Here's another data point for the mixing tube theory. The 3-nozzle burner in Tom Marshall's Stanley Model K has a center mixing tube that is 19" long, and 2 outer mixing tubes that are 16" long. I've heard this car when the burner howls with all 3 jets being fed, and it sings a duet - 2 distinct pitches.
Clearly we would expect different lengths of venturi to play a different note but how does that square with the idea that the flame oscillation frequency is synchronised by the venturi resonance? If the venturis are resonating at different frequencies how can the flame follow, unless there is more than one mixing chamber due to partitions below the burner plate. Are there? Our idea of having two lengths of venturi was that this would confuse the flame and stop the howl - as I said - it didn't!
Edited to add - of course if by chance the lengths of the venturi tubes were such as to produce resonances spaced apart by an octave then the two notes might match with a common flame oscillation frequency.
Edited 1 times. Last edit at 04/10/10 04:20PM by Mike Clark.
Having seen Mike's video, and John's reference to "first, find your resonant cavity", I'm drawn to the conclusion that we're looking at a Helmholz resonant cavity. Examples are when you blow across the neck of a glass bottle and an ocarina.
I think the "cavity" this case is the mixing chamber along with the venturi tubes. The natural frequency will change with local sound velocity. Mike's video seems to be demonstrating this variation between when the temperature of the mix is held down because the fuel aerosol is still evaporating, and when it rises because the fuel has already acquired all the latent heat of vapourisation it needs before exiting through the jet.
Interesting point about the overblowing. Can you "overblow" an ocarina? Sorry, I don't know.