In the course of consulting on a channel requiring a low noise non-inverting gain - there was also a need to inject a DC offseting level through the inverting Rg path - this needed to be broadband low noise where the ADA4896 seems ideal for this gain of 1 buffer. For non-inverting signal path flatness,the closed loop output impedance of the ADA4896 would be critical.
The tool I am using has a library model for the ADA4896 ver. 2.1. However, when I look into the net list (very well organized and easy to read) I see only a fixed 1ohm open loop output resistor
That is extremely unlikely having seen (and modeled) quite a few higher speed RR output stages (see figure 6 in that attached recent article).
Questions? Is the online PSpice model on the ADI web site updated to a more accurate open loop output impedance? Lacking that, can I least get a more realistic number than this 1ohm place holder. 1ohm is only possible if the output stage is itself closed loop but normally then you would see its own loop gain causing an inductive open loop output impedance (I have modelled plenty of those as well).
So I switched over to the ADA4899 model - it also is a fixed 30ohm resistive output open loop model - looking inside, it is more a transistor level model were it appears those transistor models are very simplified with no AC terms - so if this non RRO device is typical class AB output buffer, that 30ohms is two 1/gm at 60ohms each for the up down output stage npn/pnp. So apparently a 0.43mA standing current in these emmitter followers - seems kind of low. It does run better than the ADA4896, so I will switch to this - don't need the better ADA4896 model and the approach I am now taking does not need the more accurate open loop impedance model - I would also note the simulated noise seems low in this model vs PDS. Can always fix that with a small series R into the V+ input.
Thanks for pointing it out. We will try to look at it in the future.