gistfile1.txt

1.9.3p374 :001 > "\x90".force_encoding(Encoding::WINDOWS_1250).valid_encoding?
 => true 
1.9.3p374 :002 > "\x90".force_encoding(Encoding::WINDOWS_1250).encode("UTF-8")
Encoding::UndefinedConversionError: "\x90" to UTF-8 in conversion from Windows-1250 to UTF-8
  from (irb):2:in `encode'
	from (irb):2
	from /Users/charleshornberger/.rvm/rubies/ruby-1.9.3-p374/bin/irb:16:in `<main>'

Supposedly you can pass options to encode for what to do in the case of bytes that don't translate:

x = "\x90"
x.force_encoding(Encoding::WINDOWS_1250)
x.valid_encoding?
x.encode("UTF-8", :invalid => :replace)

Wait, I figured it out. You need to pass the :undef option:

x = "\x90"
x.force_encoding(Encoding::WINDOWS_1250)
x.valid_encoding?
puts x.encode("UTF-8", :invalid => :replace, :undef => :replace)

You get a replacement character, but you can define the replacement character: