ack's system - end of round 1

PeterA said:

Wow, I can sense your enthusiasm from the post. Congratulations, Ack. I guess you will not be buying a new DAC for a while. So now I'm pretty curious, do you prefer digital or analog these days, and how would you describe their differences in sound?

Click to expand...

I am going to keep you in suspense for now, Peter. These subjective relative evaluations take a long time to be made, and it's hard to find the same recording on LP and CD where the mastering is the same. I think the Mahler #2/BSO/Philips qualifies for that, but I need to listen quite a bit more. For one thing, I pay particular attention to the attack, body and decay of timpani, and they now sound pretty much the same, and with all the glory we hear at Symphony Hall. Seriously.

But I can tell you, there isn't a single CD that I've played so far, where my reaction wasn't of the same disbelief. Also, I have mentioned before the CD from Chesky at the bottom, where the piano is real b*tch to reproduce well. In particular, track 5 starts with a heavy-handed note, and immediately following it there has always been a woosh of noise (like white noise) before the next note (a second later), also present with the Rossini player I had in for evaluation, and it's now gone (it's still present from the unmodified balanced out). That was my first cue that there is something really positive going on here. Then, track 1 is so damn clear and powerful! Back to the Mahler, another thing I was looking for is smoothness of the voices of the chorus and the organ in the finale, and it's all there - but I think the Rossini still did better in rendering a more appropriate size of the chorus and organ, so this mod is not a giant killer, but rather a giant _noise_ killer. I am definitely hearing new details in many recordings.

Like all of my mods, I have been thinking about this one for a long time, probably 3 months now, and I am now itching to study the new/modified output section of the Berkeley Alpha Reference 2.

Steve Williams said:

I laud you in your quest. What an amazing tweak all for $2.00

Click to expand...

Thanks Steve. Like most mods, the real expense is in the time spent reasoning and researching, not the parts (the speaker mods being an exception - about $10K in parts alone). I am listening to the Mahler again right now, and next time they play it at the BSO, the Boston group needs to go together. The sheer dynamic explosions alone, live and now at home to a relative degree, are something to behold.

I haven't looked at the spectra with a scope; and no info on output impedance from Berkeley. Also, I wouldn't describe anything as a design fault, rather, as I said, most likely the interaction between it and the ultra-wide bandwidth of the Spectrals. Nonetheless, a low-pass filter does exist in the output, by design.

YashN said:

Very cool mod, ack.

You may be further interested in some findings shared by a Mr. Rasmussen over at diyaudio, whereby he recommends a cap at the DAC output which technically 'overfilters' slightly, with a drop already apparent at 20Khz but the sounds takes on a very good SQ. This was independently verified by other people, including doubters (I haven't tried yet).

If memory serves me well, he did talk of high frequency hash or oscillations that his mod tamed.

Look for "The Rasmussen Effect" over at diyaudio, long thread, with the usual naysayers, but very worth a read through and perhaps if so inclined, further testing in your setup.

Click to expand...

Thanks. There are 60 pages in that thread, lots to read.

YashN said:

For sure, but you can skip the negative posts, making it more convenient.

Click to expand...

I didn't have to go too far to find what you are referring to; the mod is aptly described on page 1, with the meaty follow-up by Ken Newton (who's also here) on page 3. It appears to be applicable to delta/sigma DACs (like the Alpha) and they suggest a single capacitor after the DAC (as you said), perhaps a couple of resistors as well, to tame high-frequency noise, at the expense of treble rolloff at 20kHz.

The main post about the mod is here http://www.diyaudio.com/forums/lounge/249418-dac-filtering-rasmussen-effect.html#post3781628 and I highlight some key parts:

So far, this is expected to work only on delta-sigma DACs and may be related to noise shaping and dither functions of this type of DAC, which totally dominates the market place. I am yet to see it work with NOS DACs and Ladder DACs.

