Matt,
The Cavalon SB makes a lot of sense to explain your problem. The iS pumps move a lot of fuel, and if the fuel feed line is not large enough the pump will lower the pressure in the feed line to the point the fuel will vaporize and cause the pump to cavitate and loose pressure. This risk is increased when the pump is higher than the tank, or the feed line is long. Rotax specifies a minimum fuel line size to the iS pumps for this reason. Automotive fuels, especially winter blend fuels, will vaporize more easily than Avgas, so if you are using Mogas you could try a tank of Avgas as a test. Downstream of the pumps fuel vaporization for the injected engines is nearly impossible because of the high fuel pressure, but vaporization on the suction side is a known risk.
In my experience, both pumps running will often overcome and mask minor vaporization in the feed line. So it makes sense you may not notice this with both pumps running. So why only when you turn off pump 2? Here is a theory. Even though pumps 1 and 2 are the same, their dynamics are a bit different because of the placement of the check valves. Pump 2, running alone, pulls fuel across the pump 1 check valve. However, pump 1 pulls fuel directly from the feed line. Maybe this check valve can act as a bit of a buffer for the fuel inlet to pump 2 and slightly lower the demand on the fuel feed line. This could explain why you can turn off pump 1 and the engine does not stumble. Also, this theory fits with the fact that after a longer flight the effect is more pronounced… maybe from heat soak of the fuel system. Of course this is all speculation, but it seems like both you and the OP have covered everything else.
Take a look at the attached page from the 912iS install manual. Reading the information at the top of the page shows that Rotax is concerned with vaporization at the suction lines.
