They only look the same with long data sets and averaging. With shorter data sets and less averaging higher peaks will be seen in FFT provided the data capturing device is up to the task. Also you seem to think that FFTs are snapshots. Most FFT software are able to display online spectrum where even small variations can be seen.
In addition to FFT there are other easy ways to look at uncorrelated noise. E.g. by recording the output and studying the recording with DeltaWave both in time and frequency domain. That also makes it easy to compare DACs in this respect. Instead of speculating why don't you try it yourself with your fave DACs. Although there is a risk that the results would not be to your liking.
I do reclock only for latches in old DACs, like AD1862.
This is not necessary for modern DACs - they are sensitive to jitter of only the MCLK signal.
There is a problem when oscillators are in USB interfaces and there is galvanic isolation - then the clock comes to the DAC with increased jitter. But my oscillators are right next to the DAC, and go back to the interface through the isolator (jitter is not important for it).
Yes, instead of old AK4490EQ, which has Digital Filter inside, new 4499EXEQ (which is improvement of the old AK4498) required the separate DF - AK4191.
There are some benefits using the separate DF, and also AK4191 is able to work in Async mode (like ASRC), that is very good for SPDIF/Toslink/Bluetooth/etc. inputs.
Me too 🙂
I think that good measurements shows good schematic, god components and good PCB layout.
However, the final verdict for audio devices is listening! With possible correction of the circuit and/or component values (for example, in I/V and LPF) during debugging. And it's okay if after that some parameters get a little worse (for example, THD).
Many modern DACs have very outstanding parameters, but many people also like all sorts of horrors (TDA1540/1543, discrete p2p, discrete op amps, etc.) with much worse ones.
But for example, the top ESS, AKM and Rohm have the highest parameters, which are quite close. But they sound different, and some people more like ESS, some more like AKM. Many also noted that with all its wonderful parameters, Rohm is slightly worse than ESS and AKM when play PCM, but excellent when play DSD.
Alex.
Yes. With same schematic and components it is possible to build a well-measuring or worse-measuring device. IME with same schematic and components well-measuring device has always sounded better.
That's true to an extent. Where I still have a problem with deltawave even if using it to do short-time FFTs, is that deltawave also has an audio compare function that, for example, easily shows when differences are around 60dB down when comparing two samples. And the the analog difference signal sounds just about like the real audible difference that can be heard on a good system. However, in order to square the fact that the audio difference does not agree with the FFT analysis, they use an excuse of clock wander to wish-think away the audible difference in favor of the FFT difference which shows much less (like it alone has magic immunity to clock wonder, or maybe deltawave is just averaging out clock wonder as FFTs tend to have an averaging effect on everything). So then what they need to do to reconcile that difference in deltawave outputs is to get an ADC and a DAC both running on TCXO Acko clocks and get rid of the hand-waving clock wander excuse (or at least minimize it as much as possible using current technology). Until then, at least so far as I'm concerned I still have considerable doubt as to the validity of their explanations as to which deltawave output shows closer to the actual differences, FFT or audio.
Last edited:
As I have explained to you many times the audio compare function is highly dependent on the matching. That has nothing to do with hand-waving. As DeltaWave recordings are made from analog signals it is practically impossible to have exact matching even with TCXO Acko clocks. Instead of hand-waving why don't you post recordings that have an audible difference but show no difference in DeltaWave?
Probably Deltawave can be used to show differences between two dacs or whatever. Especially so if a small area of a recording is focused in on. To what use though? Some small differences may be audible and some other small differences may be inaudible.
Even if we take as a given that Deltawave can always show a difference if there is a real difference, especially if focused in on a very small segment of audio. And if we further take as a given that any ADC used to record the output of two dacs will capture the differences with perfect accuracy, then all we have to do is find some way to use what Deltawave shows.
Is it good for showing a difference between two dacs is audible, or else inaudible, and if so for what fraction of the population? I don't think so. Is it good for showing one dac has a better set of sound stage cues? I don't think so for that either.
-----------------------------------------------------------------------
On the subject of whether Delatwave's analog audio difference output is a more or less reliable indicator of real differences between two matched samples, or if the FFT plots are more or less reliable, when there has been a difference between the two types of outputs the FFT seems to be favored by some people, although I am not persuaded they are correct in whole or in part (although correct in part might be the more likely).
Even if we take as a given that Deltawave can always show a difference if there is a real difference, especially if focused in on a very small segment of audio. And if we further take as a given that any ADC used to record the output of two dacs will capture the differences with perfect accuracy, then all we have to do is find some way to use what Deltawave shows.
Is it good for showing a difference between two dacs is audible, or else inaudible, and if so for what fraction of the population? I don't think so. Is it good for showing one dac has a better set of sound stage cues? I don't think so for that either.
-----------------------------------------------------------------------
On the subject of whether Delatwave's analog audio difference output is a more or less reliable indicator of real differences between two matched samples, or if the FFT plots are more or less reliable, when there has been a difference between the two types of outputs the FFT seems to be favored by some people, although I am not persuaded they are correct in whole or in part (although correct in part might be the more likely).
Last edited: