Delete comment from: Ken Shirriff's blog
Bell Labs came up with wire bonding in the mid-50s: https://nepp.nasa.gov/index.cfm/20911
One of my colleagues designed a hybrid in the late 80s with about 200 dice. He and another colleague manually wirebonded the first two prototypes. I seem to recall there were over 2000 wires on the completed assembly. Most of the chips were commodity parts but the manufacturers hadn't released them in an SMD package yet. By the time they did, we had an ASIC designed and fabbed to do the same thing.
I designed a couple of hybrids using custom ASICs myself, fortunately they were regular enough to allow for automated wirebonding. When the first article of one came back from fabrication, our assembler accidently brushed it and caused some short circuits so I had to mount it on a stage under a microscope and carefully tease a few loops apart. It took a few passes to not introduce more shorts as I cleared them out. If memory serves, they were on a 10 mil pitch in two layers or levels.
Mar 9, 2019, 3:04:09 PM
Posted to Op amp on the Moon: Reverse-engineering a hybrid op amp module