Find the implementation that suits your situation (DAC etc) best, then vary the value of a single cap to create the low-pass effect and find the sweet spot. I have successfully used -1.3dB @ 20KHz to work well with Sabre DAC (in "current" mode with very low few Ohm Z I/V) and Cirrus Logic DAC ("voltage", but with 1:1 transformer and in this instance adjusting the Zobel network on the Secondary).

...

As far as I can tell, this is applicable to differential (two phases out) delta-sigma DACs, whether "current" or "voltage" types. It comes in the form of a single capacitor across those two [+] and [-], but generally, I would think, in most cases also the addition of resistors (2) are needed.

So it is an add-on filter, one that is "dominant" in the sense it will come in at a much lower frequency and thus dominate even other filtering higher up. The lowest frequency filter is the dominant pole. But the roll-off is very slow and likely only to be circa -3dB down at 40KHz.

Click to expand...

Ken's thoughts on the mod, describing the sound, are here http://www.diyaudio.com/forums/lounge/249418-dac-filtering-rasmussen-effect-3.html#post3781866 and here are some highlights:

I have been subjectively evaluating the 'Rasmussen Effect' over the past few days. In my initial subjective description of the effect in the Oppo thread, I had noted that I hadn't yet decided whether the high treble was being unacceptably sacrificed in order to gain the effect. However, after further listening evaluation, I now feel that the effect produces a high treble that's much more like that produced by live instruments. To my ears, live music rarely sounds 'bright'. Instead, instruments such as bells, brass, and violin have a kind of great dynamic clarity which cuts through the sound of other instruments, yet they rarely sound 'bright' to me, at least, not in the HiFi sense of that word.

...

I'm finding that a -1.5db @ 20KHz point better removes the upper midrange glare than does a -1.3dB point. With such seemingly minor frequency response differences in 1st order filter response producing such relatively significant changes in perceived tonal balance (including in the bass), I feel that there must be more responsible for the effect working as it does than could be due to simple frequency domain filtering. I suspect there may be some sort of time-domain related effect at work here. The effect itself 'turns on' quite abruptly as the output filter response is only slightly changed. As Joe has pointed out, there seems to be a sort of 'knee' point in the activation of the effect.

Click to expand...

In the sister Oppo thread, apparently somewhat related to it all, the following statement is made and points to the delicate nature of the mod:

What both Coris and I are saying, to low a value cap and it gets worse, but at a certain point, increasing the cap value and suddenly, rather than getting even worse, it turns around and gets better. It is not intuitive and hence confusing.

Click to expand...

This calls for a very careful approach. Nonetheless, interesting stuff, thanks for pointing me to it!

What do we really consider "natural" sound?

What do we really consider "natural" sound?

I've been pondering this question for years, and I have come to a conclusion.

But first, let me diverge a little bit and ask: Never mind sound, let's ask, what is really a "natural" visual experience? Don't laugh, it may not be as straight forward as you think, so a simple "everything we see" is not going to cut it for me. When we look at the stars with a naked eye, is that a natural visual experience? I'd say Yes, but to an incredibly minute and highly irrelevant degree. Why? Because I don't see really much, other than tiny dots in a big wide space; collectively, they look beautiful, but each individual one is as boring as a pig. When the first pictures of Saturn's rings were published decades ago, was that a more natural visual experience, albeit only two-dimensional? What about subsequent pictures showing even more rings, to the point that we started assigning numbers to them? What about even finer pictures since then, depicting things that looked like rocks in the rings, but we still couldn't tell what the rings are really made of. What about Cassini's crystal clear pictures from recent missions? We now know there is so much more going on with Saturn's rings that what we originally thought, or what is observable with the naked eye. Which one of these visual experiences is more "natural" and why?

But enough with the allegory... back to music. What if we ask similar questions about musical instruments and the sound we experience. Is the sound of a live violin as we experience it from 20m away "natural"? How about 100m? What if you put your ear right on it, much like a microphone, or a violinist. At what [far or near] distance does it stop being natural? How do room reflections change one's perception. Is the violin's sound perhaps un-natural if you stick your ear to it? Is it bright? Aggressive? At which distance does it lose its presence? Is it always sweet? How far into a Hall do you have to sit to lose the sense of naturalness and presence?

