Least Squares Optimization

Question

If $$\mathbf{\begin{bmatrix} y_1 \\ y_2 \\ \vdots \\ y_N \end{bmatrix}} = \begin{bmatrix} b_{11} & b_{12} & b_{13} \\ b_{21} & b_{22} & b_{23} \\ \vdots & \vdots & \vdots \\ b_{N1} & b_{N2} & b_{N3} \end{bmatrix} \mathbf{\begin{bmatrix} K_1 \\ K_2 \\ K_3 \end{bmatrix}}+\mathbf{\begin{bmatrix} e_1 \\ e_2 \\ \vdots \\ e_N \end{bmatrix}}\tag{1}$$ then K can be found as \begin{equation} K = (B'B)^{-1}B'Y \tag{2}\end{equation} however B is unknown and is a function of $ e_{n}(t-t_{d}) $ The only known parameter that I have is $e_{n}(t)$. It is mentioned in the article that I am reading that $t_{d}$ can be found by letting $S=e'e$ and then partial differentiating S as $$\frac{\partial S}{\partial t_d}=0$$ However, this doesn't make sense to me. Because from my reading, that is how the normal equation (2) is derived. I would be grateful if someone can guide me in finding $t_d$ and I apologize if the question seems too noobish as I do not have a strong mathematical background. I have been reading on linear equations for a while but I haven't yet encountered a similar question. Thank you