Potting compounds usually have a higher expansion rate as things heat up, compared to the expansion rate of electronic components. If an epoxy is hard enough, and expands at a greater rate than the component that is embedded in it, it can pull the component apart. Way back in 1981 I had a rash of failures with the 990 when I was potting it in a clear epoxy. One or another of the eleven 1N914B diodes would become open-circuit as the module heated up and things expanded. I changed diode suppliers and I also changed the heat-curing approach, using a lower curing temperature for a longer time period to build up less stress. This solved the problem.
Shortly after I began manufacturing the "A" and "C" versions of the 990 in an aluminum potting shell around 1988, I began using a soft silicone encapsulant. It has an expansion rate that is even higher than the typical epoxies, but it is so much softer that it does not build up the stress that a hard epoxy can do.
Encapsulants have a specification known as "dielectric constant". If you fill the space between components with an encapsulant instead of air, it can change the performance. This is because the capacitance between adjacent component leads will increase if the dielectric constant is higher. Some circuits will not be significantly affected by the change, but others will be.
Encapsulants can help with heat transfer, depending on the thermal conductivity of the encapsulant. They can make the assembly stronger and more reliable. They can hide things.
John Hardy
The John Hardy Co.
www.johnhardyco.com