Not familiar with this exact issue, but I do know that when you get into high pore-volume throughput conditions (ie small cells) in a given timestep we often get problems. Fully implicit simulation methods are meant to be unconditionally stable, but in reality significantly exceeding mobile pore volume flux per cell in a given timestep will often cause problems.

I think the problem you have heard of is less related to big cell connecting small cell, and more related to small cell with high flux