DAC architecture section of AD568x datasheets describes their internal structure as "string DAC followed by an output amplifier". According to description resistor string is fixed and buffer amplifier gain may be set to 1 or 2. But in this case it is difficult to explain why reference input impedance changes when buffer gain is changing.
Probably in real internal structure is string of 2^N resistor with value R and sequentially connected resistor of value 2^N*R at upper tap of string. The last one may by bypassed by gain selecting switch. And buffer has fixed gain of 2. Such internal structure have competitors DAC, for example DAC8562/3.
If so it is clear why reference input impedance is inversely proportional to gain selected.
Unfortunately AD568x datasheets reference input impedances do not match reference currents in both documents (AD5689 rev. A, AD5686 rev.B) so it is not clear what current level and impedances have AD568x devices in real.
So question is does AD568x DAC really consist of resistor string with programmably bypassed upper resistor followed by fixed gain amplifier and datasheet description is incorrect? Or is description correct then how reference impedance change may be explained?