Comment by Neywiny
3 days ago
Just so long as we remember to check it'll fit the need. I recently inherited a design that used some parts from the 70s and they were not up to the task. Drop-out voltages too high, gate threshold voltages too high; whatever spec could be violated was. Just because it's been used for 50 years doesn't mean it's the right part for the job
In fact, unless ubiquity in availability is a really core part of the design requirement, they almost certainly aren't the right part for the job.
Most parts in most designs aren't anywhere close to being specification-critical. Specifying the 3904 is a great way to say "I need an NPN transistor here, and it doesn't really matter which one" (because, oh man, they can ship a lot of different things in that "3904" bin spec, and they do). So the "jellybeans" are often ideal choices.
When they are not, that is when the design engineer earns their pay.
These parts that become "the part" often have this issue relatively early into the lifecycle. The 741 op-amp is another example. They are often bad and expensive, but they are a default so people put them in.
These days there's also a wide variety of manufacturers making them including very sketchy counterfeiters.
So there are TONS of pin-compatible-ish parts from different manufacturers that have related specs with a ton of variation.
Even 20 years ago, places like TI and ADI were offering parts that are spec-dominant over the 741. Literally every parameter on the datasheet is equal to or better than the 741, and pinout is the same.