Impulse Response Function Fluctuation

Have you seen this type of Impulse Response Function result?
Plz. have look an attached pdf file.
The several IRF results are something wrong, they are too fluctuated like the file.
I checked all shock variable, and AR(1) parameters.
Those anyone have any Idea?

Many Thanks.
twosec_o_IRF_eta_eps_r2.pdf (3.57 KB)

I have seen something like that before. It occured when the timing of a variable was wrong. In my case, I erroneously set a non-predetermined variable as predetermined. This resulted in fluctuating impulse responses (apparently a complex eigenvalue that should be real).