We present high-speed, three-colour photometry of the eclipsing cataclysmic variables CTCV J1300-3052, CTCV J2354-4700 and SDSS J115207.00+404947.8. These systems have orbital periods of 128.07, 94.39 and 97.52 minutes respectively, placing all three systems below the observed “period gap” for cataclysmic variables. For each system we determine the system parameters by fitting a parameterised model to the observed eclipse light curve by χ2 minimisation. We also present an updated analysis of all other eclipsing systems previously analysed by our group. The updated analysis utilises Markov Chain Monte Carlo techniques which enable us to arrive confidently at the best fits for each system with more robust determinations of our errors. A new bright spot model is also adopted, that allows better modelling of bright-spot dominated systems. In addition, we correct a bug in the old code which resulted in the white dwarf radius being underestimated, and consequently both the white dwarf and donor mass being overestimated. New donor masses are generally between 1 and 2σ of those originally published, with the exception of SDSS 1502 (−2.9σ, Mr = −0.012M⊙) and DV UMa (+6.1σ, Mr = +0.039M⊙). We note that the donor mass of SDSS 1501 has been revised upwards by 0.024M⊙ (+1.9σ). This system was previously identified as having evolved passed the minimum orbital period for cataclysmic variables, but the new mass determination suggests otherwise. Our new analysis confirms that SDSS 1035 and SDSS 1433 have evolved past the period minimum for cataclysmic variables, corroborating our earlier studies. We find that the radii of donor stars are oversized when compared to theoretical models, by approximately 10 percent. We show that this can be explained by invoking either enhanced angular momentum loss, or by taking into account the effects of star spots. We are unable to favour one cause over the other, as we lack enough precise mass determinations for systems with orbital periods between 100 and 130 minutes, where evolutionary tracks begin to diverge significantly. We also find a strong tendency towards high white dwarf masses within our sample, and no evidence for any He-core white dwarfs. The dominance of high mass white dwarfs implies that erosion of the white dwarf during the nova outburst must be negligible, or that not all of the mass accreted is ejected during nova cycles, resulting in the white dwarf growing in mass.