Problems that require solving the same equation for multiple Right Hand Sides (RHSs) are ubiquitous in physics and chemistry. They typically result in large scale matrix equations, which can be either solved once for each RHS or all-at-once. Block Krylov methods where especially designed for such problems. By simultaneously solving the problem for all RHSs block Krylov methods can achieve faster convergence, making them an attractive approach for solving matrix equations. However, they are even more sensitive to round-off errors than their classical counterparts. While multiple regularization techniques have been proposed to improve the convergence of block Krylov methods, their efficacy remains case-dependent. Previous work has shown that in some synthetic cases, increasing the working precision improves the convergence rate of classical Krylov methods. In the current work, we evaluate the impact of extending the working precision on block-Krylov methods. We first compare different implementations of the BLock-BiConjugate Gradient (BL-BiCG) method with various working precisions on various problems from the SuiteSparse matrix collection. We then study the impact of working precision on a synthetic problem relevant to medical imaging. We show that increasing the working precision enables the convergence of the BL-BiCG in all of our considered cases. Moreover, we show that in extended precision, regularization techniques only provides a marginal improvement. Finally, we analyze, on a case study relevant to medical imaging, the convergence of the BL-BiCG for several RHSs and several precision.