Ozone beta-build listening
Posted 8th January 2014 at 06:18 AM by abraxalito
Updated 8th January 2014 at 07:15 AM by abraxalito
Updated 8th January 2014 at 07:15 AM by abraxalito
I've had the beta build Ozone (no input stage yet, fed from my QA550 via the I2S transcoder to down-convert to 32fs) playing out 24/7 for a few days now.
Overall I'm very happy with how it sounds, just a minor gripe about sibilance on some operatic vocals which I'd like to understand better. On the upside the jump-factor (read dynamics) and soundstage stability (holographic on the right disks) are about the best yet. I'm using a couple of Decca double CDs for this - 'La Traviata' and "La Boheme'. They're about the most transparent sounding and demanding disks I have. Demanding in the sense that they have lots of emotional drama which should positively engage my attention if the DAC's really up to snuff. I never much enjoyed opera until I got into building my own NOS DACs, but now I really enjoy my dramatic fixes and these two recordings are really top of the pile. They're about 50 years old but so far I've not found anything newer which touches them (not that I've looked very far yet - however I just for fun pulled out a 2012 Decca of Gounod's Romeo&Juliette and that totally sucks dynamically in comparison).
I have tracked down a circuit detail that seems to affect the amount of sibilance and it concerns how pin9 (VOCM) on the AD605 is connected. This is used to determine the output common mode voltage for the two output channels - fig35 on p13 of the DS shows how this pin is wired internally. Its buffered and then forms the reference for the feedback resistor to the gain stage. If you follow the DS recommendation and merely decouple it to the -ve supply pin this introduces power supply noise and the soundstage goes completely to pot. That is unless pins 4/5 are your star earth anyway - but this is topologically impossible for a stereo pair. So pin9 needs to go to a nice quiet, low impedance 2.5V reference which over-rides the resistors inside the chip and hence attenuates the power supply noise fed through them by the ratio of the impedances - its very fortunate ADI chose such high values (200k).
Originally I had this pin going direct to a 100uF bypassed TL431 (same reference for both channels) but it seems that's a tad too noisy so I now have an RC filter - 5ohms and 1000uF and this has reduced the sibilance considerably (not just from the LPF noise reduction - the cap goes to the 0V star now with its own dedicated wire) but not totally eliminated it. I'm wondering whether an LC filter will do better by virtue of presenting a lower impedance to the two pin9s. But I'm also thinking that I'll get better performance from having two totally separate channels rather than a stereo DAC. I've even given this variant a name 'Free radical'.
Picture attached shows the DAC engine, sandwiched between two PCBs full of caps with newly munged passive filters (6 pairs of inductors rather than 8, X7R caps rather than NP0s, shock horror!) feeding the analog stage to the right. Off camera are output trafos doing bal-SE to feed my Xindak integrated amp. The green lytics aren't real Sanyo, rather 'Shanyo' ones I bought from Taobao last summer with the intention of using them in my high-end chipamp. Turned out they weren't cut-out for 15.4V operation (despite being labelled 16V) and several exploded. However running under 7V they work fine and have impressively low ESR (20mohm @ 1kHz) given they're so cheap. I have some even lower ESR ones (5mm taller) to try on the next build, which will be discrete channels to try out the 'Free radical' concept......
Overall I'm very happy with how it sounds, just a minor gripe about sibilance on some operatic vocals which I'd like to understand better. On the upside the jump-factor (read dynamics) and soundstage stability (holographic on the right disks) are about the best yet. I'm using a couple of Decca double CDs for this - 'La Traviata' and "La Boheme'. They're about the most transparent sounding and demanding disks I have. Demanding in the sense that they have lots of emotional drama which should positively engage my attention if the DAC's really up to snuff. I never much enjoyed opera until I got into building my own NOS DACs, but now I really enjoy my dramatic fixes and these two recordings are really top of the pile. They're about 50 years old but so far I've not found anything newer which touches them (not that I've looked very far yet - however I just for fun pulled out a 2012 Decca of Gounod's Romeo&Juliette and that totally sucks dynamically in comparison).
I have tracked down a circuit detail that seems to affect the amount of sibilance and it concerns how pin9 (VOCM) on the AD605 is connected. This is used to determine the output common mode voltage for the two output channels - fig35 on p13 of the DS shows how this pin is wired internally. Its buffered and then forms the reference for the feedback resistor to the gain stage. If you follow the DS recommendation and merely decouple it to the -ve supply pin this introduces power supply noise and the soundstage goes completely to pot. That is unless pins 4/5 are your star earth anyway - but this is topologically impossible for a stereo pair. So pin9 needs to go to a nice quiet, low impedance 2.5V reference which over-rides the resistors inside the chip and hence attenuates the power supply noise fed through them by the ratio of the impedances - its very fortunate ADI chose such high values (200k).
