I found It critical to have a sloped downward run for the oil drain from the separator to wherever it is going. 

 

I question whether or not the valve cover is low enough given the limited height available in the cowl. 

 

I have an  AirWolf unit (no great shakes, I had to fuss with it a lot to get it to work), and it drains down to the left rear push rod tube in which is welded a T section to accept the hose from the separator.  Until I got about 3-4 inches of  drop with minimal bends and a  45 degree downward angle, I found oil pooling in the  separator until it got full, then the gas  stream carried the oil overboard.  Once you shut down, oil drains out slowly and when you inspect, no problem.   Very tricky.

 

Here is the trick with the IO-550.  Late model Continentals have 7 bolts holding the cylinders while earlier units had 6.   Adding bolts reduced room for crankcase vent hole.  So the current design has a 1/2 inch vent hole exiting the crankcase  horizontally at the  crank centerline to the left and discharging into the oil fill port which is a casting  that bolts to the crankcase.  The gas discharges from the 1/2 inch hole tangentially into the fill port which is about 1.5 inches diameter  thus creating a little cyclone that separates droplets from gas.  Droplets hit wall, run down, then down the dipstick hole into the sump while gas goes up and then out the breather, 3/4 inch diameter.  Get a flash light and a step stool, peer down the oil fill port, and you can see what I mean.  It is hard to see, but persist and get an education.

 

This means that the pressure drop arising from the gas spinning in the cyclone must be allowed for.  It is not much, only inches of water, maybe less.  But because of the cyclone in the oil fill port, the pressure in the add-on air/oil separator at the end of the  3/4 inch vent line is LESS than the pressure in the crankcase or valve cover or push rod tube all of which are effectively at the pressure of the crankcase/sump. 

 

You can probably now see the problem.  The  oil flowing downhill out of the air/oil separator must overcome the cyclone pressure drop by virtue of gravity and the greater density of the oil compared to air.  If the gravity drop is insufficient, the gas flows up the oil drain hose into the separator, oil pools in the  separator, and then when it gets full,  the gas bubbles through the oil and carries it overboard. 

 

Further trap.  If the oil in the sump gets below the bottom if the dipstick tube additional problems can  occur.   Remember that the engine has  2-3 quarts in circulation and draining back to the pan when it is running, so operational oil level is below static oil level, particularly when cold and oil is thick.  I had an old Toyota truck with a straight six diesel. On cold mornings once started and warming, the low oil level warning light would come on even though the pan was initially full because the thick diesel oil pooled in the valve cover because it could not drain fast enough when cold.   The pan would empty and the low level light (float operated) would come on. Once warm, no problems. 

 

If you uncover the bottom of the dipstick tube due to low pan level, then gas runs up the dipstick tube and will  not let oil drain downward from the cyclone to the pan.  The cyclone in the oil fill port now fills with oil which can not drain down the dipstick tube until it gets deep in the cyclone.  Oil is carried out with the gas stream into the  separator which overloads and you have a dirty belly, possibly gained in the first minutes of flight (maybe during run up?) until the oil heats and the pan level rises.

 

Answer: keep the oil level UP to prevent loss of oil. 

 

Related: "Everybody knows" that when you fill the engine near to capacity you blow oil overboard.  But if you look at how  the oil pan is separated from the spinning crankshaft (the oil in the pan can not even "see" the crankshaft except through a few narrow slots way overhead), it becomes unclear what mechanism is at work to cause the oil loss.  I wrote Mike Busch with this question.  He replied that after thinking about it, he did not understand the  mechanism either, and would research.  No answer yet. 

 

So here is a case where  "Everybody knows" but in  fact nobody knows anything at all.   No  explanation invoking simple physics has come forward.  All ideas welcome. 

 

Fred Moreno