You need to do it in 2 pipes. In the first one, use an user input module (location input would be a good choice), link it to a fetch feed module, then a sort, then a truncate after 2, and pipe output.
In your second pipe, feed your list of sources to your 1st pipe inside a loop module (to do that, I suggest typing your sources in text input module as a pipe-separated list, eg source1|source2|source3, and connecting it to the string tokenizer module with the pipe (|) as delimiter. This last module would then be connected to the loop module).
In the sub-module, I fetch one feed, and truncate it after n elements (in case, n=2). You probably won't need to use it directly but clone it to your 'pipe directory' in case I delete it.
In the main pipe, I take a list of your feed, break it down to the individual URL, and run the sub-module on each one of them, thus getting only 2 elements out of each feed. then, usual sorting by date and optional truncate if there is too much feeds. The trouble with that is the obscure input, so I set up a 'remnant' method using pastebin, as you can see in the second version of the main pipe.
If you have question about those pipes, feel free to ask!