Great to see your posts here, Mr. Jung.
Based on the measurements Iâ€™ve done using AD844 open loop I can confirm that there is a problem with its output stage since Iâ€™ve found it adding about 0.5% (at 2V RMS and 1kHz) to the overall open loop THD figure which is quite poor for the diamond buffer, even taking into account relatively low bias current. I hope you can, being the insider, shed more light on the reasons behind.
Funnily enough, its subjective performance is still not bad at all and I clearly liked this chip better with than without this embedded buffer at signal path, and the following device Iâ€™ve used had relatively high input impedance (100kOhm). Iâ€™ve also tried external (discrete JFET) buffer against the embedded one and while I do think the better buffers can be done, it is still not the trivial task.
All the above assumes AD844 with no feedback applied. With feedback and when used like an I/V converter AD844 wasnâ€™t that great and was certainly inferior to LM6181 (also CFA). IME OPA627 is soundwise better than both.
I'm not an insider at ADI, really never was, even when I worked there. Barrie Gilbert designed the AD844.
CFAs with husky outputs can do well, the 811 and 815 come to mind.
Yes, you can run the 844 as a DAC terminator, with roughly a 50R Zin. The current IO of the DAC becomes reflected at pin 5, so an R from 5 to GND recreates a voltage. R > 50R makes for gain. You still need a buffer to drive any useful loads, however.