So if opamps were used as regulators for Vref, I don’t see any large caps following them.
Really ? Maybe Iam looking at the wrong chart here, I have randomly picked a pretty standard WIMA MKP10 data sheet...have a look at the very bottom, the last, dotted curves stand for the metallized series and they have a nice low impedance even at 1mhz.
Kurven - WIMA – Competence in Capacitors
In the tdk white paperunder 2.6 we have similar curves on mkp where you would anticipate their ESR goes down until 3mhz before it starts to raise again:
https://www.tdk-electronics.tdk.com...1513474ba/pdf-generaltechnicalinformation.pdf
...so, yes the impedance goes up ...at 100mhz...but this may not our topic anymore, no ?
And I have yet to ask the guys from mundorf who claim their stuffis non-inductive or very low inductive at least etc...but if a MkP10 can behave that nicely in ESR at 3mhz, I would guess a Mundorf can do better, no?
And I am with Peter, I have never found it to sound satisfactory when you mix capacitors at the same position in a circuit. But I found no issues to mix them when they are the exact same type, but in different values, so a 10uF MKP10 mixed with a 0,1uF MKP 10 should sound like one cap.
So, which one are the ugly little ceramics toworry about ? Peter, I guess youare already pulling them ? BG with ceramics in parallel sounds really like a bizarre idea...
Last edited:
Yes, although this is not an argument i have any real interest in. You are quoting ESR, which is not impedance. Once you include the effects of ESL and the lead inductance the picture looks very different.
Otoh cannot imagine any good reason for the heavy bypassing of the polymer caps. It achieves what exactly? Slightly less noise in the MHz region? Does it even matter? I am also with PD and never bypass electrolytics in an analogue circuit. Let the bypassing stay where it is needed - in the PS of the shift registers.
So I’m feeling adventurous tonight and would like to remove some of those ceramic caps and replace them with 0.1uf BG NX. There are 8 sets of four caps and it seems like each quadruplet is in parallel with a large electrolytic. Are they all the same value? Why 4 per each set?
BTW, after recent discussion here in this thread, I went back to original electrolytics, just to make sure that I was not imagining things, but I didn’t like how the DAC sounded. The harmony and naturalness was missing and I quickly put back my “magic” caps.
I don’t expect much from removing those ceramics, but at this point I need to check it out.
BTW, after recent discussion here in this thread, I went back to original electrolytics, just to make sure that I was not imagining things, but I didn’t like how the DAC sounded. The harmony and naturalness was missing and I quickly put back my “magic” caps.
I don’t expect much from removing those ceramics, but at this point I need to check it out.
Attachments
I'm with you on that, and the caps I use in the DAC filters are NP0/C0G, but apparently a lot of people here disagree....
Could be a perfect capacitor when it comes to measurements and simulations, and it could be only OK for audio, when it comes to listening.
At one time I was inspired by using certain combinations of capacitors for bypassing, but after more careful listening I came to conclusion that this is not a way go: The mods never end, adding Goudreau's triplets
While superficially it might have sounded OK or better, it took away something and made music sound flat and lifeless. So after short period of fascination with stacking and bypassing caps, I decided to stay away from it. As a rule now, I try to never parallel caps, or add small caps to larger caps to make them sound better. A single cap, carefully chosen, should always sound better than multiple caps in parallel.
At one time I was inspired by using certain combinations of capacitors for bypassing, but after more careful listening I came to conclusion that this is not a way go: The mods never end, adding Goudreau's triplets
While superficially it might have sounded OK or better, it took away something and made music sound flat and lifeless. So after short period of fascination with stacking and bypassing caps, I decided to stay away from it. As a rule now, I try to never parallel caps, or add small caps to larger caps to make them sound better. A single cap, carefully chosen, should always sound better than multiple caps in parallel.
OK, so back now to the mods. I remove 8 ceramic caps as presently I'm using unblanaced output only. It was an easy and clean job. While I said earlier that I try not to bypass the caps, I keep here the BG NX 0.1uF as I think it may actually sound better with them (I didn't listen without those caps yet and I will try it tomorrow).
So how does it sound? Well, the difference is rather subtle, but I would say quite important. I didn't observe much changes in overall tonality, but what stroked me first was increased depth to the sounds and much better audibility of lyrics.
I just hear more, there is more layers, the separation is better, each sound follows it's own melody (if I may use that expression), there is more space around instruments, more air, more detail, more background sounds (no matter how faint they are, I can still notice them). I think overall presentation is more relaxed, less forced and there is less edginess. I cannot complain about the bass at all, it's well defined with very well pronounced PRAT.
And while I said in the beginning that the DAC sounded promising, I think it finally delivered.
OTOH, it may be only momentary fascination after doing few mods, and eventually everything becomes normal again and the excitement will be gone, so let me enjoy those few moments
So how does it sound? Well, the difference is rather subtle, but I would say quite important. I didn't observe much changes in overall tonality, but what stroked me first was increased depth to the sounds and much better audibility of lyrics.
I just hear more, there is more layers, the separation is better, each sound follows it's own melody (if I may use that expression), there is more space around instruments, more air, more detail, more background sounds (no matter how faint they are, I can still notice them). I think overall presentation is more relaxed, less forced and there is less edginess. I cannot complain about the bass at all, it's well defined with very well pronounced PRAT.
