Positron emission tomography (PET) is the imaging modality most extensively tested for treatment monitoring in particle therapy. Optimal use of PET in proton therapy requires in situ acquisition of the relatively strong ¹⁵O signal due to its relatively short half-life (∼2 min) and high oxygen content in biological tissues, enabling shorter scans that are less sensitive to biological washout. This paper presents the first performance tests of a scaled-down in situ time-of-flight (TOF) PET system based on digital photon counters (DPCs) coupled to Cerium-doped Lutetium Yttrium Silicate (LYSO:Ce) crystals, providing quantitative results representative of a dual-head tomograph that complies with spatial constraints typically encountered in clinical practice (2 × 50°, of 360°, transaxial angular acceptance). The proton-induced activity inside polymethylmethacrylate (PMMA) and polyethylene (PE) phantoms was acquired within beam pauses (in-beam) and immediately after irradiation by an actively-delivered synchrotron pencil-beam, with clinically relevant 125.67 MeV/u, 4.6 × 10⁸ protons s^-1, and 10¹⁰ total protons. 3D activity maps reconstructed with and without TOF information are compared to FLUKA simulations, demonstrating the benefit of TOF-PET to reduce limited-angle artefacts using a 382 ps full width at half maximum coincidence resolving time. The time-dependent contributions from different radionuclides to the total count-rate are investigated. We furthermore study the impact of the acquisition time window on the laterally integrated activity depth-profiles, with emphasis on 2 min acquisitions starting at different time points. The results depend on phantom composition and reflect the differences in relative contributions from the radionuclides originating from carbon and oxygen. We observe very good agreement between the shapes of the simulated and measured activity depth-profiles for post-beam protocols. However, our results also suggest that available experimental cross sections underestimate the production of ¹⁰C for in-beam acquisitions, which in PE results in an overestimation of the predicted activity range by 1.4 mm. The uncertainty in the activity range measured in PMMA using the DPC-based TOF-PET prototype setup equals 0.2 mm-0.3 mm.