Of course, this was DS with the original firmware, and the DS has moved up a few notches since, so my hypothesis of progress in digital technology being the great equalizer may have been invalid anyway.
Exactly my point. While all components are evolving / getting better over time, the biggest leaps are in digital playback. Amps donāt get way better year over year, nor do preamps, power conditioners, etc. IMO digital performance OTOH improves close to the rate of PC performance gains. Iād bet there are $1k DACs that sound as good as DSD with original software. Obviously we all benefit because Ted & Paul continued to upgrade the software.
If I bought a $6k DAC and it was equaled 2 years later by a $3k DAC I can live with that. If I paid $20k for a DAC and a few years later that performance was available for $6k Iād be pissed at myself. Itās why I rarely buy anything new, though I did with both the DMP and DSD. I will absolutely audition the TSS but I would be hard pressed to pull the trigger knowing that level of performance could be available a few years later at 50% less. Others will feel differently.
JMHO.
Based on my recent visit to RMAF, I have come to the conclusion that the biggest leap in SQ can be achieved with active digital crossovers. This is innovation in the digital domain, but not improvement of individual DACs per se (although this is a factor), but a completely different architecture. The reason I am saying this is the best system I heard (ever) was the Verity Montserrat, which was build with active digital cross-over. The build their own DACs so I can hardly imagine they were SOTA (thenagain who knows). They also used a modest digital server. Yet, what I heard far surpassed anything I ever heard and the amount of information retrieval had to be hear to be believed. So I have come to believe that this is the future of SOTA SQ, and while optimization of individual digital components like DACs is important, the real prize is completely re-architecturing the system end to end. Maybe a nice challenge for Paul and crew to bring together his efforts in optimization of the analog and digital piece of the puzzle with speaker design.
If only speakers were improving at a similar rate. They remain the biggest source of distortion and inaccuracy.
2 Likes
Very timely comment Elk. See my post. Active digital crossovers tackle this issue head on. We can only get so far by optimizing individual components.
1 Like
Iām sorry, I missed your post when I was writing. You made an excellent point.
1 Like
Verity Monsalvat loudspeakers no doubt sound awesome but at $325k they should sound pretty good. At some point that technology will trickle down.
I agree with your premise though. The biggest gains are to be made with loudspeakers in terms of digital x-over and eq. Imagine how much better any loudspeaker would sound in a perfect room. Iāve used digital eq components but the level of resolution wasnāt there. Now that was a few years back and weāre talking $1k digital eq. Obviously thereās better/more expensive technology.
This just validates my point. Digital is changing SO quickly that to pay 20k figures for any digital is tough. I heard 5 or 6 year old DCS stuff that was originally $40k that isnāt as good as the DMP/DSD is right now. Thatās not a knock on DCS. When that was built it was state-of-the-art but that status changes very quickly with digital, much like with processor chips.
It is my firm belief the Montserrat (more like $1M - not $400K) will trickle down. Most of the cost is sunk R&D, not physical hardware components. It mode me realize there is a whole different level of performance achievable. Complete surround sound from a stereo set. Sheer audio voodoo. As Ted keeps perfecting the digital processing and Bascom builds better amps, someone needs to build the end to end architecture though and bring this to the masses for 5 figures. Why not PS audio? They are well positioned.
That does seem to the ultimate solution, but I wonder how far something like Roonās parametric EQ, for example, can improve sound.
By way of comparison to a $1M system, have you heard the Kii speakers (I havenāt)?
I use DSP / room correction myself. The Verity system did not use room correction. Just digital crossover.
Fun fact, in room response for my center channel is very poor for some reason (I have an $8K center channel so hardware is not the issue). Thankfully mains have miuch better in rooms response. Yesterday I was watching a movie and I had trouble following the dialogue. I switched to another movie and same problem. I checked my server and found Dirac live DRC had accidentally been disengaged. Switched back on and dialogue fell right back in place. This stuff really works.
Was told by PSA months ago the Octave will not carry convolution filtering; reason I went the Nucleus+ and eventually PF 2.16 route. I hope they change their mind, I love that technology in a streamer and my room sounds so much better. However, it is not perfect solution. The combination of room treatments and CF makes for a really good sounding room.
I enjoy the speakerās name, Monsalvat. Monsalvat is a the castle where the Holy Grail is kept in Wagnerās Lohengrin.
1 Like
I mistakenly called them Montserrat. As you are probably aware Verity has a Lehengrin in the speaker lineup.
I want to highlight why what Ted has done is more significant than the generic kinds of improvement in digital technologies that are being referred to in recent posts. This is just my take based on plenty of reading and a few forum/email exchanges, but hopefully I donāt mess it up in any important respect.
Advances at the level of microchips deliver improvements to everybody in terms of the number of instructions per second which can be executed within constraints of cost, size and electrical power. We needed a certain level of performance to make the DS possible, but adding a faster FPGA to the DS wouldnāt instantly make it better. Nor does having even faster computation immediately give any other DAC a meaningful advantage over the DS.
What makes the DS special is analog. It starts with recognising that linearity is key (meaning a mathematically linear mapping of input data to output voltage), and that the simplest and most effective way to get that from a DAC circuit involves three things:
- 1-bit digital switching between ultra-stable positive and negative voltages
- Focusing on a clockās tick-to-tick consistency not its long term stability
- Oversampling and noise shaping in the digital realm combined with a gentle low-pass filter on the analog output to remove noise while preserving phase in the audio band.
