CuteStudio
Active Member
- Joined
- Nov 23, 2017
- Messages
- 163
- Likes
- 93
Oh I don't know, I'd love a lower noise floor...
... currently they are digging up the road outside...
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
New 28-bit DAC coming out.
- Thread starter GXAlan
- Start date
... currently they are digging up the road outside...
KSTR
Major Contributor
For those who have read the paper, it is important to emphasize how much effort is required to realize the all-import analog switch that disconnects the high-path once the signal level is low enough, plus the DSP pre-processing, including a feedback means to measure and compensate the individual gain and offsets errors of the paths, for a glitch-free and seamless transition.
This is the main difference compared to a multipath ADC where all that delicate stuff is in the digital domain, much easier to handle. The analog section only has to make sure that the high-gain path is actively (soft-)clipped with instant recovery before entering the ADC chip.
This is the main difference compared to a multipath ADC where all that delicate stuff is in the digital domain, much easier to handle. The analog section only has to make sure that the high-gain path is actively (soft-)clipped with instant recovery before entering the ADC chip.
I've no doubt it is an amazing achievement.
Just utterly pointless for audio reproduction purposes.
It's my dream to see HDR-A architecture applied to every element in the audio signal path: microphone, preamp, ADC, (32-bit-perfect DAW), DAC, and power amplifier. This will improve today's best audio capture and delivery by 40dB. You are right -- for common home / consumer audio it may be "utterly pointless." That's another conversation (which I believe I would convince you otherwise, but for another time -- see THD+N graph about 20 comments ago, an objective, measured, perceptual improvement via HDR-A).
In pro audio, noise build-up is still a problem, especially now that most people are listening on some kind of headphone. Headphones can attenuate room noise by 30-35dB, or even more. Which brings our real-world listening experience back to absolute threshold of hearing (-8dB SPL @ 4kHz). That's my baseline, and HDR-A architecture takes us to this baseline -- or very close. For professional engineers, this is a welcome improvement. On the other side -- the loud side -- recording engineers put microphones on snare drums and trumpets every day. Very common. These instruments (and others) can exhibit peaks in the 155dB SPL range. Today, we attenuate these signals before the micamp, to not overload the front-end or signal path. But with HDR-A, there is no need for attenuators. In fact, with HDR-A, there's no need to set gain levels. The systemic dynamic range meets or exceeds -8dBSPL to +155dBSPL (163dB or 27 bits).
I will predict that, one day in the far future, most professional audio engineering will adopt multi-path HDR-A architecture. Like all tech growth, I think it's inevitable. But like most new tech-paradigm shifts, it starts slow (expensive), followed by an adoption curve (the "S" curve), with a rapid ascent after some slow initial phase. For HDR-A, that "initial, expensive phase" is due to discrete realization -- i.e., it takes a LOT of components to achieve this architecture. The D-1 DAC has something like 1,500 components. But here's the thing -- all of these HDR-A sub-functions (DSP, gating, attenuation, calibration, feedback networks, offset control, etc.) can ALL be realized in silicon. At some point, this entire architecture will be realized in an IC, or perhaps a chip set, which reduces a DAC component count from (say) 1,500 to 150, and a massive cost reduction.
It's at these cost-performance inflection points that design engineers then have a choice --- do I design with IC's that give me 120dB dynamic range, or IC's that give me 160dB systemic dynamic range, when both devices are essentially the same price? The choice is obvious, and signals the beginning of a new design paradigm. You might enjoy a keynote lecture I gave at the 2013 AES Conference on this topic. It was picked up and made into a feature article by Stereophile. https://www.stereophile.com/content/audio-engineering-next-40-years
Last edited:
Digital_Thor
Addicted to Fun and Learning
Had 10Vrms in my Pre-amp for years. Combined with a lower gain amplifier, it does seem to lower hiss in my active system compared to 2Vrms from for example minidsp.
I think it's definitely an idea to focus on specific user cases, rather than on specific details.
I think it's definitely an idea to focus on specific user cases, rather than on specific details.
I can understand why - it would be *very* profitable.
(just a little fun - I'm sure that's not your only motivation)"Measured", I agree. "Measured perceptual" you seem to be lacking some evidence for.
I can accept you might have a market in music production. But music production has been managing well for decades to produce music with noise below audibility with current 20 to 24 bit capable gear (in reality - even with 16 bit reproduction). And even in older recordings, if noise is just barely audible when cranking the volume - is is not sufficient to have any impact on enjoyment of the music. (That is opinion - I have no more hard evidence for that statement than you do for yours)
You might be right But if so I suspect it will be more due to ease of use (eg not having to compress incoming signals prior to digitisation etc) than delivered dynamic range in the finished product, for which current 24 or 32 bit DACs with 20+ bit output dynamic range is more than sufficient for human perception.
Unless the technology is pushed at end users on the basis of "FUD/FOMO" marketing that is so prevalent in the audio world. See (the thankfully dying on its knees) MQA - and thousands of other pointless 'products' in the audio reproduction world. I'm sure you'll forgive my scepticism.
Last edited:
That can work as long as your pre-amp noise/dB is lower than your amp noise/dB
KSTR
Major Contributor
Multipath (Mic-)Preamp + ADC is already a long established reality in the pro world. German pro audio company StageTec in Berlin (founded 1993 by ex Neumann engineers) sort of pioneered (and patented) multipath MicPre+ADC under the moniker TrueMatch XMIC+
Also quite a few field recorders use multipath ADC, and even guitar pedals etc, solving the headroom vs. noise issues, notably for live recording.
