It's a knwon fact, but distorsion is not the only parameter, I avoid it as often I can , never went lower than 0605 size though... which is more often seen with X7R.
Maybe the scope is better, but many other dielectric are giving a better subjective result at ears ! If they continue in those direction we will have soon spectacular measured stuffs... which sound... bad but look good at scope ! like the Nichicon bipolar Muse... subjectivly a nightmare as well !
It's ok when I listen to such music : https://www.youtube.com/watch?v=ZtAZTgaog9o, not anymore with more complex classic or acoustic only !
Maybe the scope is better, but many other dielectric are giving a better subjective result at ears ! If they continue in those direction we will have soon spectacular measured stuffs... which sound... bad but look good at scope ! like the Nichicon bipolar Muse... subjectivly a nightmare as well !
It's ok when I listen to such music : https://www.youtube.com/watch?v=ZtAZTgaog9o, not anymore with more complex classic or acoustic only !
Last edited:
I reckon the precise point of attachment of the // cap is going to be more critical than the cap type. Needs a very, very quiet 0V. I agree NP0 is the way to go, lowest ESL hence highest SRF.
Last edited:
If a shunt NPO near a digital stuff sounds bad, one has to jump than the layout is bad (I mean the pcb drawing) ?
One day I tried with TDA1545 and TDA 1541 some NPO : didn't liked the result ! Some PPS smd caps were better subjectivly at ears ! Bad PCB/Design ? Or maybe the chip (TDA1541 : either Rogic & Loesch avoid NPO for the 14 caps arounds the TDA1541) ?
With the Subbu for instance I tried some NPO where there was X7R... there was subjectivlya sort of ringing I didn't like at ears and went back with class II (iso size)
One day I tried with TDA1545 and TDA 1541 some NPO : didn't liked the result ! Some PPS smd caps were better subjectivly at ears ! Bad PCB/Design ? Or maybe the chip (TDA1541 : either Rogic & Loesch avoid NPO for the 14 caps arounds the TDA1541) ?
With the Subbu for instance I tried some NPO where there was X7R... there was subjectivlya sort of ringing I didn't like at ears and went back with class II (iso size)
I've never liked the sound of a cap to 0V on the output of a DAC. I originally tried the experiment of removing it on the Muse 4 * TDA1543 (there its 2.2nF) and it made a considerable improvement. I think I shared this finding on HeadFi and at least one other user confirmed it.
I think perhaps the problem of bad sound could have been in the layout in that a DAC output has a huge HF energy (due to extremely fast edges) and its too easy for a 0V to get corrupted by this. When you didn't like NP0 compared to normal leaded caps it could be due to the series inductance of those leaded caps attenuating the HF hash.
I think perhaps the problem of bad sound could have been in the layout in that a DAC output has a huge HF energy (due to extremely fast edges) and its too easy for a 0V to get corrupted by this. When you didn't like NP0 compared to normal leaded caps it could be due to the series inductance of those leaded caps attenuating the HF hash.
Ah, thanks ! So too short caps (inductance) not so mandatory ?
Which is always odd to me is that glitchs are in the audio band or I couldn't hear it !
Which is always odd to me is that glitchs are in the audio band or I couldn't hear it !
No, the glitches aren't in the audio band, they're way above it. But they degrade the SQ because all subsequent stages have non-zero IMD. So the dynamics always suffer when your audio path gets contaminated with ultrasonics.
guys,
I think you would like to see this project http://www.diyaudio.com/forums/digital-line-level/254935-signalyst-dsc1-23.html#post4977595
I think you would like to see this project http://www.diyaudio.com/forums/digital-line-level/254935-signalyst-dsc1-23.html#post4977595
Thanks Abrax,
I have to say than the InterModulation Distorsion is above of myunderstanding. So does an OT a good choice instead a cap or a trimm to zero like in the shematic, or maybe a multipole filter like you made with cheap coil (bu the transparency VS the glitch = trade off?).
Samoloko :since the first R2R Dam1201 stock which had sota measurement but bad sound for my subjective tastes, I'm a little suspiciou s, let say carefull now when it's just about anouncement and good measurement only ! 95% of the brand have decent ingineer, but most of the DAC sounds not so good... for most they don't got it despite all the good measurement ! So experience old engineer and tweaker is more reliable ! I'm certainly naive, who knows !
