Hexacap-based Ozone DAC
Here's my current bench lash-up DAC based on three hexacaps. Total capacitance of the order of 1.3F.
32fs EIAJ is fed in, balanced, from the left, the central hexacap feeds two TDA1545ATs. Filters are TDK inductor based, six stage balanced with X7R caps (gasp!). Caps in the centre provide the mid-rail power (2.5V) which the I/V resistors are fed from. The two hexacaps on the right power the post amps based on AD815ARBs, gain set roughly to 26dB with a little NOS droop correction. They're the postage-stamp sized boards roughly in the middle. Output is differential and fed to ferrite-cored transformers to convert to SE.
I'm blown away by the dynamics of this, given the right recording. Next up is an amp based on paralleling 8 or 10 AD815s. Each fed from its own hexacap, of course....
Update1 - the DAC supply has now been upgraded to two tiers of the 4700uFs, interconnected by six 1.5mm diameter solid copper wires. Capacitive reactance measures 2.5mohm @ 50Hz and ESR is below 1mohm at this freq, falling at higher audio freqs to around 0.5mohm. My LCR meter's probes don't make consistent contact down at this level so the reading is subject to a fair degree of uncertainty. Adding the extra tier improved the perceived LF noise so it seems I haven't reached the limits of the chip yet.
I also found the supply used to terminate the I/V resistors benefits from lower noise. I started out with just 6000uF on this, upped it to 20,000uF and figured I'd be better off if I was using the DAC's own supply rather than a separate (and bulky) bank of caps. So I'm now experimenting with running the TDA1545ATs on a lower voltage (2.5-3V) and returning the I/V resistors to this supply voltage direct from the hexacap. This violates the output compliance spec for the chip (VDD-1V) but seems to work and sounds lower noise than having a separate supply.
32fs EIAJ is fed in, balanced, from the left, the central hexacap feeds two TDA1545ATs. Filters are TDK inductor based, six stage balanced with X7R caps (gasp!). Caps in the centre provide the mid-rail power (2.5V) which the I/V resistors are fed from. The two hexacaps on the right power the post amps based on AD815ARBs, gain set roughly to 26dB with a little NOS droop correction. They're the postage-stamp sized boards roughly in the middle. Output is differential and fed to ferrite-cored transformers to convert to SE.
I'm blown away by the dynamics of this, given the right recording. Next up is an amp based on paralleling 8 or 10 AD815s. Each fed from its own hexacap, of course....
Update1 - the DAC supply has now been upgraded to two tiers of the 4700uFs, interconnected by six 1.5mm diameter solid copper wires. Capacitive reactance measures 2.5mohm @ 50Hz and ESR is below 1mohm at this freq, falling at higher audio freqs to around 0.5mohm. My LCR meter's probes don't make consistent contact down at this level so the reading is subject to a fair degree of uncertainty. Adding the extra tier improved the perceived LF noise so it seems I haven't reached the limits of the chip yet.
I also found the supply used to terminate the I/V resistors benefits from lower noise. I started out with just 6000uF on this, upped it to 20,000uF and figured I'd be better off if I was using the DAC's own supply rather than a separate (and bulky) bank of caps. So I'm now experimenting with running the TDA1545ATs on a lower voltage (2.5-3V) and returning the I/V resistors to this supply voltage direct from the hexacap. This violates the output compliance spec for the chip (VDD-1V) but seems to work and sounds lower noise than having a separate supply.
Total Comments 16
Comments
-
And here I was, thinking it was a Jackson Pollack ...
I can always rely on you, Richard, to provide me with some light relief on this terribly, terribly serious forum, ;).Posted 20th March 2014 at 05:33 AM by fas42
-
Thanks Frank
Its terribly serious because they're all barking up the wrong trees rather than having fun following their curiosity. Audio is terribly religious isn't it? Posted 20th March 2014 at 05:36 AM by abraxalito
-
You're not kiddin' ... !!
