Following the introductions and explanations here, I am thinking along the following lines.
Notwithstanding the - way too coarse and generalized! - saying of ”a DAC does not have a sound, it is the implementation (e.g. amp) and transducer (e.g. IEM) that determine the sound“, we have experienced that IC-DACs have some sort of sound of their own. We do distinguish AKM, ROHM, ESS, TI, and many others - some are ascribed a ”house sound“.
With the N7 now, it occurs to me that one of the principal elements that is determining the sound might be the Audio Bridge. I dare to go out on a limb and suppose that it is an integrated circuit (IC), manufactured by a third party. If I am not mistaken, nothing much has been said (yet?) about its specifics. Would Cayin say more about it?
Why do I ascribe a crucial role to the Audio Bridge? It does the all-encompassing conversion of PCM to DSD in the N7.
Why do I think there might be many ways / algorithms for such conversion? Just as an example, think of the many algorithms for sample rate conversion within the PCM domain.
Just some thoughts…
If I can speak quickly on behalf of Cayin, it's the programmable logic aboard N7's FPGA preceding the audio bridge that plays a big role in how the final 1-bit stream sounds. It's what generates the I2S (data, bit clock, master clock, word clock) into the audio bridge. How it handles that, now and in future upgrades, could play a big part in how N7 sounds.
But there is something inherent to all 1-bit converters, in the way they do their decode. Scarlet Book standards call for a 7th order noise-shaper. Executed in workable engineering, that's a low-pass filter in the analog domain. Musical frequencies pass through, what's above the Scarlet Book definition of 50Khz (or where ever a designer sets his low-pass filter) gets blocked.
If DSD DAC designers follow this formula, which is going to be really common owing to the physics of a 1-bit decode, then results from a DSD DAC should be more or less closer to the next one than further. Ie: the filtering is very simple. Compare that to a PCM digital filter in FIR – with lots of ringing, slow or fast roll-off slopes, Nyquist, corner frequency, sharpness of the knee etc etc. – that has a lot of variables, and that's why various DAC IC manufacturers have a sound.
Not because of the silicon, but because of the digital filters written aboard. Oh, they are rarely 1-bit designs, being 2-6 bit delta-sigma (or PWM in the case of ESS) designs. Having many modulators complicates things greatly – and errors during the dynamic element matching process abound in chip IC silicon.
Laying that basis, I'll try as simply as possible to answer
@111MilesToGo's question:
- The audio-bridge and 1-bit DAC are like the heart of N7. The heart beats and pumps blood involuntarily. In N7, these two items in the signal path convert 1-bit audio involuntarily.
- It's in the FPGA logic, the brains of N7, that dictates how a DSD decode could sound since it can be taught new things with new programming, and certainly influence lots of how PCM is set up to be converted to DSD in the audio bridge.
- DSD, owing to very simple filtering before becoming analog, can (will) sound more consistent than PCM can, because PCM's digital filters can get very, very complicated and that's where a lot of the IP in a good PCM DAC lies.
- The digital decode is the start of the analog chain in N7, but there are many things that follow such as the discrete low-pass stage handling the small voltage signal influencing how N7 really sounds.
Below I'll leave you my snippet of why we think N7's
discrete gain stage sounds so special ...
N7's driver section counts on all the benefits that come with a discrete bipolar design – low voltage noise, low open-loop gain and the resultantly lower negative feedback necessary, high slew rates and unity gain stability. This accurate linearity means that whatever entered N7's DAC as digital, comes out exactly the same in analog. There's a reason N7 portrays elements within the soundstage with such solidity, the aural images it generates so convincingly rendered – its small signal post-conversion is of the highest quality, and inherently stable without going into oscillation caused by too much gain product.
... and I'm very eager to read more of Andy's inside stories about Cayin's development – the next I believe exactly about the discrete small-signal stage – because it's been a fascinating journey with them.