Hello! I am looking for ideas on how to improve SFDR of a transmit DAC beyond what a single chip can achieve. One idea that came up is to parallel two or more DACs. Is that feasible? Are there any application notes on this topic?
It is a well known technique to add the outputs of amplifiers in quadrature to improve the linearity of the amplifiers. With DACs this technique doesn't work as well because the distortion is governed by the transfer function of the DAC, not a square law like an amplifier. Square law devices (like amplifiers) are very predictable and adding them in quadrature does improve distortion.
Adding (or combining) the outputs of two DACs in quadrature will likely improve SFDR, but it won't be as predictable as an amplifier.
For a DAC adding a dither tone to the digital code and removing it later will spread the signal around the transfer function which will reduce SDFR at modest power levels.
Another technique would be to use predistortion. You would need to sample the signal coming out of the DAC and predistort the code (or inject another signal with another DAC) to cancel the distortion in real time as it is coming out of the DAC.
Retrieving data ...