Monolithic perovskite-silicon tandem solar cells are promising next-generation photovoltaic devices, yet they pose unique characterization challenges that go far beyond established silicon and thin film photovoltaic metrology, mainly through the combination of i) tandem and ii) perovskite-specific absorber and contact aspects. Two external contacts provide access only to the sum of subcell voltages and the minimum of subcell currents, while the current-limiting subcell cannot be identified directly. Mobile ions cause metastability that makes device parameters depend on preconditioning, scan speed, and spectrum, and poor and strongly operation-dependent charge carrier conductivities can introduce non-ohmic transport losses. Critically, the perovskite subcell can operate as a depleted PIN-type device - lacking a quasi-neutral bulk region and exhibiting high-injection conditions in the contact regions - thus violating fundamental boundary conditions of the classical diode model, rendering standard equivalent circuit-based loss analyses inapplicable. This contribution connects these physical effects to analysis methods and shows how they can be transferred to industrially viable measurement procedures.

We address the fill factor deficit that currently separates practical tandem cells from their theoretical limit. Drift-diffusion simulations show that charge carrier resistivity in the perovskite bulk couples to recombination in a voltage- and illumination-dependent way. We call this modulated transport loss. It appears as a photoshunt in illuminated J-V curves that is absent in the dark, a quasi-Fermi level splitting that already exceeds the terminal voltage near the maximum power point, and a voltage-dependent apparent series resistance. The two-diode model cannot reproduce these features. We therefore introduce the Modulated Resistance Model (mod-R), which replaces the empirical second-diode ideality with a physically motivated dark-conductivity parameter. The mod-R agrees well with drift-diffusion results across wide variations of cell parameters.

For subcell-resolved analyses, we use Suns-V_oc and Suns-PLI to extract pseudo and implied open-circuit voltages and fill factors of each subcell separately, which allows quantitative separation of recombination, transport, and selectivity losses. Photoluminescence imaging yields spatially resolved maps of implied voltage and series resistance. As a contactless method, it enables subcell-selective shunt detection by exploiting the voltage coupling between subcells. The method is validated by device simulations and extended to triple-junction cells.

Finally, we demonstrate industrially compatible implementations. LED-based EQE extracts spectral response within seconds using multi-channel simulators. For J-V measurements, open-circuit preconditioning above two seconds stabilizes anion distributions and minimizes hysteresis artifacts. Most relevant for production, we present a millisecond subcell current measurement that uses the silicon bottom cell capacitance as a charge reservoir. An ultrafast reverse sweep after short preconditioning traverses both subcell-limited regimes, yielding individual short-circuit currents from a single measurement. Results on encapsulated industrial tandem cells match spectral metric analyses and show high reproducibility without cell degradation.