8
\$\begingroup\$

Many transistors are very similar to one another, so much so that substitutions are fairly easy to make. Ft around 300 MHz, gain characteristics very close, etc etc. Can anyone speak to why so many subtle "flavors" are manufactured? Not being an engineer, perhaps I simply do not know enough about the tiny differences. Some are obvious, for example a 2N3355 offers vastly higher Ft than a 2N3904; a 2N2222 handles a little more power, etc. Noise figure and capacitance are often close, too.

Are these subtleties truly enough to drive the industry to produce the wide variety we see? Just curious if someone with design manufacturing background can offer some insights, thanks.

\$\endgroup\$
9
  • 7
    \$\begingroup\$ Keep in mind that these are developed and produced for decades, each year developing new products because new technologies become available. \$\endgroup\$ – PlasmaHH Nov 9 '17 at 14:25
  • 3
    \$\begingroup\$ Think about things that are available in daily life like your smart phone how was it 5 years ago; how much has it evolved. In short term, there are many updates for softwares for the same device, every update fixes some smaller bugs that pop-up. This explanation goes along with any technology (be it BJT or MOSFET or any other). Efforts are to make better things with time certain changes are subtle (but over a long period of time it could have major changes) and certain changes are drastic :) \$\endgroup\$ – rsg1710 Nov 9 '17 at 15:11
  • 3
    \$\begingroup\$ As was already pointed out, the "problem" is that some products are available for decades, even if there are already "better" alternatives. The reason is, that even minor differences "may" matter and "may" break an existing application if something was designed "on the edge". Some things may even be only working because of a flaw that negates another design issue. So even if a component is superior in every way could break things. PCB designers really HATE to search for and use replacement parts for a product that was working for years. \$\endgroup\$ – Rev Nov 9 '17 at 15:18
  • 2
    \$\begingroup\$ Why are there more than a thousand different toothbrushes? How come that my store alone carries 10 different brands of toilet paper? \$\endgroup\$ – pipe Nov 9 '17 at 15:32
  • 4
    \$\begingroup\$ As well as performance differences, there have been historically three popular pinouts in the old TO-92 through-hole package. One was popular in the US, one in Europe and one in Japan. That tended to multiply the number of part numbers. For example, a 2SC1815 was roughly equivalent to a 2N4401 or a BC548, the first was E-C-B, the second E-B-C, the third C-B-E. Some transistors were produced in multiple pinouts with the very similar or identical part number markings (obviously without the blessing of standards organizations). \$\endgroup\$ – Spehro Pefhany Nov 9 '17 at 15:47
16
\$\begingroup\$

In the early days of BJTs, every year the processes would improve and there would be new applications for these parts with new demands. As a result, semi-conductor companies were frequently coming out with new parts. Some were just better versions of older parts, and some were to get into a new niche that didn't have a BJT for it before.

Now consider that quite a few companies were all doing this simultaneously. Each wanted to have a product line that covered most of the market. In some ways it was a game to make your new transistor with slightly better specs than a new transistor the competition just came out with.

All this resulted in a large number of transistor models, with many of them being largely equivalent, or being outright supersets of others.

Eventually a few models got used in high-volume products. The high volume drove down the cost, so others started using that model too when the specs were good enough. When there is a particularly cheap transistor for a common application, you're going to use that one for new designs, even if 20 other transistors would work just fine too.

Now decades later, we are largely left with the few models that happened to get used in volume, have their production cost drop, and become the standard or "jellybean" part for a certain range of demands. The 2N2222A, 2N3904, and the like are in this category, although I use 2N4401 and 2N4403 for my jellybean NPN and PNP.

There is now much less development of new discrete BJTs. Look at the product line of Ztex(?) (meanwhile bought by Diodes Inc or ON-Semi or something like that). They were one of the last really specializing in and doing development on BJTs. It is one broad product line with parts covering a wide spectrum, but each being different enough from the others to make sense. The total number of parts is much smaller than is out there, but covers most of the applications. It shows what is possible when one group thinks out a single consistent product line, and that then doesn't get pressured by lots of competitive offerings popping up.

\$\endgroup\$
2
  • 2
    \$\begingroup\$ +1. The market popularity/ customer acquisition thing makes a huge difference. When a big player in the market, like IBM back then, or probably Apple today, choses a part to buy by the millions for some new product, that part suddenly becomes the popular part. Classic example is IBM's choice of the Intel family processor, that event more or less killed of the competitors, Motorola and Rockwell's CPU offerings. \$\endgroup\$ – Trevor_G Nov 9 '17 at 15:37
  • 5
    \$\begingroup\$ @Trevor: And eventually IBM's own Power PC risc processor. \$\endgroup\$ – Olin Lathrop Nov 9 '17 at 15:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.