I have to say than the InterModulation Distorsion is above of myunderstanding. So does an OT a good choice instead a cap or a trimm to zero like in the shematic, or maybe a multipole filter like you made with cheap coil (bu the transparency VS the glitch = trade off?).
Samoloko :since the first R2R Dam1201 stock which had sota measurement but bad sound for my subjective tastes, I'm a little suspiciou s, let say carefull now when it's just about anouncement and good measurement only ! 95% of the brand have decent ingineer, but most of the DAC sounds not so good... for most they don't got it despite all the good measurement ! So experience old engineer and tweaker is more reliable ! I'm certainly naive, who knows !
OT is always a good choice for reducing noise on the interconnecting cable (between DAC and amp).
The original Soekris DAC as far as I recall had poor stability of the reference voltages didn't it? So DIYers were sticking extra caps on those Vrefs to reduce the signal-correlated modulation of those voltages. But doubtless the measurements weren't affected.....
The original Soekris DAC as far as I recall had poor stability of the reference voltages didn't it? So DIYers were sticking extra caps on those Vrefs to reduce the signal-correlated modulation of those voltages. But doubtless the measurements weren't affected.....
How about using big air core inductor as filter to get rid of the HF noise? IIRC, zanden dac adopted this approach. Maybe there is a need for buffer after the inductors?
Air Core is nice but how big a mh ? Auditioned the Zanden
once. Very nice for simple music but sounded lack lustre
on complex pieces. Think was due to a mixture of passive
i/v & the many passive inductors that was used to make
the filter network
once. Very nice for simple music but sounded lack lustre
on complex pieces. Think was due to a mixture of passive
i/v & the many passive inductors that was used to make
the filter network
I've seen a pic of the Zanden filter, very complex with a dozen or more large inductors. But I doubt very much they're air-cored as with air core its very hard to get a high enough Q factor in the audio band. As far as I recall there was a patent on that filter - it was a series of tuned circuits to give nulls in the FR at multiples of the sample frequency with an aim to avoid excessive phase shift.
Yes Abrax complicated network of inductors etc.
Makes me wonder how much signal has gotten
lost within. LOL
Makes me wonder how much signal has gotten
lost within. LOL
Does the over sampling of our Library with a soft could be affective to remove the glitchs at higher levels for suppressing the shunt cap // with the I/V resistor ?
Can't get your question sorry Eldam. Oversampling is going to increase glitching which is I believe the reason why NOS is mostly preferred.
I meaned : despite it's always going through the audio band as you said, does not the oversampling (with sox soft for instance) of a red book material move the amplitude of these glitchs at a higher frequency (so always noise but less important in the 20 Hz 20K Hz?)
Or what ever the sampling rate will we have the same level of noise through the audio band one can hear ?
Or what ever the sampling rate will we have the same level of noise through the audio band one can hear ?
Hi Eldam
Me thinking what if we have a very clean transport with
minimum errors, would this not reduce issues down the line ?
Me thinking what if we have a very clean transport with
minimum errors, would this not reduce issues down the line ?
Yes but not enough ! I have problem to calm down the Taiwan 98 1541A chip in simultaneous mode despite all the Ian's stuffs which has certainly as low if not good measurements than the SD-Trans364... For me the Single crown is still superior ! In my setup the Taiwan 98 has a loudness pleasant sound with many detail but also this harchness in the mid-high.
Maybe we're meaning different things by 'glitches' ? To me they're introduced as artifacts by the DAC they're not on the audio file so the oversampling can't affect them. There are 'images' (frequency domain reflections of the audio material in mirrors placed at various multiples of the sample rate) which are moved further out of the audio band by oversampling, is this what you mean by 'glitches' ?
I reckon we get more noise in the audio band the more we oversample. Coz the glitches out of the DAC are broadband, some of their energy is low enough frequency to be heard. Or perhaps its just that by adding more and more HF energy the unavoidable IMD in subsequent circuits results in greyer tonal colours.