John Curl's Blowtorch preamplifier part III

Status
Not open for further replies.
Jump to Latest
#33,441
Hans Polak said:
Jan Didden once participated in a test with a large group while playing music with lots of content above 20 Khz. There was a switchable steep analogue LPF installed after the D/A to reduce BW to 18Khz when I remember correctly.
Outcome as he mentioned was felt as quite embarrassing because the audience preferred the 18Khz limited sound.
Jan, are you listening ?

Hans
Click to expand...

In case he isn't: it was a 15kHz brickwall filter; phaseshift like a spinning ballet dancer.
 
#33,443
Tournesol said:
hen 24X96, then it was possible to near forget the recording artefacts, compression, reduction in size ... just pushing the fader up. I can believe the piano is in the room, with all its size and weight.

We could argue during decades to understand if it is the added dynamic of the 24 bits, or the added speed (slopes) and trebles that the 96 bring, but it works.
Click to expand...

T., What you describe was common with most ADCs at least those from a decade ago or earlier. I have one from then that sounds pretty much the same at every sample rate (more like exactly the same, actually), but it was quite expensive for a 2-channel mastering ADC (it had a DAC in it too, but that was clocked differently from the ADC and I didn't like the sound of the dac). Therefore, seems like it may have been a technology sound quality issue more than a frequency response issue that you heard.
 
Last edited:
#33,444
Bill, these experiments were mainly done on cats AFAIK. There is a remarkable resemblance between the architecture of the feline and the human brain responsible for auditory processing, up to the level of the neocortex. Wernicke's area is present in a cat brain. Remarkable since it is this area that decodes for speech in humans.

Pickles would be a great name for a cat.
 
#33,448
jneutron said:
I found it odd that Lavry depicted a brickwall in his paper, but it attenuated 17Khz 50%.

Is that consistent with all brickwalls? I thought by definition, they remained flat out to close to 20K.

From his discussion, it almost looks like he says NRZ had lots of error energy above nyquist, and by implication, the brickwall removing that error energy reduces the final waveform by half..

Note I read between the lines for that last statement, am looking for a viable explanation.

jn
Click to expand...

It was just a bad filter choice in the article if you care about being flat to 20 kHz. You can check this by looking at any decent DAC datasheet since all modern parts have integrated digital filters. The TI PCM1792As filter, for example, is completely flat (+\- 0.00001 dB) until 0.454*Fs, which is conveniently 20 kHz with FS=44.1 kHz.
 
#33,449
vacuphile said:
Bill, these experiments were mainly done on cats AFAIK. There is a remarkable resemblance between the architecture of the feline and the human brain responsible for auditory processing, up to the level of the neocortex. Wernicke's area is present in a cat brain. Remarkable since it is this area that decodes for speech in humans.

Pickles would be a great name for a cat.
Click to expand...

My old cat could decode human speech, but only when it was dinner time 😛.
 
#33,452
chris719 said:
It was just a bad filter choice in the article if you care about being flat to 20 kHz. You can check this by looking at any decent DAC datasheet since all modern parts have integrated digital filters. The TI PCM1792As filter, for example, is completely flat (+\- 0.00001 dB) until 0.454*Fs, which is conveniently 20 kHz with FS=44.1 kHz.
Click to expand...

He clearly showed a brickwall that was flat to 20khz. That is why I was wondering.
That is why my conjecture that an NRZ tends to be attenuated as he showed.

Jn
 
#33,453
gpapag said:
Does making sine start/stop less abrupt (apply fade in/fade out) turns the sine file “legal” in terms of Nyquist limit?
Click to expand...

Depends. A sinc(wt) signal (also kind of fade in/fade out sine) has the spectra of exactly a box rect(f/w)/|w| (aka "sinc and rect are mathematically dual functions", or in EE terms, "an idealized low-pass filter, and its non causal impulse response are dual"). No infinite spectra in this case, of course.
 
#33,454
vacuphile said:
Most everybody who read my post understood perfectly, I guess. And also understood the context in which it was placed. Perhaps you might reread because "..., distortion?" means that you did not get the point.
Click to expand...

I understood perfectly well.

You've provided absolutely no data to prove the assertion that redbook maintains ITD interchannel with a multi frequency stereo signal pair, nor even a sine wave. The "ppm" level accuracy statement you offer as proof despite the fact that it does not prove the question.

You state that localization is limited to below 3.5kz despite the fact that human testing shows otherwise.

You show testing on an entirely different species as proof humans localize via zero crossing.

You patch a lot of assumptions together and claim success? Really?

And then in the next post you try to pull the "attack the person instead of the content???

Again, really? Are you so flat as to use that silly tactic? Did you really think attacking me was the way to stop asking actual technical questions?

As I previously said, several of the posters on this thread have chosen to make it a trash heap, and it appears contagious.

Please, stick to discussion.

Jn
 
Last edited:
#33,457
JN,

The point is that redbook allows for the reproduction of a perfect sine wave at frequencies most relevant for ITD and that there is a full continuum for these sine waves to have their t=0 at any godgiven point in time. Without any relationship to sampling rate.

As to the 3.5kHz: did you read Hans's post?

Lernen, mein Liebchen.
 
#33,459
vacuphile said:
JN,

The point is that redbook allows for the reproduction of a perfect sine wave at frequencies most relevant for ITD and that there is a full continuum for these sine waves to have their t=0 at any godgiven point in time. Without any relationship to sampling rate.

As to the 3.5kHz: did you read Hans's post?

Lernen, mein Liebchen.
Click to expand...
As I pointed out previously...
1. Nordmark showed discrimation out to 12khz with synthetic jitter stimulus, and out to 5khz without synthetic jitter. Simply stating previous data is wrong without any semblance of proof doesn't float my boat. Also, other researchers made it out to 5khz.

2. Since zero crossing has been claimed using data from another species, what do cats say w/r to redbook?😀 But seriously, several neurosci guys did hypothesize zero crossings, but never tried to detail the fact that we can localize (with additional sound in play to mess up the zero crossings) also present. As I said, I wondered about the cochlea LP filtering as a possibility. Also, some neuro guys claimed rectification.

3. While I am a firm believer in linear theory, superposition of many different frequency signals so close to the nyquist limit with an NRZ envelope caused by an S/H, plopped into a brickwall of 20 kHz, that concept indeed needs proof via test. As in, I believe extension of a sample at one value by NRZ is not really a linear function. And, when I say close to nyquist limit, I mean within an order of magnitude or so. I'm a test guy at heart, and always want resolution and accuracy of measurement at least that far beyond desire.

Oh, almost forgot...the most important...

4. Thank you for not continuing along the line of "trashing the individual". It is most kind of you.

I have indeed stooped to condescension when someone needs it (like Tt and markw who deserve it no matter what they say...oops, wrong playbook..my bad.) But, I prefer civility. Too many are attackin others based on their name, not their message. Almost like a dodgeball game.

Cheers, jn
 
Last edited:
Status
Not open for further replies.
Top Bottom