Maybe I misinterpret the question, but... Steep low-pass filters overshoot by design when you apply a step as input signal, and it is a linear phenomenon: the overshoot is proportional to the step. You could attempt to correct for it by inverse prefiltering or some other technique, but then you reduce or eliminate the filtering. If you are willing to accept the consequences of that, such as stronger ultrasonic images and/or worse frequency response, you might as well use a smooth rather than a steep filter from the start so you don't have any overshoot to correct.
That is the topology for a pre-distortion DAC
Also I didn't know dacs without some kind of processors existed.
Could you maybe stop misinterpreting every single word of my post? OBVIOUSLY when I say "add error correction code", I meant as in having that code be executed by the processor, NOT putting a sticker on the amplifier with "I want error correction"
And just as a side note, there is almost no functional difference between an SDM DAC and a class D AM modulator.
Also I didn't know dacs without some kind of processors existed.
Could you maybe stop misinterpreting every single word of my post? OBVIOUSLY when I say "add error correction code", I meant as in having that code be executed by the processor, NOT putting a sticker on the amplifier with "I want error correction"
And just as a side note, there is almost no functional difference between an SDM DAC and a class D AM modulator.
I'll go one better than that and cease responding to every single word of your posts.
This has already been done, and they even sell amplifiers using this technology. I’m too lazy to look up the name of the microcircuit; you can’t buy it anyway, and amplifiers that cost $1,000 or more are sold using this technology. Essentially, this approach requires specialized ADCs, and it also imposes very serious restrictions on the topology of the printed circuit board. In short, it is very difficult to squeeze out high characteristics from this topology.
I believe you may be referring to this one : https://www.diyaudio.com/community/threads/zetex-ddfa.117627/
Well after crunching some numbers, it seems that it's WAY easier to just upsample more and more than to add in error correction. Since yeah it would require a complex model of the response as to not defeat the function of the filter. At which point it likely will require more processing power than a higher upsampling frequency will, and will have the risk of errors in the model.
It's some sort of autocalibration loop for digital predistortion for a transmitter. Its main purpose is to correct for non-linearity of the PA; RF PAs are usually designed for high efficiency rather than good linearity. You could make something similar for a class D audio amplifier, but applying it to a DAC reconstruction filter doesn't make much sense to me.
With an attitude like that, you won't get many useful replies.