... the return airflow enough to cause the fire to go out after five minutes? The thermopile checks as good. When I open the glass front then the fire stays on for the long haul, otherwise it slowly weakens then dies. recently had a new roof and vents installed and a new cap on the direct vent top as well. The top is not blocked but I thought a new shape could create an air current interference with hot air coming out the middle that could block or impede the return air path. Anyone ever hear of this?