This paper reviews the performance of many Roberts evaporator designs, notes their shortcomings and presents a design that corrects them as a novel SRI design that may well maintain the popularity of the Roberts type. A very large (5300 m2) new vessel was constructed by Ingenio San Antonio (ISA) staff from design drawings of the principal author in the second half of 2000. Its commissioning phase took place in January–March, 2001. Trials carried out by SRI and ISA engineers during a testing period at the beginning of March 2001 are described. Observations of the boiling pattern and heat transfer performance were made. The heat transfer coefficient (HTC) and the ‘heat transfer performance relative to standard Roberts HTC values at the same conditions’ (HTR), results were determined. The tests later in this series were run with higher operating juice levels and did not achieve the high HTR values found in the earlier tests where the juice level was being kept lower. Despite this, the performance is, in general, excellent, with an average HTC of 3014 W/(m2.K) for the whole 12-day operational test period. Initial tests averaged 23% above the typical heat transfer coefficient achieved in Australian evaporator bodies under the same juice outlet brix and temperature conditions, and later operational tests performed slightly better. These results are all the more remarkable because they were achieved at relatively low evaporation loadings and on juice prepared by the sulphitation process. Both of these factors usually retard the heat transfer performance. After observing the behaviour and noting the HTC results, some ideas for future design modifications were formulated, mainly concerned with the reduction in juice volume beneath the calandria, and in isolating the condensate removal line from the juice space. These have since been explored in variations of the basic design.