Patent 9.590.648 clears things up: “In all embodiments, DSP(120) will maintain (what at least to listener's perception will be) linear, glitch-free path-to-path level transitions at Summing node(104).”
So the goal is not 20-bit+ linearity, but only a high dynamic range. Some psychoacoustic trickery is thrown in for good measure.
The patent describes that it’s not the resistive elements prior to the summing node that are used to mix the DAC outputs with the desired ratio; they both have the ‘same’ value, e.g., 200 Ω, to keep their thermal noise contribution in check and to present a sufficiently low output impedance to downstream consumers. Instead, the amplification and attenuation of the signal happen earlier on, in the amplifers following each ‘multipath’ DAC output.
If you were trying to achieve an ultra-high degree of linearity that approach would further complicate things ‘cause you then need accurately matched resistors in three places (amp 1, amp 2 and the 200 Ω averaging resistors). Of course, as mentioned on p. 41, you can use an internal ADC to perform regular auto-calibration, but that then limits the linearity of the system to the ADC’s INL spec. It reminded me of AN-86 by Jim Williams, in which he uses an ADC to linearize the output of two averaged DAC outputs.
So the goal is not 20-bit+ linearity, but only a high dynamic range. Some psychoacoustic trickery is thrown in for good measure.
The patent describes that it’s not the resistive elements prior to the summing node that are used to mix the DAC outputs with the desired ratio; they both have the ‘same’ value, e.g., 200 Ω, to keep their thermal noise contribution in check and to present a sufficiently low output impedance to downstream consumers. Instead, the amplification and attenuation of the signal happen earlier on, in the amplifers following each ‘multipath’ DAC output.
If you were trying to achieve an ultra-high degree of linearity that approach would further complicate things ‘cause you then need accurately matched resistors in three places (amp 1, amp 2 and the 200 Ω averaging resistors). Of course, as mentioned on p. 41, you can use an internal ADC to perform regular auto-calibration, but that then limits the linearity of the system to the ADC’s INL spec. It reminded me of AN-86 by Jim Williams, in which he uses an ADC to linearize the output of two averaged DAC outputs.
The smallest signal that such a range would be capable of would be entirely inaudible in a living room, and the highest level, system willing, would blow the windows and your eardrum out.
With what exists today, I cant see any meaning with this product other then bragging rights. My guess is that it's actually worse in reality than what already exists. And I'm not sure they can bragg until measurements that prove the specification exists. Are they published?
To put things into perspective... (it's the Lounge... 😉 )
//
To put things into perspective... (it's the Lounge... 😉 )
//
The blending may not be obviously audible but I wonder how the linearity measures. Do you have some data or a spec sheet on that?
If you say 'gain disparity' probably means the linearity and possible even the monotonicity is shot.
Jan
Difficult to say. A simple stepped level measurement vs thd gives a very low distortion throughout the range. Those are rather long test tones though and any artefact at the start will be filtered out if it sticks to one channel through the tone burst. SINAD sees a rise in the noise floor as the signal level rises through a transition zone though stays well better than 100dB through that zone. It’s still monotonic as well. However they’ve done it it seems to work very well to the ear.
In normal testing by ASR ("audio science review") their DAC testing combines distortion + noise. An "inexpensive" DAC like the SMSL Su-1 (<$80 US) has a combined distortion/noise characteristic achieving "19.5 bits of dynamic range". Under all normal usage this DAC is dead quiet, meaning that the Imersiv D1 provides an added 40dB of psychoacoustic redundancy in adding their "secondary" multi path section.
What is interesting (as is questioned in throughout this thread), is that no measurement claims seem to exist for "improvements" in distortion, rather suggested by the exclusion of the word "noise" in "we're about 7-bits better ("noise") performance [40dB] than any DAC made today" warranting their superlative global claim as to be a "paradigm shift".
Notwithstanding noise performance improvements of such psychoacoustic redundancy it is difficult (if not impossible) to imagine that such a 40dB improvement is equally true as specific to its distortion (as also repeated suggested in this thread). This also raises the questions; If noise is already inaudible what contribution can harmonic distortion improvements provide (if any can be heard at all) below the "19.5 bits of dynamic range" achieved by an SMSL Su-1?
