Portable spectrometers designed for users with little technical training must be more robust than spectrometers designed for professionals. Common measurement errors (and in fact any imaginable non-random error) must be accounted for, trapped, and the user gently directed to manipulate the system to yield a valid, useful measurement. Such measurements are likely cost-sensitive, as consumers typically want infinite performance at zero cost. These constraints require either simplicity, automation, or software-driven operation using inexpensive components whose limitations are compensated algorithmically. We describe progress towards an instrument of the latter description, providing absorption, diffuse reflectance, and luminescence measurements to be carried out in a hand-held grating spectrometer. The gratings are plastic films, stacked and mutually rotated to generate hundreds of orders, which in turn are detected by a consumergrade camera chosen to mimic those in typical low-end smartphones. Low camera dynamic range is compensated by observing orders with a wide range of throughputs, such that overall dynamic range in a single exposure may be 12 bits from an 8 bit image. This requires significant effort for wavelength calibration and inter-order intensity normalization. The paper discusses progress in automatic calibration algorithms, software modularization, and method development for quantifying nitrate and phosphate in agricultural runoff. While the overall approach has been established for some years, significant details requiring careful raytracing, numerical analysis, algorithmic modularization and error handling, materials choices, and mechanical engineering for manufacturability are of current importance.