So the conclusion I have come to is that, to me, it's primarily about timbre. If I can distinguish and recognize an instrument from any distance and still deem its sound realistic, as if I were a musician, then that's natural. To that extent, I use timbre to identify instruments and their degree of naturalness - not presence, not dynamics of any kind, nothing else. As such, everything else is secondary and perhaps unimportant, and MAY include the room, reflections, position, relative distance to other instruments, dynamics/loudness, et al. This definition of naturalness covers having your ear right on it - like a microphone - as well as being as far as possible to still conclude timbre is real. More than that, the close-up sound of a violin has tremendous resolution and macro- & micro-dynamics, which is not necessarily true if you are sitting far far away, especially in a large hall; it just isn't the same, yet they are both natural to me, but to varying degrees and for varying reasons. The same goes for just about any instrument, other than perhaps organ.

Which brings me to the next topic: is what I am hearing at home, from my system, natural? And how does my analog fare in that department against analog, and vice versa - that's what PeterA effectively asked upthread. Well, the only answer I can give right now is that it varies, much like the visual analogies I gave above. Whereas before my analog was far superior to anything digital with just about every recording, this is no longer the case. With things like strings, I still prefer my analog in most cases; probably voices too; with a lot of other instrumental recordings available both on CD and LP, I would say it's like splitting hair and it may vary on the mastering. And when it comes to RBCD HDCD, my digital is now far superior to analog with the same Reference Recordings, period.

So what is going on here... well, the latest DAC modification, which really builds on top of prior analog-section shielding documented in these pages, has dropped the noise floor so much that the gap between my analog and digital is a virtual tie. Part of it, I think, has to do with the slight residual hiss in the XP-25 and LP surface noise. So I am predicting that the new Spectral phono - when God blesses its release - or similar will tip the balance yet again. However, again, HDCD is so clearly superior right now, that we will have to wait and see how a better phono stage will do. Folks who have heard my analog understand how good it is, and therefore, it may come as a surprise what I've just said about HDCD. Consequently, my modifications have tremendously improved timbre and the sense of space in the Alpha DAC; regarding the latter, instrument positions in the stage are very clearly delineated, set in jet-black background; and I am certainly hearing tiny things I have never heard before.

The reality as I hear it is this: my analog sounds really good, yet a little rounded off and polite; you get the kind of information (and I am talking about timbre, micro and macro dynamics) as if one were standing at a fair distance from the instruments, certainly not on stage, but not too far away. HDCD sounds much closer to what the microphone would pick up, and that includes all the additional micro-detail and micro-dynamic information I could care for, as well as wider dynamic headroom. It's incisive with close-miked recordings, yet still natural by the definition I gave. Yet, it can still be a lot of fun, like Sunday At The Village Vanguard. Effectively, I am getting varying degrees of naturalness from both media, and I would argue that's a great thing. Frankly, these comparisons no longer make sense, if timbre and other important characteristics can be perceived to be relatively true to a highly believable degree. At the same time, we must not lose all perspective, and neither type of sound is the best one will ever hear; far from it. The intent here really is to convey only the varying degrees of naturalness that I am hearing...

The conclusion is basically this: I am finally now having fun with digital, as well as analog. More on the latest mod and its audible effects coming soon...

-ack

Yeap - the title of this thread, Timbre & Articulation

Thanks, and I agree Peter - yes, "range of believability" is appropriate. Exactly why exposure to live unamplified music is so important. Equally important is to remind ourselves that what we hear is not necessarily the same as what the microphone "hears", as discussed many times on the forum - especially when it comes to close-miked recordings.

microstrip said:

It seems to me we are just turning in round when referring that timbre is the number one for sounding "natural" or "believable". How do you defined timbre technically? I have listened to many systems that are correct in timbre and that IMHO do not sound natural at all.

Click to expand...

