Unicode Broke

Hello, my server is red hat linux 9 postgresql 7.3.2,  my client is
windows xp professional odbc postgresql+ unicode beta (postgresql
unicode beta some times)
my database has unicode encoding.  When I try to read a recordset of
more than 16KB I get strange characters and I loose some of the
beginning of the recordset too but if the recordset is less than  16KB
it works perfectly.