How to properly build a 24-bit R-2R DAC?

Question

Electronics

I know that it is realistic to build a DAC on a 16-bit R-2R matrix for a power supply up to 0-30V 0-5A, two such DACs for such ratings is quite a restful task. It turns out approximately steps 0.00045776V (at the required 0.01V) and 0.00015259A (at the required 0.01A). I came across schemes on the Internet 'with the successful implementation of such a task.
I would like to clarify what difficulties await me when building a 24-bit R-2R matrix?
I saw poats where they build audio DACs with a similar resolution, it was noticed that there is a small inductance at the output of the 74HC595, what is it for and how can linearity be improved and what errors in the construction should be avoided? Thanks in advance for specific advice.
I will build from 2kOhm smd 0603 1% of one batch. What will provide the linearity of this DAC.

Answer 1

So a 1% error in the most significant digit immediately overlaps 16 least significant digits, doesn't it?

Answer 2

24-bit R-2R DAC

Do you know that every 10 bits is 1000 times, and thus, a DAC with a bit depth of 20 or more at the output will already operate with microvolts? In order to really use its low-order digits, you will have to enclose it in a Faraday room, and power it not from a noisy-ringing 220 volt network, but from a chemical source. And this is first.
And secondly, the design. DACs with a high bit depth, in principle, cannot be built on discrete elements, due to their spread (unless you fool around with their selection to an accuracy of 0.001%). They are made only as a monolithic device on a single chip, and this is already an integral finished device, the correct design of which you do not need to worry about, its developers have already taken care of it - you read the datasheet and do as it is written there.
So forget about 24 and immediately take what you can - say, 12. But even then, take resistors an order of magnitude more accurate.