We present a study of the central engine in the broad-line radio galaxy 3C 109. To investigate the immediate surrounding of this accreting, supermassive black hole, we perform a multi-epoch broad-band spectral analysis of a joint NuSTAR/XMM observation (2017), an archival xmm observation (2005) and the 105-month averaged Swift-BAT data. We are able to clearly separate the spectrum into a primary continuum, neutral and ionized absorption, and a reflection component. The photon index of the primary continuum has changed since 2005 ($Gamma = 1.61 substack{+0.02 -0.01} rightarrow 1.54 pm{0.02}$), while other components remain unchanged, indicative of minimal geometric changes to the central engine. We constrain the high-energy cutoff of 3C 109 (E$_{text{cut}}= 49 substack{+7 -5}$,keV ) for the first time. The reflector is found to be ionized (log $xi$ = $2.3 substack{+0.1 -0.2}$) but no relativistic blurring is required by the data. SED analysis confirms the super-Eddington nature of 3C 109 initially ($lambda_{Edd} >$ 2.09). However, we do not find any evidence for strong reflection (R = $0.18 substack{+0.04 -0.03}$) or a steep power law index, as expected from a super-Eddington source. This puts the existing virial mass estimate of 2 $times 10^{8}$M$_{odot}$ into question. We explore additional ways of estimating the Eddington ratio, some of which we find to be inconsistent with our initial SED estimate. We obtain a new black hole mass estimate of 9.3 $times 10^{8}$M$_{odot}$, which brings all Eddington ratio estimates into agreement and does not require 3C 109 to be super-Eddington.