With this setup are the raw voltages coming from SMPS? And, have you played with, or considered using the Jung regulator or similar, which gives ultra-low impedances at LF? As a comparison, perhaps?Posted 20th March 2014 at 06:05 AM by fas42
-
Raw voltages are coming from my bench supplies (two dual output ones) which are linear, cheap Chinese ones. Since my experience with the chipamp I'm rather wary of regulators - adding caps to that (LM338) still improved the sound, despite the DS showing impressively low Zout at LF. Similar experience with shunt regs - I used a transistor assisted TL431 which according to sims should be in the region of 30mohm, adding caps (20,000uF) to that still improved it. The shunts were powering TDA1387 DACs.
Tentative conclusion - Zout definitely matters but its not the metric which tells how good a reg is for audio. Peufeu's idea of seeing how much distortion is generated when feeding a signal in is on the right track I feel. But realistically a reg should be fed noise in and then see how much noise it spits out.... My hunch is this is the acid test and the real reason why feedback is avoided by those seeking the ultimate SQ...Posted 20th March 2014 at 06:37 AM by abraxalito
-
Okay, the bench supplies are an unknown quantity in terms of how 'dirty' they are, the hexacaps may be doing quite a bit in quashing the noise coming out of them. Since you have so much raw energy storage in those arrays, you could do an experiment in running the setup off effectively battery power: have all warmed up, play a key track, and halfway through disconnect the bench supplies, and pull the mains plugs on them. Does it make a difference, either way?
As regards Zout, you most likely have these already: [URL="https://waltsblog.waltjung.org/?page_id=533"]References & Regulators WaltsBlog[/URL]. If not, [URL]https://waltjung.org/PDFs/Regs_for_High_Perf_Audio_2_B.pdf[/URL] gives precise, actual measurements of performance here - these figures are what I've used as a reference point ...Posted 20th March 2014 at 07:13 AM by fas42
-
I have series inductors between the supplies and the caps, to act as LPF and so as not to compromise the stability of the supplies. I could do this experiment just by turning off the supplies and disconnecting them - the caps play for long enough to listen for a while. But I'm not curious about this so I won't do it coz I am totally uninterested in experiments listening for differences.Posted 20th March 2014 at 07:52 AM by abraxalito
-
Interesting comment about being uninterested in differences - unless you misunderstood what I meant. Differences to me say that there is an aspect to the system environment which is affecting the sound: if you have sound A and B, then either A is correct, or more correct, or B is, or they are both roughly equally distant from optimum sound. Usually either A, or B, is better, which I determine by using difficult recordings - which variation is delivering the more cohesive, better resolved reproduction.
I should now have better understanding of whether I still have a weakness in the reproduction chain relevant to that aspect, which guides my next step. [I]If[/I] there is [I]no[/I] difference in sound that is the ideal result, because that area of the system is sufficiently 'robust' at this stage of the refinement ...Posted 20th March 2014 at 10:15 PM by fas42
Updated 20th March 2014 at 10:18 PM by fas42 -
Listening for differences is a mug's game. I'm only interested in improvements and those can't be told from short-term listening. As you note it needs lots of musical examples.Posted 21st March 2014 at 12:50 AM by abraxalito
-
I guess everyone does their listening in different ways - for me it is almost instantaneous, the system is either in the 'zone' or it's not ... I guess my brain's got so attuned to listening for the sort of faults that cause the real problems, from doing it over so many years, that it takes mere seconds for me to register whether the sound is better or worse - it's become instinctive.
The key is to do things the way that suits the individual - so long as overall genuine progress is made ... :)Posted 21st March 2014 at 02:03 AM by fas42
-
Sure, I'm still a noob at this without your many years of trainingPosted 21st March 2014 at 02:16 AM by abraxalito
-
Ahhh, but I'm sure you're a fast learner ... !! :p, :D
At a serious level, you've mentioned Bitstream recordings being hard to take - and I know what you're reacting to, there's a very fine edge, where they can easily cross into quite unpleasant territory to listen to - that's the sort of recording I would be using to assess progress made; how do they shape up, situation A vs. situation B ...Posted 21st March 2014 at 02:28 AM by fas42
-
Not Bitstream, rather DSD. They don't really sound unpleasant, just they're veiled at the top end, lispiness on vocal sibilants. Bass dynamics though sounds great in contrast to HF dynamics which suck big-time. And then on top of all that is the 'auditory smoke' which is the artificial spaciousness effect - that's not unpleasant either, but no substitute for real spaciousness which is not tipped up towards the HF.
