I'm trying to understand how DAC chips work

Hello .
Please forgive my ignorance, but I would like to understand how a multibit DAC behaves when it receives information from a delta-sigma DAC. This is the case when using a converter (with DAC delta-sigma inside) from usb to s-pdif..
Is it possible for a multibit chip dac to behave like a transparent system, i.e. to bring a minor change in the sound if there is another delta-sigma type chip in front of it? And here I mean the conversion, for example, from s-pdif to usb followed by the multibil dac. That is, if the delta-sigma DAC sequentially transmits bits of information, maybe that the multibit DAC will take information fragments and it will pass them on "slices" of delta-sigma type information. That is (just as an example) if delta sigma sends 4 bits of information to the multibit dac, the latter will practically try to process all those 4 bits of information and not maybe 16 (depends on chip) as much as it can, because that's what was transmitted to it.

I'm waiting for you to tell me where I'm wrong.
 
Input to a DAC is digital, output is analog, you don't chain DACs in line like this. Perhaps you are confusing delta-sigma modulators with DACs? A modulator is often the first stage of an ADC and the last stage of a DAC. The rest of the chip is rate-conversion, clocking, precision reference, digital filtering.

So I think you are on the right lines, but its only the delta-sigma part that actually converts between analog and digital, the rest is all DSP stuff, especially rate-conversion (decimation and up-sampling (aka interpolation) for ADCs and DACs respectively). The key enabling technology is noise-shaping which pushes nearly all the quantization noise well above the Nyquist limit, and is probably the closest thing to magic in DSP! You need some understanding of feedback and control theory as well as the more pedestrian aspects of DSP to understand how it enables upto 24 bit resolution.

Most sigma-delta (people say it in either order) modulators are proprietry technology so datasheet typically only hint at the internals of these chips, saying things like "fifth order sigma-delta modulator" - higher orders do more noise-shaping, but are harder to tame.
 
So I think you are on the right lines, but its only the delta-sigma part that actually converts between analog and digital, the rest is all DSP stuff, especially rate-conversion (decimation and up-sampling (aka interpolation) for ADCs and DACs respectively). The key enabling technology is noise-shaping which pushes nearly all the quantization noise well above the Nyquist limit, and is probably the closest thing to magic in DSP! You need some understanding of feedback and control theory as well as the more pedestrian aspects of DSP to understand how it enables upto 24 bit resolution.
You can decimate and interpolate and by extension rate convert without noise shaping.
 
...would like to understand how a multibit DAC behaves when it receives information from a delta-sigma DAC.
As others have been trying to explain there are delta-sigma modulators, Dynamic element Matching output resistor arrays (or if not resistors then maybe current sources, or maybe scaled transistors), all of which are often included in one DAC chip. Maybe something like this (where 'Hyperstream' is a modulator):
1675563540382.png

Regarding SPDIF in particular, it is not really what I would consider to be an ideal digital interface. The above diagram is for an ES9038Q2M DAC chip which includes a SPDIF decoder and an ASRC to attenuate incoming timing jitter from the SPDIF signal. However, this particular DAC chip can be configured to use asynchronous USB, which then can be fully synchronous from the perspective of the DAC itself. In that case the data can remain in bit-perfect form, and without the need for any special ASRC or FIFO buffering (one or the other of which is often used for SPDIF in order to help maintain good sound quality).
 
Last edited: