I have now fixed the error in the example you quoted (the "." was missing from the .tran statement because earlier Ngspice versions of EasyEDA didn't need it).

Your project is private so I can't see it to help diagnose the model issue. Can you share it?

Sorry but this has turned into a comedy of errors. I have looked at your spice symbol. There is a problem with it because if you add your own model then you must paste the subckt into the schematic sheet and not into the Spice Symbol as you have done. I have created a new AD8422 spice symbol:  Please remove your own version of the AD8422 spice symbol. However, you do not need to add a model anyway because there is a native AD8422 model already in the LTspice ADI1.lib library. However, there is something wrong with the way EasyEDA is accessing the AD8422 model in the ADI1.lib library because it is getting hung up on a subckt for an ADA4700. * I think this may be related to the ADA4700 actually being named ADA4700-1 and I think EasyEDA is confused by the "-" (hyphen) instead of a "_" (underscore) character. Even if I copy the AD8422 model paste it into the schematicand change the subckt name to AD8422x and the name in the spice symbol to AD8422x, EasyEDA still goes looking in the ADI1.lib despite netlisting the local version and then hangs of the ADA4700. However, assuming that the library calling issue can be fixed, even when the AD8422 model is run natively in a local installation of LTspiceXVII, it does not run reliably. It works for some reference input voltages but not others (try the AD8422 test jig in a local installation of LTspiceXVII!). So, it requires that AD fix their AD8422 model and EasyEDA fixes the library calling process. :( * I have raised a Bug Report for EasyEDA and will be posting an LTspiceXVII Bug Report with AD.

Thanks for the reply and for your help!  That's what I was afraid of, and I was wondering about the model so I also tried it in LTSpiceXVII, and while it runs fine, the output is not correct at all.  Unfortunately, this is an issue with some other parts as well, because I attempted a couple other instrumentation amplifiers and differential amplifiers with the same results.  Shoot!

Thanks for the reply and for your help!  That's what I was afraid of, and I was wondering about the model so I also tried it in LTSpiceXVII, and while it runs fine, the output is not correct at all.  Unfortunately, this is an issue with some other parts as well, because I attempted a couple other instrumentation amplifiers and differential amplifiers with the same results.  Shoot!

Thanks for the reply and for your help!  That's what I was afraid of, and I was wondering about the model so I also tried it in LTSpiceXVII, and while it runs fine, the output is not correct at all.  Unfortunately, this is an issue with some other parts as well, because I attempted a couple other instrumentation amplifiers and differential amplifiers with the same results.  Shoot!

Ahhh!  I was getting an error, so I tried several times to add comment...  so there they all are!  Sorry about that.

@lemonoje, I suspect this is a problem with the AD models rather than with LTspiceXVII itself. I have had problems with AD models many times before. They do not seem to be as well written and tested as the (pre-AD takeover) Linear Tech models. How accurate do you need the models to be? Would a simple behavioural model suffice?

@andyfierman Well the behavioral model might suffice, I need to build an analog receiver chain with several ICs, op-amp circuits (primarily using LT1057), anti-alias filters, and ADCs.  Since the signal will be sometimes differential, sometimes single ended, have a wide frequency range, and is modulated - I need a pretty accurate model that shows any clipping, impedance mismatches, phase shifts, etc.  I may also need the LTC1563 and LTC1564 anti-alias filters, LTC1606, along with a couple AD parts.  I'm evaluating devices and I wanted to also model it in simulation.  I wanted to use EasyEDA over LTSpice as the overall editor and online capabilities FAR exceed LTS (by an order of magnitude).  LTS can usually get the job done, but it has that circa 1990s feel to it and no collaboration capability at all.