with in bringing something like that to the consumers. Heck, there are a lot of complications bringing it to any application...

For RF systems, I have dealt with noise floors for narrow-band systems in the -140 to -160 dBm (mW referenced to 50 ohms) range. Audio is a piece of cake. ;)

You need to specify what metric is used for "performance". For data converters, ENOB (effective number of bits) is essentially SINAD. A perfect converter with no noise except quantization noise has SNR ~ 6N dB for N bits. The noise floor is about 9N dB assuming no other noise or distortion. If you "add up" (root-sum-square, RSS) all the quantization noise across the Nyquist bandwidth that 9N dB noise floor becomes about 6N dB of SNR.
 

pozz

Слава Україні
Forum Donor
Editor
Joined
May 21, 2019
Messages
4,036
Likes
6,827
While I have in the past worked with superconducting stuff for data converters, microwave filters, and computing, I am no expert. There are a lot of practical complications to deal with in bringing something like that to the consumers. Heck, there are a lot of complications bringing it to any application...

For RF systems, I have dealt with noise floors for narrow-band systems in the -140 to -160 dBm (mW referenced to 50 ohms) range. Audio is a piece of cake. ;)

You need to specify what metric is used for "performance". For data converters, ENOB (effective number of bits) is essentially SINAD. A perfect converter with no noise except quantization noise has SNR ~ 6N dB for N bits. The noise floor is about 9N dB assuming no other noise or distortion. If you "add up" (root-sum-square, RSS) all the quantization noise across the Nyquist bandwidth that 9N dB noise floor becomes about 6N dB of SNR.
I mostly had THD and SNR in mind.

So, if I'm reading you right, you achieved roughly -150dB SNR for a specific RF bandwidth only. Does that mean that the 21-or-so-bit noise limit that's generally referenced applies just to the audio band? And then, for a real world converter, quantization noise is for all intents and purposes infinitely reduceable but overtaken by Johnson (thermal) noise at low levels?
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
I mostly had THD and SNR in mind.

So, if I'm reading you right, you achieved roughly -150dB SNR for a specific RF bandwidth only. Does that mean that the 21-or-so-bit noise limit that's generally referenced applies just to the audio band? And then, for a real world converter, quantization noise is for all intents and purposes infinitely reduceable but overtaken by Johnson (thermal) noise at low levels?

Actually around 180-190 dB total dynamic range but there was lots of processing so not directly comparable to audio. And that was not using superconductors, though one of the systems did use some superconducting microwave filters. Frankly I am in no hurry to deal with that kind of range again; my current job is much more plebeian (though not without its challenges).

Based on quantization noise, 21 bits yields an SNR around 128 dB. IIRC, Johnson (thermal) noise is around -174 dBm in a 1 Hz bandwidth so if you take that to 20 kHz you end up around -131 dBm which blowing off units and such gets you close to the 21-bit "limit". I'd guess that is (sort-of) where it comes from; I have not tried to research it. Note the commonly-accepted noise floor in deep space is around -198 dBm (again 1 Hz bandwidth).

In the real world are all sorts of noise sources including thermal, shot, flicker, etc. etc. etc. Plus there is nonlinearity (distortion) to deal with. And matching thresholds and such as you work to achieve "zero" quantization noise by increasing resolution is a challenge. As Amir's measurements have shown achieving 120 dB dynamic range is quite a challenge though DACs seem to be doing it. Whole systems, meh, especially when you start including speakers.
 

pozz

Слава Україні
Forum Donor
Editor
Joined
May 21, 2019
Messages
4,036
Likes
6,827
@DonH56 With every post I believe you less and less about your being "no expert".
 

Hipper

Addicted to Fun and Learning
Joined
Jun 16, 2019
Messages
753
Likes
625
Location
Herts., England
Very thought provoking, but should I feel a bit cheated that the soundtrack for the old movie "Krakatoa: East of Java" was lacking in dynamic range?

Because this is an audio forum, I couldn't resist checking out what would theoretically be possible to record in uncompressed form, and it seems that at least as far as the data format is concerned, pretty much anything possible on the planet could be captured with 32-bit audio with loads of dynamic range to spare:
https://www.sounddevices.com/32-bit-float-files-explained/

Interesting what stereo will do as Krakatoa is of course west of Java!
 
Top Bottom