IMO the technical performance of DAC's at or above the performance of an Su-1 is well beyond the scope of psychoacoustic correlation to traditional measurements. This also suggests that all descriptive dialog suggesting improvements over that of an Su-1 has no meaningful correlation to traditionally understood measurements. As an aside this also means that ASR seems knowingly wasting their time in producing meaningless comparisons suggesting that traditional measurements are all that is important. What is justifying their rating scale of all identified devices?
This does not suggest that unknown noise mechanisms (defined here as anything not signal as also including harmonic distortion) does not exist as yet to be clearly identifiable and correctable to psychoacoustics. So it is that regardless of such nonsense suggestive of psychoacoustic merit that Imersiv proponents and others might present this does not negate the potential for amazing sonics that such a device or others could provide. So I would like to see traditional tests and hear from those who have heard it. What do I know?
What is interesting (as is questioned in throughout this thread), is that no measurement claims seem to exist for "improvements" in distortion, rather suggested by the exclusion of the word "noise" in "we're about 7-bits better ("noise") performance [40dB] than any DAC made today" warranting their superlative global claim as to be a "paradigm shift".
Notwithstanding noise performance improvements of such psychoacoustic redundancy it is difficult (if not impossible) to imagine that such a 40dB improvement is equally true as specific to its distortion (as also repeated suggested in this thread). This also raises the questions; If noise is already inaudible what contribution can harmonic distortion improvements provide (if any can be heard at all) below the "19.5 bits of dynamic range" achieved by an SMSL Su-1?
IMO the technical performance of DAC's at or above the performance of an Su-1 is well beyond the scope of psychoacoustic correlation to traditional measurements. This also suggests that all descriptive dialog suggesting improvements over that of an Su-1 has no meaningful correlation to traditionally understood measurements. As an aside this also means that ASR seems knowingly wasting their time in producing meaningless comparisons suggesting that traditional measurements are all that is important. What is justifying their rating scale of all identified devices?
This does not suggest that unknown noise mechanisms (defined here as anything not signal as also including harmonic distortion) does not exist as yet to be clearly identifiable and correctable to psychoacoustics. So it is that regardless of such nonsense suggestive of psychoacoustic merit that Imersiv proponents and others might present this does not negate the potential for amazing sonics that such a device or others could provide. So I would like to see traditional tests and hear from those who have heard it. What do I know?
Last edited:
Think you may be misunderstanding me. I’m talking about the cs43198 which has its own dynamic range enhancement multi channel path thing going on. I’ve no experience with this new thing, I’m simply wondering how their new approach compares. There’s clearly differences in approach as the cirrus part uses two equivalent channels at different gains and blends between them while this one is a truncated 7 bit major stage that appears in the diagrams to have its own -112dbu noise floor which is somehow magically nulled by cleverness. I will be awaiting SINAD sweeps with great interest to see how that shows up in the measurements.
(If you are in fact asking about the cirrus chip though I’d say it sounds much like any high end cirrus DAC if well implemented. Accurate and true to source.)
As an aside, it isn't always desirable to eliminate noise altogether. There often exists a desire to set a sound pressure level (or SPL) in a room in relation to random noise levels that could be present there. This can be characterized as the threshold of critical discrimination ("TCD"). One example of TCD might be to set up an SPL level to support the ability to discriminate between the call by a loved one to come and separate out the recyclables from that of a call to supper.
Last edited:
A reproduction system should be silent when it is not reproducing anything.
//
//
When I turn mine off and it's not reproducing something, ya gotta love the silence. I can put my ear right next to the cone, and, dead silence!
It is never really silent, but it is very nice and refreshing bathing in the forest sounds on my long walks (best at night when there is NO traffic.
dave
I think myself a sensitive person Dave. Lovely image above. We also have countless trees in our back yard, along with a couple of enormous maple trees. This creates a beautifully canopy over our deck. Truly serene. You know that maple trees have leaves? Well these things come off in the fall. It doesn't go unnoticed by me, nor can I stop my feelings beginning to manifest of significant guilt when I see my loved one all alone outdoors in the chill of an autumn wind trying to rake up all those leaves that keep raining down on her head and all over the yard. This is while I am taking pleasure in listening to my stereo...yeah...so I send out the dog and close the blinds.
Last edited: