dBu to 0 dBFS

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
tested with RMA
now its looks how it suppose to look to me too...
also re did the test with REW again, same results tho  ::)  no change on 16/24 bt settings!


edit : this is with dithering on... without dithering, previous measurement at given level is correct!
 
i wasnt 100% wrong afterall with REW measurements...

you have to zoom in to c whats going on...
16 bit -  24 bit, with and without dithering!





 
kambo said:
i wasnt 100% wrong afterall with REW measurements...

you have to zoom in to c whats going on...
16 bit -  24 bit, with and without dithering!
Thank you for taking the time; it just proves that more bits result in better measurable performance, although I could hardly hear the difference between 0.0023 and 0.0005% THD and between -136 and -147 dB noise.
The difference in a null test would be at about the same level, i.e. about -110dBfs for the distortion residuals; that would be difficult to hear without the use of some kind of "magnifier", i.e. about 40 dB gain for raising the signal above the noise of the monitoring path. That would make the measurement questionable because of the difficulty in controlling and eliminating all sources of error.
But the null signal could be visualized on the waveform screen easily.

It's interesting to note that dither doesn't improve the 24 bit performance as much as that of 16 bit (in fact it's not surprising since 24-bit is closer to "perfection" than 16).
That may be explained by some "natural" noise-induced dithering, as JR mentioned earlier, but it seems your test has been done in the digital domain, so I don't really see what could be the noise source, maybe the digital attenuator which produces some truncation noise, but probably 40-bit below 0dBfs, so I'm not sure...
 
unless i am missing something, without the dither i dont c any difference at all...
and with these measurements, would anyone bother using dither at all ?
 
kambo said:
unless i am missing something, without the dither i dont c any difference at all...
and with these measurements, would anyone bother using dither at all ?
I've done the same test as you with Samplitude; it's not identical since the generator has only 16 and 32-bit options, and the FFT analysis does not highlight the harmonics, but I believe it demonstrates clearly enough that more bits result in better conversion. Indeed some kind of dithering must be happening, since the discrete  harmonic fringes seem to be transformed into HF noise.
 
here is the harmonics

top 16/48
middle 24/48
bottom 32/48

edit: wav file exported direct from cubase, no loopback... cubase tools tone generator used for 1khz test tone.
no dither.
 
I believe there's something wrong, I don't know what (exporting files, you don't know what happens, particularly when the DAW adds dither without warning).
I've done the comparison with RMAA between 16, 24, 32 and 32 float, and guess what? It gets better and better. I must admit the difference between 24, 32 and 32 float is marginal.
I believe the FFT in RMAA is not adequate for files; it's ok for live testing, though. Windowing may be an issue...
 
woooooaaaaaaa  :eek:
what is your sound card... my figures are not even remotely close to yours... 

clearly u have a way better sound card than i do.

IMHO, analysing wave file direct export from DAW is better way,
i did my all my past tests in mono... this is stereo loopback on output 7-8


 
kambo said:
woooooaaaaaaa  :eek:
what is your sound card... my figures are not even remotely close to yours... 

clearly u have a way better sound card than i do.
That's pure analysis of the .wav files; no soundcard involved. I wouldn't say it's "perfect" because there's always some math errors but I would say, that's as close as can be to the purest data.

  IMHO, analysing wave file direct export from DAW is better way, 
I had troubles exporting from Samplitude, and I guess you had too. Most DAW's make choices, unknown to you, when bouncing from one format to another.
 
Be aware that the shape of the noise floor around the signal is an artifact of the FFT calculation and windowing(?) decisions.

I repeat that old school (analog) bench  THD+N  and S/N measurements can be instructive.

JR
 
JohnRoberts said:
Be aware that the shape of the noise floor around the signal is an artifact of the FFT calculation and windowing(?) decisions.
Yes, that's why I made sure the settings would be the same for all the analyses, and indeed, it qualifies only for comparative assessment. Since I used internally generated files that start and end with a zero-crossing, I could have dispensed with windowing, but I thought it was not necessary, considering the FFT was stable enough after 3-4 seconds.


I repeat that old school (analog) bench  THD+N  and S/N measurements can be instructive.
Remember the issue was raised about the intrinsic capability of higher bit depth to produce better performance, so bypassing any converters adresses the issue at the heart.
Indeed, what's important is what comes out of the orifices, and analog measurement is a proven evaluation method; however, we're dealing here with differences that most audio analysers would not register. AP555, the unquestioned champion, has a residual THD of -117dB and a self-noise about the same. With it, one can certainly assess the difference between 16 and 24 bit, but not between 24 and 32. And anyway, whatever analog circuitry in the converters would tend to mask any difference. And not everybody has access to a 555, or even a 525.
I believe no one in his right mind can deny that 24bit gives superior performance to 16 bit, provided the implementation is correct. There is enough proof that well-implemented converters do exist, delivering performance close to the chip mfgrs specs.
 
abbey road d enfer said:
I believe no one in his right mind can deny that 24bit gives superior performance to 16 bit, provided the implementation is correct. There is enough proof that well-implemented converters do exist, delivering performance close to the chip mfgrs specs.

i dont mean to be disrespectful to your knowledge and experience;
IMHO
my understanding is 24 bit has better noise floor, resulting in better dynamic range, and stereo crosstalk figures.
actual recorded signal above -86 dbFS is exactly the same in 16 bit and 24 bit.
RMAA exported ref. wav files do NULL too, left over is just noise, as expected around -86-89 dBFS, thats where all the noise and distortion is...
(figures changes if u convert 16 to 24 or 24 to 16 bit)

if you look at my loopback results (my last post) with a decent enough AD/DA (lynx aurora)
i really dont see any superior performance, but better!
may be i am missing the whole picture!

edit: RMAA ref wav files dithered btw!

 
kambo said:
i dont mean to be disrespectful to your knowledge and experience;
IMHO
my understanding is 24 bit has better noise floor, resulting in better dynamic range, and stereo crosstalk figures. 
Isn't it enough to say that it results in a better conversion?

actual recorded signal above -86 dbFS is exactly the same in 16 bit and 24 bit.
Indeed.


RMAA exported ref. wav files do NULL too, left over is just noise, as expected around -86-89 dBFS, thats where all the noise and distortion is... 
If there's something leftover, that means that it doesn't null! What did you expect? Hearing the Star and spangled banner or "Paul is dead"?  :)
 
abbey road d enfer said:
Isn't it enough to say that it results in a better conversion?
Indeed.

If there's something leftover, that means that it doesn't null! What did you expect? Hearing the Star and spangled banner or "Paul is dead"?  :)
Come on man Paul isn't dead....  8)

JR
 
abbey road d enfer said:
Isn't it enough to say that it results in a better conversion?
Indeed.

indeed  :)
that exactly what i said... better... 
not superior!

"If there's something leftover, that means that it doesn't null! "

how exactly do you ever NULL a noise  :eek:
i thought it was clear  that i was referring to actual signal NULL, not including the noise floor and distortion !
 
kambo said:
indeed  :)
that exactly what i said... better... 
not superior!
From Thesaurus:
Better: 1. of superior quality or excellence

"If there's something leftover, that means that it doesn't null! "

how exactly do you ever NULL a noise  :eek:
Indeed, a purely random noise cannot be nulled, but harmonics could (or not) null. The issue there is that dithering shifts the harmonics in the upper region of the spectrum, where they can't be heard and are not seen because they're masked by the thickening of the graph.


i thought it was clear  that i was referring to actual signal NULL, not including the noise floor and distortion !
It was not clear to me, because I never had any doubt that the fundamental would null; that's mathematically evident. Again, we see the importance of terminology. The signal is the whole signal, including noise and distortion.
 
apologies for the confusion !

as for the "better", i am sure u dont use superior  for something a bit better very often... unless its woooow :)
anyway, English is not my first language...  i take your word for it :)
 
abbey road d enfer said:
Isn't it enough to say that it results in a better conversion?

Broadly speaking everyone agrees. The data posted doesn't seem to prove any point anyone has disputed so far...

Ok, so my point was really this:

kambo said:
my understanding is 24 bit has better noise floor, resulting in better dynamic range, and stereo crosstalk figures.
actual recorded signal above -86 dbFS is exactly the same in 16 bit and 24 bit.

To which you replied:

abbey road d enfer said:

Yet when I tried to make the same point the answer was just "no".
 
  At least 2 different brands lunched 32 bits converters last year, for metrology proposes, don't get excited audiophiles... Of the shelve noise floor aren't even close to those 32 bits but they are still useful for linearization, self testing, etc. over having fewer bits from the converter it self.

  In audio we don't usually do that kind of nonsense, but from a good 24 bits converter you still get much lower INL than from a 16 bits one, measurable over the analog noise floor in many cases. You are not going to get a mic to cover 144 dB dynamic range, nothing new there, neither from a 24 bits converter, a few of those bits are lost in the way, equivalent useful bits are a few bits down in a ny converter (excluding 8 bits ones and there about which have INL of ~0.5 LSB). The noise floor of the analog path is going to be higher than the digital one. Once inside the computer no argue is advantageous working with a few extra bits, so we already need to store and process the 8 extra ones, why not to already record the noise and avoid the use of dither in that stage, maybe we get a useful bit or two in the way.

JS
 

Latest posts

Back
Top