What would your expected output be if you were to change the category filter?
You're blending the data by month so it does a group by logic on both datasets to then display the data in the visualization.
What is the problem you are trying to solve though ?
What fields are eliminated ?
Blending is not the answer to everything, so you have to know what you're trying to achieve and then which fields in the data do you need.
The primary source has order information. I would want that order information to stay the same regardless of any filter on the second data source. If I put in "amount" in the second data source and select none on the filter, I would expect it to either show 0 or just eliminate the "amount" row but not the orders information from the primary source.
Blending is basically a left outer join. By choosing a value in Amount you are only choosing the records that have joined properly. You would have to choose NULL in Amount as well in order to get the records that don't join properly.