Hi Rik,
good to hear that you was able to generate a splitted sitemap for more than 600.000 links. As you may know dealing with so large sitemaps is a great challenge.
About the problem with Kunena topics, at this point i think that you was right when saying that an ASCII character is messing things up.
It could be the case that one or more posts link contain some invalid characters breaking the XML format.
My only idea would be to try to identify those posts so eventually they could be excluded. If you look at the screenshot below you see that data sources can be configured to include/exclude topics or categories by IDs.
John