I am convinced multipath DACs will eventually take over once the implementation is cost- and space-effective (single chip solution, in the end).
For many applications in the near future, intelligent paralleling will remain the more effective way of increasing SNR. Take one excellent DAC chip to start with (say, ES9039q2m) and parallel 16 of them, get 12dB more SNR. Intelligent means all the DAC channels get slightly different input data (gain & offset) so that all the D/S modulators are decoupled and hence any systematic/correlated noise and artifacts average out as well. For consumer / end-user application this certainly covers all grounds (if not already covered).
In audio production (or for measurement gear) even more SNR will be welcomed, if only for convenience (don't care about level range switching anymore).
My amateur opinion if made cheaply into a multichannel IC it would appear in home theater products where you throw away massive amounts of SNR to accommodate speaker level adjustments and multichannel EQ etc . And for convenience/sloppy programming the software is quite wasteful with levels it seems .
In my desktop setup I "waste" -3dB for clipping protection and the EQ itself another 6-8 dB as most is bass boost and thus the midrange is relatively lower in level.
But i think ordinary products handle this consumer use case good enough for all humans .
Listening to reverb tails on a DAW with 32 or 64 bit floating point source material in the studio , another use case .
I can apricate that the levels are all over the place in a studio before producing to a coherent mix
We consumer are stuck with 16 or 24 bit source material ( practically something else almost nothing is true 24 bit in practice the low bits are all random noise ).
Are not consumer source material already pre conceived with the same "problem" as this architecture solves .
Are not my mundane DS DAC's already better at -40dB than the actual CD's or files i want to play ? Meaning that my current DAC's are already transparent visavi the source material itself ? And even if I burden the system with a 10dB SNR loss due to room EQ etc.
In my desktop setup I "waste" -3dB for clipping protection and the EQ itself another 6-8 dB as most is bass boost and thus the midrange is relatively lower in level.
But i think ordinary products handle this consumer use case good enough for all humans .
Listening to reverb tails on a DAW with 32 or 64 bit floating point source material in the studio , another use case .
I can apricate that the levels are all over the place in a studio before producing to a coherent mix
We consumer are stuck with 16 or 24 bit source material ( practically something else almost nothing is true 24 bit in practice the low bits are all random noise ).
Are not consumer source material already pre conceived with the same "problem" as this architecture solves .
Are not my mundane DS DAC's already better at -40dB than the actual CD's or files i want to play ? Meaning that my current DAC's are already transparent visavi the source material itself ? And even if I burden the system with a 10dB SNR loss due to room EQ etc.
It sounds like next iteration of the PRO chip from SABRE.
KSTR
Major Contributor
More like 2+ generations, I feel. A more feasible solution could be a dedicated chipset of two or more chips because the whole system isn't just a DAC, it is DAC + ADC + DSP + precision analog circuitry and the latter is where the integration problems will manifest.
kelesh
Active Member
- Joined
- Feb 26, 2021
- Messages
- 192
- Likes
- 231
And the obligatory nod to the early days of computing, in the form of those ubiquitous gray ribbon IDE/ATA/floppy cables.
EERecordist
Addicted to Fun and Learning
- Joined
- Feb 24, 2024
- Messages
- 971
- Likes
- 1,854
I did not work in display industry research, but I had many friends doing that, I went to conferences, and followed the literature. Their AES is SID, the Society for Information Display. We have an active local chapter with regular presentations on information display and the visual system. One of our guests was Brightside Technologies, a spinout from UBC in Vancouver, Canada. They were bought by Dolby Laboratories as an ingredient for Dolby Vision. I knew several engineers in research at Sharp Labs when Japan's Sharp did research for their TVs, projectors, industrial displays, and medical displays. The most critical displays are in medicine, primarily radiology, and in color management for movies. Several of the Sharp Labs people I know went to Dolby. Dolby is also a great resume data point for friends in the brand advertising world too, with a 90% gross margin, you can afford to spend on brand. I used and continue to use for archiving their A-system, DBX tried to get into the pro world, but not very successfully. No comment necessary, just an observation. Add, let's hope Audio Precision and Rohde and Schwarz, and other test equipment vendors have multipath roadmaps.
I used my Benchmark HPA4 preamp with the Imersiv D-1 and it seemed to restrict the sound. I sold the HPA4 and I now go direct into 1 of 3 amps with the D-1. I have 3 systems. I am going to buy a passive preamp switchbox that Requisite Audio is building for the D-1. I purchased the D-1 from them. Imersiv has something called SBB that allows other sources, but I do not think it will be as good as the passive that Requisite Audio is building. I have heard the prototype on my system, it still needs work, but it does not make the D-1 sound worse, like the HPA4.
Sans the preamp, the D-1 is glorious. The best my system has ever sounded. I have owned the Benchmark DAC1, DAC2, and DAC3 and these are not even close to the sound I am hearing out of the D-1. I am a huge fan of the LA4 + HPA4 preamps. I have owned one for more than 10 years. However, the D-1 is the first source that got worse after adding a Benchmark preamp. That was a bit shocking.
Since I have no money, I sold 7 pieces of audio gear to buy the D-1. At this point I am thinking this thing is rather cheap for what I am hearing.
I listen to my 3 systems at home from 10-18 hours a day while I work. I now have a lot of hours with the D-1. I also have a RME DAC that I have not hooked up for weeks since I got the D-1. I sold the following 2 DACs (out of the 7 items), Schiit Yggi+ OG and the Schiit Yggi+ MIB. I loved the sound of the OG but the D-1 made that DAC expendable.
Last edited:
Similar threads
