This question already has an answer here:

I've always wondered, what is it about mosfets that seem to make it more a preferrable choice over BJTs? In fact, are bjts still used in today's modern electronics?


marked as duplicate by CL., JRE, laptop2d, Daniel Grillo, The Photon Nov 8 '16 at 18:53

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • \$\begingroup\$ Discrete BJT's are substantially cheaper than discrete FET's. A generic BJT in SOT-23 package is only around $0.0075 in US currency (when purchased in large volume). A MOSFET will be at least $0.02 (in volume). If you need to design analog circuitry using discrete components, BJT's are MUCH easier to use. \$\endgroup\$ – mkeith Nov 4 '16 at 6:08
  • \$\begingroup\$ @mkeith True. I buy BJTs by the 10,000 lot size in TO-92 at about 0.4 cents each ($40) and give away to students what I don't need (which is almost all of them.) I can't do that with MOSFETs. If you see any MOSFETS in TO-92 (hobbyists, remember) at those prices, anyone please let me know. Now if I could just figure out how to use those cheap BJTs as resistors, I could save even more money! \$\endgroup\$ – jonk Nov 4 '16 at 6:16

Yes, BJTs are still used. BJTs have higher transconductance, which gives them higher gain. They are ideal for amplifiers, where you want to maximize the gain. In opamps, for example, you are limited to two stages for stability reasons. You will notice that BJTs are also dominant at MW(300MHz - 300GHz) frequencies, and for low noise amplifiers.

MOSFETs are good for switching, because they use power only when they change state (if you ignore leakage that is). BJTs need to constantly draw power. That is why CMOS is prevalent today.

There is actually considerable amount of research going on in how to replicate the BJT performance in CMOS technology, especially for satellite applications. If successful, one can significantly reduce the cost, and introduce SoC solutions to the market. This is because the CMOS process is more cost effective, i.e. requires less masks when an IC is manufactured.

  • \$\begingroup\$ I love mosfets. \$\endgroup\$ – Bradman175 Nov 4 '16 at 0:40

The main thing going for MOSFET's is Silicon and CMOS.

Silicon is just purified sand. It is great for making MOSFET's because it naturally grows a insulating oxide (glass). Both of these make it cheap (for example other semiconductors require elaborate processing for a much lesser quality, unstable, leaky oxide -- GaAs). Silicon has also been the industry giant (by huge margins) for many decades so the technology is quite mature.

The structure of MOSFET's is also much simpler than BJT's. They handle digital logic well and can be used to create CMOS logic which uses only static-leakage power consumption (unlike the BJT which, since the base draws some current, always uses static power).

About five decades ago the leading computer makers got together and decided that analog computers were out, and digital computers were in. This pretty much cemented the MOSFET as transistor of choice for digital IC's, since it lends itself so well to static-powerless gates (except for non-ideal leakage) in CMOS configurations.

But BJT's are stilled used in critical analog applications for high gain or better noise.

  • \$\begingroup\$ "A couple decades ago ..." what exactly happened? i dunno of any analog computers in use even in 1996. there were analog electronic products, as there is now. but those op-amp based analog computers sorta went out before the '70s. \$\endgroup\$ – robert bristow-johnson Nov 4 '16 at 4:16
  • \$\begingroup\$ @robertbristow-johnson: couple was the wrong word. I was referring to the decision by Fairchild (and others) to favor digital over analog computers (aggravating Widlar) in the 60's. \$\endgroup\$ – jbord39 Nov 4 '16 at 4:45
  • 2
    \$\begingroup\$ @jbord39 Time is a-flying, that's the friggin' problem. :-] I take advantage of this comment to point out that one thing that I did not see in the answers so far is the fact that MOS are popular thanks to microprocessors and that is so not only because of (very important) power considerations but also because the CMOS cell allows for a higher level of integration. Olin says it better here: electronics.stackexchange.com/questions/26404/… \$\endgroup\$ – Sredni Vashtar Nov 4 '16 at 5:42
  • 2
    \$\begingroup\$ We use 'analog computers' every time an op amp integrates, differentiates, or sums or differences signals. While IC digital computers sell by the millions, op amps sell by the billions. We just don't call them 'computer' nowadays. \$\endgroup\$ – Whit3rd Nov 4 '16 at 8:59
  • 1
    \$\begingroup\$ @Whit3rd: eh, no; not quite. Calling the ANALAC or the analog computers Lockheed Martin/Boeing used before the digital revolution an op amp is like calling the microprocessor a 7404. The difference is levels of complexity and hierarchy. \$\endgroup\$ – jbord39 Nov 4 '16 at 14:21

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