bootstrap "repeated time values within panel" 2 SRI

I'm trying to run 2sri approach(control function) in Stata 15. My first stage is a probit regression and the second regression is Tobit (using xtdpf code from the website)

My first stage code is as:
global Z "a b c d "
probit X IV $Z i.year i.industry, r
predict `Xhat'
matrix `a1'=e(b)
gen Xuhat=X-`Xhat'

My second stage code is :
global Z "a b c d"
xtdpf Y X $Z Xuhat, timedummies

The problem now is that in the probit model, I control for year and industry fixed effect, I cannot control firm fixed effects. In the second stage, I control both year and firm fixed effects. My supervisor said in this way, the estimates are inconsistent. However, my instrumental variable is quite a constant variable (didn't change every year), besides Probit cannot include firm fixed effects. Does anyone know how to solve this problem? please help me, I will appreciate any help!