Seems we have rather contrasting styles of kit assessment, me I prefer to play the most dynamic recordings rather than the most problematic ones. To me an improvement to a design of DAC results in greater differences between recordings. More transparent glass means more contrast is available, provided its there at the source.
On reflection I do recall one 'Bitstream' recording (Philips, Brendel playing) and that had similar characteristics to DSD - washed out top-end dynamics, squashing (or severe softening) of the intense percussiveness of piano. Rather like multiple generations of analog tape.Posted 21st March 2014 at 05:11 AM by abraxalito
Updated 21st March 2014 at 05:21 AM by abraxalito -
My mistake, I thought you mentioned it several times, :).
Different ways of assessing is good, I predict in the end that the final outcomes, in terms of system performance, will match pretty well. I favour the problematic approach because it gives instant feedback ... and, my history has been that of listening to albums of popular music, that most would regard as poorly recorded. My pleasure is to 'rescue' these, because the transformation in subjectively perceived dynamics is quite something to behold.
Since my desire is to enjoy music, irrespective of its origins, I work towards extracting all the positive elements in what has been laid down - the "no bad recording" thing - and I've found that has always worked if I put enough effort in. Standing back, one could analyse the sound and say one recording is better than another in various ways - but subjectively it doesn't seem relevant, I can lose myself in the music of the "poor" recordings - and that's what's important to me ...Posted 21st March 2014 at 06:36 AM by fas42
-
Perhaps our approaches differ because our preferred tastes differ. I listen 90%+ of the time to classical music, there are nearly always abundant versions available of a classical piece so I'm able to make choice on recording quality AND musical performance combined. In general, Decca presses mostly the right buttons, EMI to a slightly lesser extent but usually very satisfying. Older Sony (CBS) can sound great, but newer ones are DSD and I'm less likely to listen because the dynamics are what gives me goosebumps and DSD doesn't deliver. Ditto Telarc recordings. DG is rather a mixed bag, sometimes almost up to the best Decca (often the oldest ones, pre-digital master tape), other times DSD-alike.Posted 21st March 2014 at 06:56 AM by abraxalito
Updated 21st March 2014 at 06:59 AM by abraxalito -
A nice rundown - interesting that Telarc, supposedly the dynamic kings are failing to deliver ... but I have a couple of their, classical orientated, recordings which are downright weird: obviously designed to make the typical, expensive audiophile rig sound good - and the perspective projected is very strange, a highly manipulated FR and acoustic.
Recent Yehudi Menuhin is very good for testing - I find his late violin playing barely tolerable, and the instrument tone he projects tests the capacity of the system to bring forth any postive aspects to his sound, :D.
A special delight is finding totally obscure Eastern European classical recordings, on el cheapo, no name labels, from the bargain $1 bins - with magnificent performances on them. I have one where the violinist makes Yehudi sound like a pathetic, high school wannabe ... ;)Posted 21st March 2014 at 11:04 PM by fas42
Updated 21st March 2014 at 11:08 PM by fas42 -
Eastern European can be pretty good, as can Russian, Telarc does sound quite weird sometimes, especially so after they went over to DSD. Fairly typical (psychologically) that the ones shouting the loudest about something aren't actually the best at doing it. 'He who knows does not speak, he who speaks does not know'.
I plan to pick up some 'job lots' of classical recordings copied on to DVDs by a few Taobao sellers, I noticed one has a nice collection of Eastern European/Russian titles....Posted 21st March 2014 at 11:17 PM by abraxalito
