My wife has an LED desk lamp: a head unit with two rows of LEDs on an Anglepoise-type arm connected to a mains PSU. There are four membrane switches: on/off, brighten, dim, bank-select (one, other, or both rows of LEDs). They are normally-open, momentarily-closed switches which control a PCB that seems to use power transistors and probably pulse-width modulation to drive the LEDs.
The on/off switch had started to become unreliable and eventually stopped working altogether. The idiots who designed the light had made the four-switch panel inaccessible by welding it shut (no screws to remove it). Grr!
I was able to fix it by soldering a couple of wires to the correct pins on the PCB that the switch unit connects to by ribbon cable, and then feeding those wires to a momentarily-closed microswitch that I happened to have.
But the failure mode of the switch was intriguing. The switch still worked - a resistance meter still changed state - but instead of going to "zero ohms" when it was pressed, it went to about 100 k ohms. I wonder what could fail in a switch that would cause its momentarily-closed position to be high resistance. Severe corrosion of contacts?