I had some slight longitudinal free play in my rods. Some were making noise when the head brutally changed its direction on a diagonal movement. It probably did not help making better prints also.
So I re-designed my caps once again. This time I embedded one nut in the internal side, and it becomes very easy to prevent any rod from moving on its axis.
Update (thingster derivative): you can also add an external secondary nut in order to lock the screw very tightly (as show on the first picture). Else you may like the thicker but more efficient derivative by thingster which embeds a nylock nut.
My previous designs required that the inner cone had to be taylor-made. It was not very convenient, and it was not fully durable because of the material... This revision gets rids of the issue by including a metallic M3 bolt in the center.
You'll just need an M3 hex nut in the inside slot, and one more M3 screw. Tighten it until you get no more longitudinal free play on the rods by pushing it forward and backward while you screw. Then unscrew a little bit. The resulting friction is probably neglictible b/c of the small surface contact (metal on metal) and the location on the rod, but the purists could even file the screw in a conic shape ;)
Just use the shorter M3 screw you have, so they do not protruding uselessly out of the printer frame. Mine here are 5 mm M3 screws, without counter-nuts.
I highly recommed my "nut calumet" to hold the matching nuts in the inside of the printer walls: http://www.thingiverse.com/thing:44118
You can add a second nut, on the head side of the screw to secure the position (as shown on one of the unassembled caps in a picture). I did not have to because the PLA holes are tight enough for me (so far). I have yet to see if the rotating rods could ever unscrew the central bolt, but I don't think so.