Special char problem between SS versions

I am working with sites in the Portuguese language which contains a number of funny chars. This is a source of many problems and the only solution I've come up with which work ok is to uncheck the "Tidy HTML after scraping". This way I get most of the chars extracted the way I want them. I use the char setting from the site I am scraping which is usually ISO-8859-1 in both the scrape session and as a general setting in "settings". That one work better than the UTF-8 surprisingly.

I build my scrapes on my desktop which is running Vista and when ready I export them to the Linux server for prod and this is where the "unsolvable" problem comes in. When I import the scrapes in the Linux-SS all chars in my extractor patterns get messed up. Even worse, after I have fixed them again and export the scrape to the "import folder" in order to be able to run the scrapes from the SS server they get messed up again. This time I'm unable to fix them, hence cannot use the server...

Does any one have a solution to this?