Strange QRegExp to Identify non-ASCII Characters

Question

In a legacy codebase, developed with Qt4.8, C++, for an ASCII-based application, I saw the following lines

    const QRegExp lBadCharsRegExp("[^\\x{001F}-\\x{007E}]");

    // if we find a match, our string contains some bad chars
    bool lMatches = xStr.contains(lBadCharsRegExp);

The character class in the QRegExp is weird. It looks like the author was aiming for a range of hex characters, but the curly braces are unneeded. Moreover, if I pass in a string containing every printable character, it rejects just '~'. If I instead do

const QRegExp lBadCharsRegExp("[^\\x001F-\\x007E]");

the '~' is also accepted. Can someone explain

• the odd syntax in the first QRegExp
• why it excludes just '~'