Hi Dara,
Mistakenly I replied the email below just to you, as opposed to the list. So I'm forwarding it to the list as it contains the link to the new location of the design_ekaterina.ods.
Since then, you sent me two other messages off-list. I'll reply them here:
1) About the rationale for the -1 in the subject-specific EVs in design2: the are to absorb the intra-subject variability, leaving only the between-subject variance, which is necessary for a between-subjects comparison in there are repeated measurements.
2) About the message in randomise: "Warning: tfce has detected a large number of integral steps. This operation may require a great deal of time to complete."
This means something is not right in your design (it looks you've used subtractions, but that isn't what is prescribed in design_ekaterina.ods). Also, there are only 256 shuffles possible, which suggests just 16 subjects and a 1-sample t-test, but this doesn't match with your description of 34 subjects. So, something isn't right there... no need to allocate more cluster resources, just check the design.
All the best,
Anderson