Thank you for the informative response.
I think I will try with increase the number of tuning/sampling and more informative prior. But I cannot pick lower target accept. It gives me some divergences. Also, number of observations are really low right now. This might be the main reason model fails in certain occasions (non-informative priors, lower target accept, autocorrelation etc.).