I concur with other comments. The assumptions for an 12-24V AC dimmable LED make certain compatability model assumptions on the dimmer type. Normally these designs use a buck-boost type that must store energy to sustain the power between 50 Hz half cycles and this stored energy in reactive elements includes a storage cap that must be pre-charged in 1 or few cycles to obtain instant On characteristics.
The down-side is this stored energy has an equivalent watt second or Joule equivalent energy value and when this energy is transferred on power up is a shorter time during a random phase angle where the peak Ac input is 325V nom for 230Vac/50 Hz can challenge a designs ability to withstand or soft start inrush currents. They can operate off AC or DC with these circuits.
Now that's just the LED bulb, but the same holds true for the dimmer with it's stored energy and inrush current and random phase angle and is compounded by the inrush of its load and its input storage needs to charge up without exceeding some max current.
Unfortunately there are always combinations of products that may be incompatible and over the years has caused designs to choose leading edge and trailing edge type control methods in other products to prevent hysteresis on dimmer functions.
The third challenge is your contact switch will likely have contact bounce and this during a peak AC sine can cause massive inrush current to caps with mechanical bouncing high dV/dt or arcs from V=Ldi/dt and then LC resonating effects during the turn-off arc.
This explanation won't help you but there may be in expensive ways to curb the over-stress condition if we knew exactly the root cause of the component(s) that failed.
- OVP, Turn-off over voltage: from V=LdI/dt CLC-line filter and or MOV to dimmer or maybe an X cap 0.047uF across input switch.
- OCP, Turn-on over-current: An ICL metal-oxide NTC soft-starter (cheap)
- Switch debouncer: A ZCS triac control by power switch (not common consumer product, but common in SSR's which often cost more than your cheap dimmer or maybe the X cap solution above.
This reminds me when we tested OEM power supplies, one of my test requirements was 10k AC power cycles with random phase during life test, on,off, on , off. We did the same for OEM disk drives. It should be common knowledge for any Test Engineer to do this. Obviously an oversight on this dim product.
Best guess? get an Cap and an ICL from Digikey and call their hotline tech support or email them for your shopping solution. The holding current for 50W on 230Vac means the ICL must be rated for 500V minimum and 0.5A to 1A.
Get extra, they are cheap and install in series between switch and dimmer input. You could also get ones for higher current on dimmer output for 5A. They come in different Curie temperatures and are designed to run hot from 55 to 125'C Choose the lowest range for an office environment.
Good luck shopping.