Originally I had rough running on climb and at full power. Jabiru correctly predicted it was caused by the right cylinders running rich - as proven by checking the CHTs and EGTs at vaious settings. This was caused by air swirling down the intake tube and causing poorly mixed air:fuel to spiral rich air to the right and leaner air to the left.
They advised installing a single vertical baffle in the cobrahead which stopped the air swirling through the carby and feeding rich air to one side of the engine - problem cured.
Later another Jabiru owner here learned, or worked out that adding a crossed baffle also aided the problem of heat distribution.
We both built various versions of baffles consisting of two flat plates of glass which fit into each other and help to further prevent swirling of the air. We then spent hours doing flights with various cobraheads and settings and produced some nice tables and graphs showing the benefits of more complex baffles.
We tried various lengths for the baffle with my friend getting good results with just a short crossed baffle of only an inch or so long. I tried one that is full length of the cobrahead with cutaway sections to allow for joining to the carby and to keep the sensor holes in the carby clear of the baffles.
We both had good results though at that point my crossed baffle was better.
My friend further learned that a grid baffle was used in a lot of automotive situations for similar problems. So he went on and made a grid of about half inch squares. He has a 2200 enging in a Corby Starlet so he has other issues as well, such as very limited space and unusual flow paths for the air induction. Anyway when he developed a grid baffle all his problems were cured in a single swoop.
I took his advice and built a prototype grid baffle but I didn't find it as good. It was better than the single baffle but not as good as the crossed baffle. So I ended up going back to another crossed baffle.
Any grid baffle is the sum of the area taken up by space less the area taken up by the thickness of the baffle material. My friends baffle was better made than mine - thinner plates, less resin used to smooth out the joins etc. I think mine was probably worse because the thickness of the plates stopped more air going through. I got better results from the crossed baffle so I went back to that and was satisfied.
We now have two engines running on crossed baffles and one on a grid baffle and all are doing well.
Obviously you can't do any of this with any accuracy unless you have all cylinders being monitored.