Originally I had this pin going direct to a 100uF bypassed TL431 (same reference for both channels) but it seems that's a tad too noisy so I now have an RC filter - 5ohms and 1000uF and this has reduced the sibilance considerably (not just from the LPF noise reduction - the cap goes to the 0V star now with its own dedicated wire) but not totally eliminated it. I'm wondering whether an LC filter will do better by virtue of presenting a lower impedance to the two pin9s. But I'm also thinking that I'll get better performance from having two totally separate channels rather than a stereo DAC. I've even given this variant a name 'Free radical'.
Picture attached shows the DAC engine, sandwiched between two PCBs full of caps with newly munged passive filters (6 pairs of inductors rather than 8, X7R caps rather than NP0s, shock horror!) feeding the analog stage to the right. Off camera are output trafos doing bal-SE to feed my Xindak integrated amp. The green lytics aren't real Sanyo, rather 'Shanyo' ones I bought from Taobao last summer with the intention of using them in my high-end chipamp. Turned out they weren't cut-out for 15.4V operation (despite being labelled 16V) and several exploded. However running under 7V they work fine and have impressively low ESR (20mohm @ 1kHz) given they're so cheap. I have some even lower ESR ones (5mm taller) to try on the next build, which will be discrete channels to try out the 'Free radical' concept......
Total Comments 6
Comments
-
Solid report, Richard, I'm impressed with the concept of circuit implementation practiced as form of abstract art ... :D
Remaining SQ issues, sibilance and such, could be knocked over by the right decoupling strategy, or even a solid tidying up, physical stabilising of everything. As you've already experienced, a single lead not going to the optimum point, or positioned poorly can make all the difference at this level of reproduction - simple experimentation is as good a technique as any, just keep fiddling with everything that strikes you ...
Yes, I would go for a DAC per channel ...
FrankPosted 9th January 2014 at 12:15 PM by fas42
-
I've run into the 'Frank effect' once again with this - aka 'not enough caps' syndrome. Even though I put 2 * 3,330uF on each AD605 turns out that more sounds better - more 'snap' and 'weight' to the bottom end, better foot-tapping to all instruments. So, once again active components are limited by their power supplies.
Going 'dual mono' allows more caps to be placed closer to the amp chips. Have you seen any dual mono DACs incidentally on your travels? I got the impression that the original 'Trinity' DAC might have been dual mono, but the latest incarnation seems to have reverted to a single box. I must do more research into that...
Yes, lower noise, better grounding on the next revision. I've already designed an LC filter for the critical 2.5V reference. I might also make a recording of the PSU noise when playing music to get a handle on its magnitude. I shall need a trafo to raise it out of the noise of my Sony recorder I think....
Inicidentally, listening to another Decca opera from my collection now - 'La Cenerentola' by Rossini, sung by one of my faves, Cecilia Bartoli - its a digital original recording I think, not as dynamic as the 'La Trav' - so analog is still my reference. The main difference I hear is in the LF noise - there's just more LF 'ambience' information captured by the magnetic tape, a kind of LF bloom. But why I wonder is the ADC used for remastering better than that for the recording? Hmmm...Posted 9th January 2014 at 02:15 PM by abraxalito
Updated 9th January 2014 at 02:20 PM by abraxalito -
Yes - TNT review of the 2006 incarnation of the Trinity DAC here : GTE Trinity DAC Technical Review [English]
Seems rather odd that the justification for the 3 boxes is to 'keep the power supply away from the sensitive DAC circuis'. Whereas I want the opposite - won't pay to have a cap box methinks....Posted 9th January 2014 at 02:41 PM by abraxalito
-
Have to be careful with that magnetic tape 'bloom' - there is a very well known distortion artifact generated by those old analogue recorders, which adds 'richness' to the sound - the tape doesn't move 100% smoothly over the heads, there is slight "stiction" - a subtle vibrato or tonal enhancement occurs ...
These days the studios have digital 'boxes' which almost perfectly transparently add that type of tape recorder distortion to the sound, a spicing up of the mix with 'analogue warmth' :)Posted 9th January 2014 at 10:09 PM by fas42
-
Yes - I don't think that's what I am hearing. I hear improved dynamics with the original analog tape, in comparison the digital original is slightly more 'bleached' or 'washed out'. In other words, I reckon the digital suffers from more noise modulation than the analog - which is rather ironic seeing as the major shortcoming of analog for me is modulation noise
I still remain to be convinced there is such a thing as 'analog warmth' - rather I do hear digital 'coldth'
Posted 10th January 2014 at 12:55 AM by abraxalito
Updated 10th January 2014 at 12:58 AM by abraxalito -
That sounds like there still may be problems handling the treble content - I find a powerful indicator is how cymbals come over in very strongly driving rock music, if they are dead in quality or not fully resolved to sound like the real thing, they don't shimmer like they do in real life, then the treble content is being contaminated. I have, at the moment, a very bizarre, 'experimental' techno music disk, extremely 'digital', full of real world sounds, that ring with treble harmonics constantly. If these sounds weren't reasonably correctly rendered this disk would be a nightmare: extremely dreary and pointless, or headache inducing shriekiness ...Posted 10th January 2014 at 06:25 AM by fas42
Updated 10th January 2014 at 06:27 AM by fas42
