Overflow in ARMA-GARCH-M

[sungkyu]

Hello. I am using Eviews 6 student version.
I have been trying to test with ARMA-GARCH-M model and when i add S.D. and Log(var) in mean equation, it does work without any problem.
however, the test could not carry on due to the overflow with variance in mean equation. Does anyone knows that the main reason of the overflow? Is this because of such a big value of variance? My sample is not normal distribution.
I can not think about any possible way to sort this out.
Please give me some help i am stuck with overflow.

the data set is between 06/04/2006 and 15/03/2010.