I'll fix this as it is my code.
Note that this curve has j=0 and the cases of j=0, 1728 were not
implemented with any efficiency (or, it seems correctness), but that I
am half-way through doing that.
In the meantime I'll try to put in a quick patch to correct what's
wrong here. First step,
I have posted a patch for this on trac #2849. The bug would strike
for any curve with j=0 (=1728) defined over GF(3^d) for odd d.
Assuming someone reviews this positively it will get into sage-3.0.
It is also likely that by then there will be much better support for
the cases j=0 and j=1728
