In the third year of my engineering studies there was a rumour that there would be a question on the exam about a FET with some kind of “trench”. We scurried to prepare for such an eventuality which, of course, never came up, so I don’t know if this was a real concept 40+ years ago, but today there are certainly “trenchFETs”. In all probability the commonality of the name is merely a coincidence, but I have other examples of prescience.
In my first job as an engineer there was a younger technician who laid out PCBs using dot and tape. Let’s call him P. P was much more interested in his vintage Porsche than work. (As an aside, much later in my career we had another junior engineer who also had a vintage Porsche and didn’t have much interest in his career. Note to self- do not employ any person who drives a vintage Porsche.) P had done a layout of an early microprocessor design- in those days the memory was external and the address and memory busses snaked across the double sided board leaving little space for the any other circuitry.
You may recall my blog “Do you check your PCBs?” where I describe how I check every PCB. In this case checking it was well justified because I gave up counting when there were 100+ errors on the PCB. As he started the re-work, P expressed the wish that there should be single gate ICs so that there could be a device exactly where you need it. It took several decades, but you can now get almost any function you want as a single gate. Of course, the size of even multiple gates in their surface mount packages today make IC density on the PCB much higher than back in those days. I wonder what P would think of multilayer boards as well.
It became obvious very quickly that the clock/calendar function was an essential part of many micro based designs. National Semiconductor made a device that would allow you de-multiplex the display output of a bedside clock chip (which was one of the early drivers of LSI technology and so there were many of them). The first computer readable device I came across was made up of 4000 family devices that filled up a full Multibus board (12”x6.75”). I suppose it didn’t take much foresight to expect that there would eventually be a dedicated IC for this purpose.
I was introduced to the concept of watchdog timers and defensive coding at an Intel seminar and I was thrilled. I had already experienced the despair of trying to understand why a microcontroller would go “nuts” and how to rectify the problem. In quick succession there appeared two application notes from Intel and a couple of articles in the electronics magazines and I was convinced of the concept. I designed and marketed a small mezzanine board made out of TTL logic aimed at the Intel Multibus range of single board computers. It was a few years later that Maxim brought out the first of their single chip watchdog range.
I am told that one of the worst areas for lightning strikes in the world is an area around Pretoria (now Tshwane) in South Africa. We were to design a fire protection system for an aircraft hangar using flame retardant foam. The foam is really corrosive and if erroneously sprayed on uncovered aircraft engines the results would necessitate the scrapping of the engine. In order to prevent this eventuality the system was to be inhibited during the time of a lightning storm. How the hangar was to be protected if the lightning was to start a fire was not included in the specification! We had to resort to an external (and very expensive) lightning detector and we really wanted a semiconductor solution. Not so long ago AMS introduced the Franklin Lightning Sensor It popped up in another of my blogs Niche Devices.
It seems that whenever you have a good idea, the world is such that someone else has had a similar one already. Many of the world’s great inventions have been shown to have been developed within very short time spans by different people. I have had a few ideas that never seem to have occurred to others, or are just stupid or past their prime. Microprocessor based systems used to have many external configuration pins or DIP switches and we had to use multiplexing with an array of diodes or shift registers to multiplex these inputs on to the limited number of I/O pins on the micro. I thought it would be a good idea to have DIP switches with the diode array built in to the package. Extending the concept I also wanted the pullup resistors and parallel-to-serial shift register in-built as a second option.
And another idea: it is possible to get filter capacitors for RS232 communications built into a D-subminiature connector. The thought occurred to me that one could add a MAX232 driver/receiver as well. That would save board space and quite possibly, the use of a PCB in the assembly could obviate the weird spacing of the D-Sub connector pins as well as reduce the number of actual pins.
I am sure almost everyone has had thoughts like these- don’t be shy, let us know in