In rawSolrRequest I head to put:

$data=$this->postQuery( $request, $paramsAsString, 'application/x-www-form-urlencoded;charset=utf-8' );

It solved problems with searching word with "2 or more bytes utf-8" characters. These characters where well indexed (over text/xml content post queries), but when searching solr would not get good character, so there were no results.
Same query manually executed over get was ok.

Enhancement would be to have some ini isettgin regarding this...

Steps to reproduce

Index some word with ẫ char. Then try to search it.