The digital side of this had already been developed and standardised as DSD. Tedās initial inspiration was to take the claims about DSD literally ā you really could just run a DSD stream through a low-pass filter and get an audio signal. The big challenge is to get the most perfect analog representation of that DSD stream in the form of voltage swings between a positive and negative potential.
If all we wanted to play was DSD then we technically wouldnāt need anything like an FPGA in our DACs. In the real world we have mountains of PCM, and nearly nobody had a DSD-compatible player, so a DAC project for PS Audio would need to be able to accept PCM and convert it to DSD. So, there had to be a processor of some kind for that core purpose. An FPGA is a great balance between flexibility and raw performance, making it perfect for this kind of application.
Digital technologies have an Achilles heel in the audio world, though: electrical noise. Thereās nothing smooth about digital circuits. They shake and screech and tug eratticaly on the power supply in ways have audible impacts. They harm linearity by making the +/- voltages from the DSD signal switches not exactly right, or by causing the clock to be less consistent in the time between ticks, by creating RF noise that results in electrical vibrations in audio interconnects, and by sending similar noises into the common electrical ground such that they can impact other components like amplifiers.
Some of the FPGA updates Ted has crafted for us do actually change the computation ā for the same set of digital inputs weād get different digital outputs. We havenāt benefited from an advance in digital technology during this time, Tedās just learned more and more about how to do digital signal processing in ways which correct earlier mistakes, improve accuracy in already-correct algorithms, or tune the results to make better use of the analog hardwareās intrinsic capabilities.
However, itās also extremely significant that along the way Ted has learned how to program the FPGA in ways which reduce the electrical noise that the FPGA creates as it performs its operations. This improves the consistency of the output driver voltages and the clock tick timing, both of which improveā¦ you guessed itā¦ linearity.
Now the TSS. There are a bunch of user-friendly features etc being built into this thing, but in every other respect weāre basically looking at improvements in the analog realm, not digital! More and better power supplies, more paralleled output switches, better transformers in the output stage. And in order to maximise the linearity of those components, the FGPA and all the other fancy but noisy digital components are being booted out into a separate box and the communication between the boxes will be via an optical fibre so that no electrical noise will be able to transfer that way.
Yes, because silicon has improved in the past decade weāll probably see a newer model FPGA in the digital box. But that aspect of the system will be no more than a tiny contributor to the overall improvement we might expect from the TSS over the DS. And similarly, the advance in computational capability generally doesnāt offer any other maker an advantage over Ted Smith and PS Audio when it comes to audio, because computational capability has very little to do with the most crucial aspect of any DAC.
4 Likes
I like it and donāt see anything to argue with. Thatās a fine synopsis.
1 Like
@edorr in principle I agree with you that active digital crossovers can perform far better than analog ones. But my god the cost of a full implementation is prohibitive. Youāre doubling (2-way) or tripling (3-way) your DAC and amplification requirements. And I sure hope nobody goes the DAC - preamp - ADC/crossover/DAC route because thatās just ridiculous. Preamp aficionados would have to live with digital volume control at some level unless they are willing to adjust two or three volume controls simultaneously post-crossover.
I appreciate that itās a trade-off but for me ā and I suspect most people ā the better optimisation lies in 2-way speakers with amazingly good crossovers. We can do room correction prior to the DAC if required.
That is the state of technology now. All I can say is I heard the future and it is active digital cross-overs. At some point a very good (think DS caliber) DAC can be build for a few 100 bucks and the architecture will no longer be price prohibitive. Mark my words my friend.
Heh. OK, noted. You might like to read the rest of this thread again in the meantime.
What interests me most about Tedās architecture is the separation of digital and audio - the big-picture architecture. The connection between them is glass fiber which has no length limits for home audio (unless you plan wrap your house in fiber). But the implications are enormous, not only for sound quality but also for physical placement of the components.
The architecture allows placing all of the audio elements of the system near the speakers but the digital box can be anywhere that is convenient - near the listener position if desired. Theoretically this eliminates all long (and expensive) interconnects - all that is needed is a single glass fiber. I canāt help but think that this is a pattern that will eventually be adopted by other manufacturers as soon as the SQ benefits are recognized. And, since the voltage will be higher at the analog output of the DAC, I canāt see any reason to ever need a pre-amp. From what Iāve inferred from Tedās posts, the analog box will output a signal every bit as good as the best preamp. However to add a pre or not will obviously be a personal preference of each owner. But, without a pre-amp, all of the analog elements will be passive - theyāll just make beautiful music per the bits coming down the fiber.
As for DSP, if active digital crossovers were ever deemed to be absolutely necessary to achieve the next order of magnitude increase in sound quality (say 10+ years from now), Iād think that a digital crossover box could be devised that would accept the output of the digital portion of the DAC and output to 2 or more optical outputs acceptable to the analog portion of the DAC, each for a particular frequency range. This would allow bi/tri amping where the amps could be tailored to the upper, mid, or lower frequency ranges. The technical challenges would be enormous - I understand that DSD is particularly hard to process - but who knows what algorithms will be developed in the next few years. Ted may have had such a plan in the back of his mind because he originally envisioned more than one optical output for the digital box. But, for economic reasons (as well as technical - clock synchronization I think he said) he abandoned that for this iteration.
I really want to see how this turns out. If the new architecture allows saving a few $1000 on cabling, maybe $20K isnāt so unreasonable. But thatās speculation - I may be completely wrongā¦
At some point we will be able to travel to Mars in a few days. Actually, my prediction has a better chance of coming true.
1 Like