(x^2 + 2)y" + 3xy' - y = 0 is good enough. This involves the Sturm-Liouville theory and the Legrangian to get a closed solution. That gets too advanced for this site. You might try for a Frobenius series solution. This video might help you do that: https://www.youtube.com/watch?v=Gq7MxHjtAdw Edwin