Potting compound serves a number of purposes:
1. For some products, it hides the details of the actual circuit. In some cases, the circuit is unique and the "intellectual property" deserves protection. In other cases, the circuit is just a scam, and the reality of the scam needs to be hidden. The circuit of the 990 was published by Deane Jensen, so I chose to pot the original 990 product in a clear epoxy to hopefully show off the quality of components and layout, etc. It certainly has been a conversation piece. One of my friends saw a clear 990 for the first time, way back when. He stared at it for some time, shortly after smoking a big joint. Finally, he turned it upside down and right side up, then said "It would be great if it snowed inside".
2. The potting compound can provide structure to hold things together. In the 990C, there are way more than 100 holes in a p.c. board that is roughly 1 square inch. Copper traces and pads are very small, so the added strength and structure of the potting compound is important to make it reliable.
3. The potting compound can help to thermally couple parts together. Different compounds have different levels of thermal conductivity, so performance will vary.
4. The potting compound can help to conduct heat away from the components that generate excess heat. Again, the net result will vary depending on the thermal conductivity of the specific compound.
5. The potting compound has a "dielectric constant" that can affect the performance of the op-amp. Two conductors that are separated by a dielectric (insulator) of air will perform differently from those with a dielectric of epoxy, or silicone, or whatever. Different compounds will have different dielectric constants.
6. The potting compound has an expansion rate known as the thermal coefficient of expansion. This is the rate of expansion per degree of temperature increase.
7. Potting compounds have a hardness rating, typically rated in "Shore D" or "Shore A" units.
Generally speaking, potting compounds expand at a faster rate than the components of an op-amp (or any circuit). If the expansion rate of the compound is sufficiently higher than that of the components, and the temperature change is high enough, and the compound is hard enough, the electronic components can be damaged as a result of the build-up of stress. I had a rash of failures in the clear 990 op-amps in early 1981 due to 1N914B diodes being pulled apart on a microscopic basis as the modules heated up. One or more diodes became open-circuit. Of the eleven 1N914B diodes in the original 990 circuit, certain ones becoming open-circuit because of thermal expansion stress would have no effect on the basic performance of the 990. Other 1N914B diodes would cause a catastrophic failure. I changed diode suppliers. I also changed the curing procedure for the epoxy to hopefully reduce the buildup of stress by using a lower curing temperature for a longer time period.
The 990C that I make has an aluminum potting shell. The two output transistors which create most of the heat in a 990 are bonded directly to the wall of the aluminum shell for maximum heat conduction. Aluminum has a much higher thermal conductivity than most epoxies. The heat from the transistors is quickly conducted away from them, distributed evenly around the entire surface of the shell, and dissipated into the surrounding air. The only way to improve the dissipation would be to add fins, making it a much larger package, or providing some type of air circulation. Both are possible, but, within the limits of the package format, the aluminum shell is the best approach.
Epoxies would have to have a large amount of thermally conductive filler material added to them to come anywhere near the thermal conductivity of aluminum. There would be so much filler that the epoxy would be very thick and hard to work with. The dielectric constant would probably be a lot higher. It would also be so hard that it might cause stress fractures in components due to the differences of thermal expansion.
I use a soft silicone encapsulant. It is sufficiently soft that it does not build up a significant amount of stress on the components as things heat up and cool down. The aluminum shell keeps the heat evenly distributed around the entire circuit, and dissipates the heat as effectively as possible. Components that need to be thermally coupled are located next to each other. The even distribution of heat by the aluminum shell keeps the thermally coupled parts at as closely matched a temperature as possible. The silicone has a much higher temperature rating than most epoxies. It is more expensive, and requires special procedures, but I think it is the best way to go for this particular situation. The 990C package is not as fun to look at as the original clear 990 package, but it is better in terms of basic performance.
Trade-offs. Life is full of them.
John Hardy
Are you SHURE I didn't make those????!! ;D
