I'm working on a model featuring myelinated axons stimulated via the extracellular mechanism.
So far I've implemented myelinated sections and unmyelinated nodes using two distinct arrays of sections with fixed nseg and length, and then used sectionref lists to assemble them into the axon with alternating nodes and internodes.
The internodes have no channel mechanics but have the extracellular mechanism, decreased membrane capacitance and increased membrane resistance.
Both types of sections have different nseg and lengths. I've noticed that especially when modeling smaller diameter axons (2-5 microns), the membrane potential has a saw-wave aspect where the extracellular field is strongest. This is related to nseg as changing it changes the aspect of the membrane potential, but attempting to increase nseg actually makes things worse with greater distortions, counterintuitively.
Right now, for a 250 micron internode I have 13 segments (and for the 3-micron nodes I have 1 segment). Increasing nseg to 27 and 3 respectively makes the distortion worse at the internodes.
To give an example, here are two screenshots of the problem:
The first one is with low nseg (5) per internode on a 5-micron fiber being stimulated by a sine extracellular field. The latter is the same but with higher nseg (more than 20 if I remember correctly)
It seems the low nseg case is the most logical: nodes of ranvier drive down the rest of the axon's membrane potential (after an initial rise due to the extracellular field) and there are smooth transitions between nodes. In the latter case the membrane potential seems to follow a saw-wave. What is going wrong?