I'd figure out whether you want a verbal command or a hand signal (at first, you can always add the other kind later) and then start attaching those to ears up/ears down. It's essentially backchaining, I think. So let's say you decide on "papillon" for ears up, and "phalene" for ears down.
"Papillon, Watch me" Ears go up, c/t
"Papillon, Watch me" Ears go up, c/t
"Papillon, Watch me" Ears go up, c/t
and so on, until his ears are going up on papillon because he's anticipating the cue. The I'd start fading the Watch me command until the cue is and only is papillon.
"Phalene, /hand over ears" Ears go down, c/t
"Phalene, /hand over ears" Ears go down, c/t
"Phalene, /hand over ears" Ears go down, c/t
and so on, as before. These could probably be done concurrently, but I wouldn't suggest successively because common wisdom is you don't start stringing things together until the individual parts are complete.
When they are complete, start asking for ears up, c/t, ears down, c/t until he's anticipating again, OR start asking for ears up, ears down, c/t. Were this me doing it, I would start with a couple of the first method to get my dog used to the action and as sort of a way to tell him what I want. Then start asking for both before the c/t. Somewhere in there (when he's going up and down reliably) add in whatever cue you'd like to use for the whole thing. For training the trick, I suggest a verbal cue because it leaves your hands free, but if hand cues are working better just go with that.
Method based of off Ailsby's Principle of Laziness,
http://www.dragonflyllama.com/%20DOG...g1/levels.html