My analogue design engineer colleague still uses this approach.
One rationale I have heard him say is that he prefers to do things with transistors instead of logic ICs because of the direct availability of transistors, compared to ICs. It's an advantage when doing repair. You always have transistors laying around.
However, when you need to repair ICs on a PCB, you need a specific IC doing your function, which you may not have in stock.
I wondering why the argument doesn't apply for transistors. The assumption seems to be that it doesn't matter much which transistors are used.
I don't know if anyone can validate what my colleague is saying.
shrug it's fine if you like doing it that way, you're doing a small set of functions, you don't mind the parasitic capacitance, and board space and power consumption is not an issue. And desk space: https://www.edn.com/jim-williams-desk-circa-2007/ (if it was good enough for Jim Williams, it's good enough for you)
> The assumption seems to be that it doesn't matter much which transistors are used.
Weeell .. sort of, in that the variance is so high that for most designs the parameters are designed to be irrelevant. Provided you get current capacity right. Sometimes you genuinely need matched transistors though.
(One of the classic synths relied on a specific batch of "faulty" transistors, which made its properties almost unreproducible until the full-digital era)
I get the idea, and have worked with people who hold that kind of opinion. But it's getting more and more untenable, if you want to have competitive products. It's the same why you generally just can't use leaded parts anymore - almost everything is surface mount.
ICs are doing so much more these days, that it would just take up too much space, or need whole extra boards to do what a tiny QFN part does. In terms of repair, even the individual transistors (of the few we use these days) are so small that field repair is extremely difficult - it's hard to replace them without a proper lab setup, so we mostly have spares of whole boards or modules for field repairs, and then send the faulty board back to the lab where somebody can repair it with a microscope, hot air rework station etc.
However, when you need to repair ICs on a PCB, you need a specific IC doing your function, which you may not have in stock. I wondering why the argument doesn't apply for transistors. The assumption seems to be that it doesn't matter much which transistors are used.
I don't know if anyone can validate what my colleague is saying.