How to use a db in UTF-8 encoding under Windows

This is a repost, but I am trying again in the hope that someone knows
the answer.

I have a problem with a 8.2.15 PG database under Linux when I try to
connect from a 8.2.15 Windows client (psql)

As per the Notice for Windows users I set the font of cmd.exe to Lucida
console and then change the codepage to UTF-8:
chcp 65001
Then when I try to make a select or any \d command I get "Not enough
memory"

I tried with setting the codepage to 1251 (that is the native on
Windows)
and I can do all selects, insert - however the ecnoding is wrong and I
cannot actually use the db as it contains non ASCII symbols.

I see that there was a bug for this. Is there a work around?

Kind regards:
al_shopov