[Yanec]

The resistor matching level is not particularly extraordinary. What makes it all work is our linearity compensation technology. To grossly oversimplify, it takes the input values of the digital data assuming an evenly spaced grid of 2^bd values (bd=bit depth) and maps it correctly to the actual values of a non-evenly spaced grid produced by an imperfect R2R network, ensuring output linearity whatever resistors we use (within reason). This was sort of glossed over, briefly mentioned in the ad copies, but should really be a big selling point...

Thanks! You have a very good point there.

Both Amir from ASR and Paul from PS Audio have told me at certain times that r2r would be their preferred dac if the need for high resistor tolerance can be addressed properly. Paul even made a video on this several years ago.

It seems you have discovered one way to address that, you should definitely use it more as marketing.

Is the compensation two part: hardware plus algorithm(software)? Is the algorithm part resampling or altering the signal?

 

[Joe Bloggs]

Thanks! You have a very good point there.

Both Amir from ASR and Paul from PS Audio have told me at certain times that r2r would be their preferred dac if the need for high resistor tolerance can be addressed properly. Paul even made a video on this several years ago.

It seems you have discovered one way to address that, you should definitely use it more as marketing.

Is the compensation two part: hardware plus algorithm(software)? Is the algorithm part resampling or altering the signal?

The hardware part would be to make as good of an R2R DAC as possible within the budget to start with, the software part would be improving it via software as described above. Sorry I cannot go into further detail about it (the original post actually contains most of the details if you read into it), but rest assured it is not resampling in terms of adding samples in time, for the NOS mode...

 

[Yanec]

The hardware part would be to make as good of an R2R DAC as possible within the budget to start with, the software part would be improving it via software as described above. Sorry I cannot go into further detail about it (the original post actually contains most of the details if you read into it), but rest assured it is not resampling in terms of adding samples in time, for the NOS mode...

Yep, I understand. No proprietary details needed. You answered the only concern - no resampling, so it indeed is pure r2r and real NOS. That's what I was hoping for.