A lot of (most?) current DS DAC chips are “multibit” usually the upper 5 or 6 bits, with the remaining bits done by modulating the bottom bit.
It’s the reason Schiit uses “true multibit” to differentiate their solution.
This is very interesting. It’s funny, I spoke with a number of manufacturers at the recent CanJam in NY and they told me they’ve actually had to scrap designs (or significantly delay production released) they worked on for a long time because of the release of new DAC chips. I hope they get enough lead time on this stuff to get them into their systems. I think one of the concerns is that regardless of how something sounds or is implemented, we still focus on the DAC chips being used, and will want the latest and greatest.
For me, it is interesting to see as being in the silicon business, that they are decomposing the Pure Digital: Delta-Sigma Modulator and digital signal processing from the mixed-signal: Digital/Analog section of D/A convertor.
We will see this more, and more so we can use an advanced process for pure digital logic, and I/O die back a generation. Here they can use a different foundry or process node for AK4191, and use the current foundry D/A chip. Also, they can just update AK4191 with more features and bigger DSP for more secondary processing independent of D/A section, even do semi-custom solution for bigger customers
@Resolve - did you get the impression that they were scrapping their designs because they needed their new products to take advantage of the new chips (or availiability?), or because of perception of “older vs. new” To consumers and the marketability of products with older chips.
Splitting the modulator from the DAC makes a fair amount of sense. Though it is interesting to note that the raw performance of the 4498 is on the same level as the 4497 (and not on the same level as the 4499), which has an integrated modulator.
The 4497 is a bit fickle to implement, though, particularly with regard to power-supplies. It’s also voltage-output, as opposed to the more common current-output.
So, the ability to have more capable modulators, divorced from the raw DAC output, is definitely interesting architecturally. What we need, now, is a 4499-level part that can take the modulator-output as a direct input, and see where that gets us (as there are definite advantages to having the modulator, up/over-sampler/etc. be separate … see Chord M-Scalers for an example).