This work presents DeBaTE-FI as a scalable tool to aid in assessing the susceptibility of embedded IoT systems to soft errors. DeBaTE-FI includes a method that allows precise targeting of the injection points, equivalent to common simulation methods, while providing significant improvement in the accuracy of the results, particularly concerning the propagation of errors in the control flow. A comparison against an instruction-based fault injection method is made, showing a doubling in the estimated rate of crashes/timeout outcomes using DeBaTE-FI when compared to an identical simulated campaign. We present improvements in the performance of the platform following the simulation campaigns and evaluate the system's scalability.