This paper shows the details of an implementation of variable radix floating-point complex division based on previous implementations of the algorithm. This implementation takes advantage of the easier prescaling offered by low-radix division and recodes it as necessary for higher radix iterations throughout the design. This, along with proper use of redundant digit sets, allows us to significantly altar performance characteristics relative to exclusively high-radix division implementations. Comparisons to existing architectures are shown, as well as common implementation optimizations for future iterations. Results are given in cmos32soi 32nm MTCMOS technology using ARMbased standard-cells and commercial EDA toolsets.