I don't think they do. According to https://www.akm.com/global/en/products/audio/audio-dac/ak4191eq/ the AK4191 has a THD+N of -153 dB and a noise floor of -156 dB. There is no way such low numbers could refer to analogue imperfections. It must be the digital noise floor you get from rounding to seven bits and from finite word lengths in general.
Ooooooh ok so it has nothing to do with the analog domain then? Or is it that that noise floor doesn't make it into the analog domain?
If that's the case yeah it may be easier just to go with the sabre dacs...
If that's the case yeah it may be easier just to go with the sabre dacs...
The way I interpret it, it is the purely digital noise of the AK4191. Connect a DAC to the AK4191, and the DAC will convert it to analogue and add its own noise, which in all practical cases will be much larger.
In a similar vein, has anyone got a recommendation for a 32 bit I2S ADC board? I'd been wondering about 32 bit in and out and multiple I2S DACs, with a minidsp USB adapter as traffic cop.
Try to explore another galaxy 🙂
For this time, 2023 AD, there in no one Audio ADC or DAC on this planet, with the real 32-bit. (even 24 🙂)
BTW, Sharc and sigmastudio looks and works great.
Until you see the licensing fees or otherwise the enormous pile of effort.
I would recommend the ADAU145x series instead. (or the slightly newer series).
Until you see the licensing fees or otherwise the enormous pile of effort.
I would recommend the ADAU145x series instead. (or the slightly newer series).
Well I retract my statement. I was thinking of the LTC-2500, but its ENOB is only 22
Since the beginning of this post, I have VERY much become aware of this lol. It's insane how tightly they keep the dev tools under lock and key. They DO however now make a project board with a ADSP-SC589.
Another thing I have explored since the beginning of this post. I have been eyeing up the ADAU1467. I also have decided to scale down to just 3 and 3 to save money and space. I also decided to go with ESS over AKM because now AKM acts like how ESS used to be and not publish their full spec sheets for their newer chips.
If you relax the bandwidth requirement of audio I think it might be possible. Could be some seismic applications (which I recall may only have a ~1kHz sample rate) get there.