I have a circuit composed of subcircuit A which creates a low-pass filter of an input voltage (It has an AD8221) and subcircuit B which takes the output of A and uses it to drive a feedback loop to ultimately drive a mosfet. Subcircuit B converges fine in LTSPICE if I give an input voltage. Subcircuit A converges fine if I put the output to a 1K ohm. But if I put subcircuit A and subcircuit B into the same LTSPICE Model, then subcircuit A fails with complaint about a singular matrixat a node inside the AD8221. Here is the subcircuit A, which converges just fine.
Just placing a different circuit next to this one, causes the LTSPICE calculation to fail on U4. No electrical connections between the circuits. Completely bizarre. I could understand if, say, the output of U4 was driving an input to subcircuit B. But it fails all by itself. My thoughts are that somehow subcircuit B is encouraging a certain time step size and that time step size is breaking the calculation for subcircuit A. Or maybe there is a dumb bug in the above circuit??
[I downloaded the AD8221 model as it was not included in LTSPICE by default].