3 ms·

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-

by Neywiny 6mo 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

rcxdude 6mo ago
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.
- exmadscientist 6mo ago
  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.
pclmulqdq 6mo ago
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.
- colechristensen 6mo ago
  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.
  - pclmulqdq 6mo ago
    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.
    - exmadscientist 6mo ago
      Twenty? Try forty. These things don't get chosen for performance reasons. They get chosen by: 1. Inexperienced or simply bad designers (this is real, sadly) 2. Designers minimizing cost or maximizing part availability (this is 99% of the volume these days: you just need something that does the function, and pretty much anything will do) 3. Designers who had them lying around in the parts bin when they got the circuit to work on the bench... and when it works well and cheaply you stop asking questions and ship it! (and yes the word "well" is doing some very heavy lifting there) Also, nobody counterfeits 3904s and 741s and 324s. They're so easy to meet spec on that you can just ship them as genuine with your own label on top. A fair bit of counterfeits of other chips are 358s with fake top markings. That is the kind of cockroach these guys are!

by Neywiny 6mo ago

rcxdude 6mo ago

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.

exmadscientist 6mo ago
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.

pclmulqdq 6mo ago

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.

colechristensen 6mo ago
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.
- pclmulqdq 6mo ago
  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.
  - exmadscientist 6mo ago
    Twenty? Try forty. These things don't get chosen for performance reasons. They get chosen by: 1. Inexperienced or simply bad designers (this is real, sadly) 2. Designers minimizing cost or maximizing part availability (this is 99% of the volume these days: you just need something that does the function, and pretty much anything will do) 3. Designers who had them lying around in the parts bin when they got the circuit to work on the bench... and when it works well and cheaply you stop asking questions and ship it! (and yes the word "well" is doing some very heavy lifting there) Also, nobody counterfeits 3904s and 741s and 324s. They're so easy to meet spec on that you can just ship them as genuine with your own label on top. A fair bit of counterfeits of other chips are 358s with fake top markings. That is the kind of cockroach these guys are!