Finally, a grain of sanity in this thread; exactly what I said myself in https://www.diyaudio.com/forums/the...wtorch-preamplifier-iii-3404.html#post6032210. But then of course, everybody talks, nobody is listening here.
Probably electronic noise of some kind. Mic CMC 6 body electronics are supposed to start rolling off at 26kHz. Didn't use any power conditioner at all, and some of the equipment ran on SMPS including the RME ADC.
No bit-depth reduction was used by me, could be some dither used internally in the ADC.
Jn, you keep talking in your posts about sampling at 44.1 kHz. The fact is that nobody does it that way these days.
Since the early '90s about 99% of audio A/D converters on the market are delta-sigma types with their front ends running in 2.5-6 MHz range. The final output gets decimated down/low pass filtered by built-in digital filters, so no matter what the output rate (44.1; 96; 192 or even DSD) the front part, including "gentle" band limiting filter, is always the same. And, there are no brick wall filters in front of A/D chips anymore.
From the standpoint of system response that really doesn't matter - the same function must occur before decimation. LPFs are also used in DAWs after many/all internal operations, to remove illegal values.
My complaint with this whole discussion is that it seems to try to cheat on the subject of illegal values. As if: "but what if I want to allow these illegal values".
All good fortune,
Chris
In the psaudio link you refer to, a Dr Kunchur is mentioned as the person who proved that our ears could hear differences in arrival times of 5 to 10 usec.
Howewer in the document where he published his findings, there was a severe mathematic calculation error. What he actually proved was that our auditory system has a minimum discrimination level of 0.7dB, exactly the attenuation of the sound traveling a difference of 5 usec in his test set up.
I confronted him with his error, and after some emails he admitted that his real knowledge was in a totally different discipline and that if he visited the Netherlands he would like to meet me, which did not happen.
But nevertheless, every now and then I come across this figure of 5 to 10 usec.
Hans
Fascinating. The examples given seem to be at a few 100Hz but even so looks like some could detect the difference in a couple of cycles, at least if I am reading the graph correctly.
And sorry I'm 3 pages behind today. Real world got in the way...
Cymbal was Zildjian K Custom Hybrid Limited Edition 14" Reversible Hi-Hat top.
Mic was Schoeps MK41 capsule with CMC 6 body (aka Schoeps CMC 641).
Mic and premap-to-ADC cables were custom manufactured Jam designed XLR cable wire.
Preamp was Grace Designs M101.
ADC and master clock source was RME ADI-2.
SPDIF from ADC into Focusrite Scarlett 6i6.
Recording software was Reaper 64-bit with project sample rate set to 192kHz (to avoid any unwanted resampling).
Hope that's enough info for you...
Brilliant, thanks! Data and all the measurement info. Perfect.
Nyquist denial??????
I would take great pleasure in providing you some new orifices, however, that would be counterproductive (I believe, do not prove me wrong).
The beautiful aspect of my analysis is..wait for it....
It is testable
What part of that do you not understand???
All that I speak of, all that I do, the primary goal is, testability.
If you have a problem with testability, tell us.
I believe you confuse me with someone else.
Jn
Jn, you keep talking in your posts about sampling at 44.1 kHz. The fact is that nobody does it that way these days.
Since the early '90s about 99% of audio A/D converters on the market are delta-sigma types with their front ends running in 2.5-6 MHz range. The final output gets decimated down/low pass filtered by built-in digital filters, so no matter what the output rate (44.1; 96; 192 or even DSD) the front part, including "gentle" band limiting filter, is always the same. And, there are no brick wall filters in front of A/D chips anymore.
My major concern is....if there are sources which exceed nyquist as a consequence of envelope modulation, that any filtering which impacts the envelope modulation splash removes information.
Jn
Ah, ok...If I say 'F=BIL' you might work out who I was talking about
The worst thing that can happen to one is, not being remembered.
Sigh..
Jn
Nyquist is not just "sampling twice the frequency".So let's recap ...
Exactly how many angels is it?
If you (and others) do not understand that, just ask.
Jn
I don't understand why anything has to be tested in 2020. Sampling has been established law since I was in diapers. To work right it requires bandlimiting. Bandlimiting has known, predictable effects. There is no other there there.
Much thanks, as always,
Chris
- Status
- Not open for further replies.