Of course not. Although Cestrian has done A/B testing with two identical dacs and reported some results in Marcel's RTZ dac thread.
Also, I don't see support for all of the claims made by other people in this thread to the effect that measurements prove its all impossible. Mostly they are just insisting they are right without providing any published studies in support of their arguments.
Also, I don't see support for all of the claims made by other people in this thread to the effect that measurements prove its all impossible. Mostly they are just insisting they are right without providing any published studies in support of their arguments.
Instead of these useless comments from sighted listenings why don't you show e.g. the noise floor of your dac with your best clock compared to e.g. Crystek. As has been shown in RTZ thread noise from bit clock can be seen as elevated noise floor.
Noise floor isn't what people are listening to. If they are listening to any noise at all it is more likely correlated noise in the skirts (which we only call noise because it doesn't appear in an FFT as its own spectral line or lines correlating strongly with one or maybe a few distinct bin frequencies).
Moreover, a lot of what people are hearing has to do with stereo imaging, which nobody at all has shown they can measure with FFTs.
Moreover, a lot of what people are hearing has to do with stereo imaging, which nobody at all has shown they can measure with FFTs.
Measuring noise floor of your dac is not about listening but it would show more objectively how well expensive clock works with RTZ. Isn't that what you are constantly trying to "prove" with your fairy tales from listenings.
I don't care about noise floor fairy tales, like they are the only thing that really matters. What they may be is the only thing you can think of to measure with an FFT, so it seems important to you and other people with the same point of view. For people who listen to music, particularly music in stereo, FFT noise floor is a red herring. It doesn't work as an argument that stereo imaging can't be good, because there is no published science relating imaging to some minimum required dac noise floor.
Last edited:
So if you are not willing to take any measurements what are you doing in this thread? And what are your credentials for educating other members on this topic?
@Markw4 I've said it before and I'll say it again: This thread is about FFT and multi-tone analysis. Please don't turn this thread into another one of your subjective vs objective crusades.
Your subjective anecdotes are exactly that. Subjective. And your appeal to authority (so-and-so famous on the internet said...) will go nowhere here. Please accept this.
Try to put yourself in the shoes of someone years from now who stumbles upon this thread from a search about FFTs in, say, mechanical systems. Then ask yourself: "Will my input on this thread be useful to this person?".
Tom
Your subjective anecdotes are exactly that. Subjective. And your appeal to authority (so-and-so famous on the internet said...) will go nowhere here. Please accept this.
Try to put yourself in the shoes of someone years from now who stumbles upon this thread from a search about FFTs in, say, mechanical systems. Then ask yourself: "Will my input on this thread be useful to this person?".
Tom
Tom,
The most interesting application I am aware of offhand for use of FFTs in acoustical analysis of mechanical systems would be the identification of passing ships by submerged submarines. Its done by FFT analysis of ship engine and driveshaft sounds. The method involves taking FFTs of FFTs to produce identifying patterns of wavenumbers.
I would just add that the method involves the analysis of multiple tones.
https://en.wikipedia.org/wiki/Wavenumber
The most interesting application I am aware of offhand for use of FFTs in acoustical analysis of mechanical systems would be the identification of passing ships by submerged submarines. Its done by FFT analysis of ship engine and driveshaft sounds. The method involves taking FFTs of FFTs to produce identifying patterns of wavenumbers.
I would just add that the method involves the analysis of multiple tones.
https://en.wikipedia.org/wiki/Wavenumber
Last edited:
Not what I was referring to, but there are various ways of analyzing underwater acoustics to derive useful information.
What I talking about is the patterns of squeaks and groans, etc., produced as big ship engines slowly rotate, and usually operate a driveshaft with large (possibly journal) bearings.
So, if we take a short-ish time or a sliding window FFT of the sounds a ship makes with the engine running, then reorganize the data for each bin of successive FFTs into its own time series, an FFT of that new time series for each bin will show the frequency or frequencies with which a particular type of, say, squeak-tone, repeats over time as the driveshaft and engine rotate.
What I talking about is the patterns of squeaks and groans, etc., produced as big ship engines slowly rotate, and usually operate a driveshaft with large (possibly journal) bearings.
So, if we take a short-ish time or a sliding window FFT of the sounds a ship makes with the engine running, then reorganize the data for each bin of successive FFTs into its own time series, an FFT of that new time series for each bin will show the frequency or frequencies with which a particular type of, say, squeak-tone, repeats over time as the driveshaft and engine rotate.
http://www.vibrationdata.com/tutorials_alt/weight1.pdf
and a great book on frequency analysis, everything still valid and should be studied here:
https://www.bksv.com/media/doc/bn1445.pdf
and a great book on frequency analysis, everything still valid and should be studied here:
https://www.bksv.com/media/doc/bn1445.pdf
What are you talking about?
You got exactly what you said you wanted: a useful, practical type of multi-tone FFT analysis as applied to a mechanical system (plus it includes acoustics!)
Last edited:
Not dumb, okay. Paranoid, maybe.
Perhaps you should just be more clear about exactly what want? Personally, I don't know about the practicality of applying a 20-tone test signal to a mechanical system and getting useful results. Is that what you were suggesting could be of interest years from now? I guess it might be possible for electrostatic speakers. Maybe that would that make you even more dissatisfied?
Last edited:
Really good Crystal clocks are BVA: https://rubiola.org/pdf-articles/conference/2007-ifcs-xtal-osa.pdf However these cost as much as a DCS dac and you will need to wait 6 months to 1 year to get one. The ultimate low jitter source is a SAPPHIRE WHISPERING GALLERY: https://hal.science/hal-02993778v1/document
Back to reality- Some DAC archtecture is more sensitive to clock jitter. Its entirely possible that the R2R with the clocking at sample rate and a sample and hold (I hope) is the type that is more sensitive. The switched capacitor type seems least sensitive to clock jitter. However that may be a limitation in ultimate performance.
And I must point out that humans get emotional joy from music through MP3 which has many identifiable and audible errors but not so bad as to prevent connection to music, which I thought was the point. I suspect that the lossy music guys know more about what matters and what doesn't than the ultra audiophile obsessives.
I would call those exceptionally good. At least 10dBc/Hz better at 10Hz offset than the audio clock phase noise plot I posted... Okay maybe closer 4dBc/Hz better at 10Hz offset (allowing for the difference in clock frequencies).
Quite so. @MarcelvdG RTZ is a discrete DSD dac, and a rather clever one at that.
Yes, but I cannot listen to satellite radio in my car. It is just too awful. 256kbps MP3 can be pretty good though.
May I ask if you know how to measure soundstage localization cue reproduction accuracy in audio electronics devices, or do you know of anyone who does know how to do it very precisely?
I can see how FFTs could be useful, but conventional spectral analysis seems likely insufficient.