A while ago, fellow Planet Analog blogger Dennis Feucht talked about integration onto ceramic substrates in his post Between Discrete & Integrated Circuits.
The comments went into a brief discussion on hybrid circuits (Figure 1).
I would like to dive into that old world that is still in use today. Companies such as Apta, Dynamic Hybrids, and MicroHybrids are still designing and assembling hybrid circuits in large volume for military, space, and other applications. One of my current jobs is to recreate old hybrids, but on a small volume scale and at a low cost. A typical hybrid, once the lid is removed, looks something like that in Figure 2.
What one sees are some resistors (black segments), traces (in gold), some diodes and/or integrated circuits, and even surface mount capacitors, all on a ceramic substrate (white portion). The resistors are built and trimmed to specification. Other discrete parts are bonded onto gold pads. The semiconductor devices are typically bond wired to a gold trace. Around the edge of the ceramic substrate are posts. The posts serve as bonding pads that allow the connection of the circuit to be made to the pins of the hybrid (Figure 3).
In order to get a hybrid made, one would have to purchase the semiconductor parts unpackaged, and most semiconductor companies would prefer to sell packaged parts since die handling is difficult in general. Also, one would not be able to buy a small quantity of die, but would have to purchase a half or whole wafer's worth, which could be thousands of parts. This is not the best low-cost method.
Since I have to recreate these hybrids that were last made 20 to 30 years ago, the challenge is two-fold. The first is finding replacement parts for the old ones. The second is how to package for a low-quantity production.
My solution ended up being quite simple with today’s component technology. Granted, for the first challenge, finding an op amp that meets the exact criteria as the old can be very difficult, but not entirely impossible, with additional circuit modification, if necessary. Same with obtaining odd-ball resistor value, not in the 1% value chart, but this can be accomplished by careful selection of two resistors in series or parallel.
The second challenge, packaging, ended up being the easy part — we made a circuit board instead of a ceramic circuit substrate. There were further complications and tricks regarding how to package everything onto the PCB. The circuit board would then be inserted into the hybrid package, as shown in Figure 4.
Now a little bit of adjustment of dimensions would get back to that final packaged part.
So, there is an old problem part moved into the modern era with a low-cost solution. What have been your unique solutions to old problems?