The Glass design was wrong! The design philosophy must make the beginning of the process and the beginning of the machine design forgiving.
You can medicalize the glass. For instance, you could evaporate copper, over plate with nickel and then finish with gold. If the wiring philosophy is reworkable, which is designable you can do this.
No, we knew we weren’t gonna get great eels in the beginning, we knew that the main frame that wouldn’t always be perfect electrically or logically. So we had a total rework philosophy.
This is the key, total rework philosophy. This is why your glass would have a metallized pad that would be gold coated if you wanted, but I’m not familiar to know whether this is possible.
But the key is total rework philosophy, it’s needed in the bring up stages of the process and the machine design of the main frame/GPU
No, The deeply embedded wiring net was totally deleted with laser ablation on the top surface of the MLC substrate. We used a wiring redundancy design to make this deletion. Here is the key: Werewired the ENTIRE NET using DISCRETE wires that were ultrasonicly bonded to “thick” Gold pads. We discreetly rewired the entire net on the top surface. This was done during the bring up of the modules and substrates as they populated the main frame. Subsequent design releases re-embedded the wire into the ceramic substrate.
@JSCC2020Lee You are welcome!
(Trust me, we did repair wiring, it was the lunch pin to everything. Our philosophy was, you had to be able to rework and repair every single chip and every single substrate.)
IBM produced hundreds of thousands of the MLC substrates. These substrates and the TCM module were the heart of the main frame for over 20 years.
You can trust what I say. You shouldn’t talk about wiring tricks
MLC is Multi Layer Ceramic: Aluminum Oxide with glass and Molybdenum wiring encapsulated in the Thermal Conduction Module. The semiconductor chips were attached with the IBM C4 PbSn reflow connection technology.
From the electronics packaging laboratory & semiconductor facilities in East Fishkill, NY
From Grok:
https://t.co/jMke8TGRYL
@JSCC2020Lee Well, the substrate technology offered exceptionally high yield. We did have an insitu discrete wire repair process and full net strip for the defective net interconnect. Full net strip was required because of the potential for ringing and reflections in the substrate.