The unijunction transistor came and went as well 

The unijunction transistor came and went as well 

unijunctions have their place- I designed one into something quite recently, and then spent a disproportionate length of time at the review meeting, explaining to younger chaps how it replaced the microcontroller they seemed to expect in a simple timing circuit.  It seems they are not taught to those on modern electronics or solid state physics courses. As I had one on veroboard flashing lights beside my Hornby train set as a kid, I had prior knowledge.

Memristors have surfaced before as the missing link. I suspect a few useful applications will emerge, but not as world shattering as the hype.

MEM switches are another area of great interest to those of us that sometimes want to move a really small current really fast.

Mike

There is at least one product (made by a past employer of mine) which is based on unijunction oscillators which (if I've remembered the right product) I know is still being provided into a safety-critical and high reliability application. I never came across them being used for anything other than oscillators, but they are a very neat solution for those.

I had completely forgotten about memristors until I saw this thread, I do now remember now them being discussed in the press and possibly in my degree in the late 70s / early 80s - it was the sort of thing us geeky types were into, never mind if it has an actual application let's see if we can make one! But then it's very hard to predict when a "solution in search of a problem" might suddenly hit the big time, some never do, but others e.g. lasers and fibre optic cables most definitely do.

well if you need to design a UJT into something new and surface mount, then central semi make some nicer ones although they keep dithering about making them end of life and then making another batch after all.
As well as free running oscillators UJTs are also seen in mono-stables and ring counter designs. (as a solid state improvement on the dekatron... now those I have seen and marvelled at but never actually designed with...) As you say, some things rise and fall and have their moment in the sun..

Mike

Sadly (if it is the product I think it is) I happen to know it's soon going to be replaced with a "black box compatible" digital based system, which replaces a 10 transistor circuit with two (at least) massive FPGAs providing identical functionality!

It's an irony about fail safe systems, they need to be either extremely simple (where every failure mode can be tested) or massively complicated (where every failure mode is detected and mitigated).

The problem is not just the UJT availability, it's all high quality analogue components (this was my world for 20-odd years) including capacitors and particularly pot cores. Whereas once it's implemented in e.g. FPGA then the design can be ported to a new device without a massive rework of the safety argument.

Doesn't have the elegance of a one-transistor oscillator though. Possibly my favourite is the old transistor (or indeed valve) TV flyback circuit - that always stuck me as sheer brilliance at getting a very complex and precise wave shape out of a tiny handful of components. They don't make 'em like that any more! Imagine setting that as an exercise for the student: "you need to create two synchronised ramp waveforms, one at 50 Hz, one at 15.625 kHz, at the end of each 50 Hz ramp you must create this precise weird waveshape, and each 50Hz cycle must have a slight offset so that the 15.625 kHz ramps are interlaced. No you can't use a processer or an FPGA, you can have around 10 transistors and an inductor or two..."