Characterising ores in-pit and in real-time presents a significant value opportunity for the mining industry. Hyperspectral imaging systems are proposed as an enabling technology to achieve this real-time characterisation. This paper considers the application of a hyperspectral imaging spectrometer mounted on a shovel. We define the performance requirements for such a spectrometer and associated subsystems during a typical dig-load cycle. The relatively high swing rotation speed of the shovel requires approximately video-rate exposure times to achieve good spatial resolution. We analyse a selection of components against these requirements and find that suitable commercial spectrometers, detectors and optics appear fit-for-purpose. Automated, real- time spectral classification systems are also available, with only relatively modest requirements for processing power. The paper argues that at near peak solar illumination, a system for field deployment at the required sensitivity is viable using only commercial off-the-shelf components.