That is correct. For consistent behavior you need to use unicode characters instead of HTML entity characters.
What you could do is set up a translation table (Or, even better, a dictionary) between HTML entities and Unicode characters. You can use the table at http://www.alanwood.net/demos/ansi.html as a starting point.
if you want to show html entities, use <mx:text HTMLText='&xxxx'/>
But how is it possible that some ascii code are working and others are not?
You must use Unicode values, not Windows extended ASCII values.
The Unicode value of a copyright symbol is hex 00A9 or decimal 169. This happens to be the same as where it is in a Windows' extended ASCII character set, but this is irrelevant. The Unicode value of a copyright symbol is hex 2122 or decimal 8482. In Windows' extended ASCII it has a different value, hex 99 or decimal 153, but again this is irrelevant.
So the reason you think it "works" for the copyright and not for the trademark is that the two character sets agree on the value of the copyright but disagree on the value of the trademark.
Moral: Find a good Unicode table.
Adobe Flex SDK Team
Actually, let me revise the moral: Store Unicode in your database. Or build a mapping table of Windows extended ASCII values to Unicode values.
Thanks Gordon....it's clear now.