I'm
wildly guessing here, but I'd bet you a beer I'm in the ballpark...
When MC converts a file using an internal encoder, it goes through the normal audio engine to the encoder, because MC knows exactly what formats it can handle and not handle.
When you convert with an external encoder, I bet it works similarly to bitstreaming: if the external encoder can handle the input (or maybe just "if it is already PCM") then it passes it through directly. Since DSD encoded data is packed in a PCM stream, and the encoder can take PCM, it passes it through. So, for this to work, the
external encoder would need to be able to handle DSD input.
When MC actually bitstreams DSD, it has to
look at the data inside the PCM stream and decide on the fly if it is DSD formatted or PCM audio. I'm thinking it doesn't do this detection here, so it shunts the PCM through (as it would with a WAV file or whatever).
Assuming this was done on
purpose, it certainly doesn't help you in this case, but I can see how that would be a good "rule of thumb" for external encoders. Perhaps some user's entire
point for using the external encoder is that it can handle a formatting that MC cannot, or something?
But I have no idea if this is on purpose. Might just be an oversight.