And while I said in the beginning that the DAC sounded promising, I think it finally delivered.
OTOH, it may be only momentary fascination after doing few mods, and eventually everything becomes normal again and the excitement will be gone, so let me enjoy those few moments
Attachments
Last edited:
Just further on the caramics... I think Cyril Batemans series of articles on capacitors for audio first identified NPO ceramics as a good possibility.
I do agree though, theory and ears must go hand in hand.
Soekris, one way to progress the sonic side of things is to find a trustworthy listener and provide them with 2 units each with a different variation of the design. Ask just which they prefer. It's as close to double blind as you can get as there is no external indication of any difference. More listeners can be better.
I do agree though, theory and ears must go hand in hand.
Soekris, one way to progress the sonic side of things is to find a trustworthy listener and provide them with 2 units each with a different variation of the design. Ask just which they prefer. It's as close to double blind as you can get as there is no external indication of any difference. More listeners can be better.
Thanks, Peter...so for sure this is an area to influence our experience...awesome job...I will try this mod later myself...need to think which cap...BG at current prices is a nogo...
btw, when you use unbalanced why did you not gave your balanced dac the setting to combine two balanced dac strings to one unbalnced and half this way the output impedance ? Typically lower Z sounds better as well...
btw, when you use unbalanced why did you not gave your balanced dac the setting to combine two balanced dac strings to one unbalnced and half this way the output impedance ? Typically lower Z sounds better as well...
Last edited:
I think i am going to scream next time i read this line again somewhere. Which of course will be no later than tomorrow
Many thanks to PD. Very curious if the absence of bypass won't sound still better.
I was experimenting earlier with TDA1543 in parallel and the single chip sounded better. Also, LM4780 in parallel, didn’t sound as good as balanced, so I’m reluctant to put circuits in parallel.
Besides, the circuit on a right from a DAC board is a headphone amp and this one works in balanced mode, so occasionally when I switch to headphones, I need balanced DAC.
Meanwhile i tested something else: my favourite JLSounds usb board vs the inbuilt Xmos.
To match the test fairly the isolated side of the JLSounds was powered from the same +5v which supplies the 1941 and the usb side off the usb incoming power +5v. This is not how i normally use the usb board: it gets it's dedicated Salas shunt for the isolated side and a simple, but independent +5v PS for the usb side and yes, this makes a substantial difference.
Anyway, for the purposes of comparison both were powered up in exactly the same way and stayed powered for the duration of the comparison. Switching the usb cable between one input and the other resulted in fast and error prone autodetection, so all that was needed to do between tracks was change the drivers.
Having tried most of the available usb boards i was surprised to notice relatively small differences. Timbrally the 1941 xmos had a mild mid bass bump and sounded slightly darker in the upper ranges. The bass bump also explained my initial impression of increased lower range solidity of the 1941 vs 1021.
After some extended listening i am happy to say i can live with both. My slight preference goes towards the JLSound board but the differences are not dramatic. It is clearer in the upper ranges and has substantially better 3D. This clarity can also get a bit tiring if the program material is unsympathetic.
Now i need to look again at the board and figure out how the PS of the usb part can be separated and improved. The input usb part is probably easy to figure out, but i am not so confident about the rest. The JLSounds topology makes such a decision easier as there is a reclocker on the clean side of the galvanic isolator, so this is obviously where the second PS goes to.
To match the test fairly the isolated side of the JLSounds was powered from the same +5v which supplies the 1941 and the usb side off the usb incoming power +5v. This is not how i normally use the usb board: it gets it's dedicated Salas shunt for the isolated side and a simple, but independent +5v PS for the usb side and yes, this makes a substantial difference.
Anyway, for the purposes of comparison both were powered up in exactly the same way and stayed powered for the duration of the comparison. Switching the usb cable between one input and the other resulted in fast and error prone autodetection, so all that was needed to do between tracks was change the drivers.
Having tried most of the available usb boards i was surprised to notice relatively small differences. Timbrally the 1941 xmos had a mild mid bass bump and sounded slightly darker in the upper ranges. The bass bump also explained my initial impression of increased lower range solidity of the 1941 vs 1021.
After some extended listening i am happy to say i can live with both. My slight preference goes towards the JLSound board but the differences are not dramatic. It is clearer in the upper ranges and has substantially better 3D. This clarity can also get a bit tiring if the program material is unsympathetic.
Now i need to look again at the board and figure out how the PS of the usb part can be separated and improved. The input usb part is probably easy to figure out, but i am not so confident about the rest. The JLSounds topology makes such a decision easier as there is a reclocker on the clean side of the galvanic isolator, so this is obviously where the second PS goes to.
As the vref, and impedance at that point, plays a critical role in the operation of this dac it would be interesting to know what the result of this mod and the reduced capacitance does to the DACs performance. Is resolution lost? Is there some frequency dependence introduced by an impedance that chances with frequency?
Soekris, could you comment? How low does the impedance here need to be to ensure the DACs performance capability is reached, and what effect will changes in impedance have?
Soekris, could you comment? How low does the impedance here need to be to ensure the DACs performance capability is reached, and what effect will changes in impedance have?