I recently saw a single supply opamp circuit that had what looked like an interesting virtual ground topology.
The incoming signal needed to be differential for the rest of the circuit, so there's an opamp phase splitter arranged with a voltage divider on one of the noninvering inputs such that the outputs of the phase splitter center at half the supply at quiescense. Equal value resistors are then taken from each op amp output of the phase splitter and the non inverting input of another opamp configured as a unity gain buffer is connected there. That amp's output is then used as the virtual ground for the system.
I guess this arrangement saves using an electrolytic bypass cap across the virtual ground voltage divider at the cost of an opamp section. I haven't simulated this yet as I'm mobile right now; I wonder if there could be issues with this arrangement, like startup problems? Thanks.