This may be a subject for a new thread, but the best definition of timbre - basic and technical - is on wikipedia. Your definition of naturalness is obviously different.

microstrip said:

IMHO the critical questions are what are "the other important characteristics" and how are they connected to our equipment?

Click to expand...

I referenced some, like micro- and macro-dynamics, and presence. For example, I don't need to hear the power of the violin the way the violinist hears it, for me to be able to tell that I am listening to a violin. I can tell by the tone (timbre). And let's make sure we recognize that I mentioned the "other important characteristics" in the context of comparing analog vs digital, and said that if they are all highly believable, then the comparison makes no sense anymore.

What you and micro describe is basically my looking-at-the-stars analogy - natural but extremely uninteresting. I have to agree, but at the same time again, just degrees of naturalness. The extreme you described is not the type of naturalness I would care for either.

Berkeley Alpha mod - the sound

I wanted to chronicle the vast improvements I've been hearing with the Alpha mod sooner rather than later, because I am about to listen to the 4000SV player soon with the same Spectral electronics (but different speakers), and I think it will be an interesting comparison. BTW, the Vishay 1837 are in, and I can't say I hear any difference from the 1839, at least they have not been obvious.

1. There isn't a single CD I've played so far that has not tremendously benefited; transfixed CD after CD
2. The increase in vividness and you-are-there feeling (with the right recordings) is arresting
   1. palpable images
   2. jet-black soundstage, even deeper than before
   3. instruments in the back of the stage easily discernible; all very well defined around the stage
   4. uncongested stage, can easily tell musicians re-tuning their instruments in live recordings while the audience is applauding; can easily tell a singer turning to another and whispering something to another, while the audience continues to applaud
   5. really fast sound - very quick rise and settling... are the Berkeley guys using JFETs????
   6. exceptionally low noise - e.g. Krall Live in Paris: who the heck is coughing all the time. New information is popping up everywhere, contributing to the vividness
3. Larger dynamic headroom as mentioned before, seemingly uncongested dynamics - passed most orchestral with flying colors, totally killed my Guilmant Symphony #1/Organ, Orchestra/Chandos wrt scale, much like the Rossini player I had in
   1. Special mention to the extra-ordinary Dorian recording of Proteus 7 "For Your Ears Only" (unfortunately long out of print); total killer dynamics of the trumpets and other winds
   2. Remarkable rendition of my Mahler #2, BSO/Ozawa/Philips, probably as good as the LP if not better (this is an original digital recording)
4. Significantly improved timbre with every single instrument - piano like this DAC has never done before. Really tight, powerful and well-rounded bass. very crisp trumpet micro-dynamics
5. Very literal sound - e.g. you can easily hear the air blowing - the "pfff" - with closely miked flute
6. Has largely removed a dryness in the Alpha sound, transforming the sound to very sweet, with the right recordings
7. I can easily go louder and louder without fear of excessive distortion; there is an uncanny ease to the sound

As compared to my analog, it's everything I said before and more:

1. RB HDCD is something to behold; it can beat digital Reference Recordings on LPs
2. It still cannot beat my best analog recordings, especially the old analog Reference Recordings
3. Mastering is a huge factor in the format I end up liking - e.g. Bruch Scottish Fantasia/Oistrakh/LIM XRCD24 sounds so boring with limited presence compared to a Decca LP re-issue, total night and day
4. Strings and voices still sound more natural with analog, but some Chesky vocal (Rebecca Pigeon) sounds strikingly real as well (haven't tried opera yet)
5. Overall, I now enjoy both formats equally, for the first time

I will keep the filter at the current 193kHz -3dB point for now, for fear of introducing phase issues if I go lower. Interestingly enough, madfloyd emailed the other day that his Soulution preamp has a 50kHz input filter for digital; well, I can now understand why, but it feels too low thus aggressive.

Bottom line: this is like a whole new DAC. I would like to compare it at home against the 4000SV and dCS Rossini again, but that won't be possible because I have suspended my customer ties with Goodwin's High End after repeated insults and attitude by its owner.