That's inadequate heatsinking for a 50 watt driver !
I too purchased a 50W LED floodlight from eBay and even though the bracket was mediocre, I was impressed with the construction and when connected, the light output. Time passed before I was was actually able to mount the device on a specially prepared mast. Much to my dismay, it was only then I discovered it was radiating significant electronic hash over the top of my favourite AM radio station and even worse, the light (and the EMF radiation) cut out after approximately 15 minutes of operation.
The control unit is a sealed box siliconed (no thermal past here!) onto the inside of the back cover and I suspect it is this component which contains thermal sensing. Turning the supply off for a period of time enables the light to operate again for 10 minutes or so but will not restart without an interruption to the power. Because the unit is mounted on the top of a mast on the peak of a shed, when I take it down, I would like to effect a permanent repair to both the radio radiation problem and the cutting out. Information on the voltage to the LED unit and any RF filtering such as capacitor sizes and configuration would be appreciated please. Alternatively, assuming it is the control unit and not the LED element causing the cutout, has anyone successfully overcome the cutting out problem by improving the heat sinking of such a unit?
Update - January 2014
After some delay, I pulled the mast on which the light was mounted and dismantled the beast. The sealled power supply was siliconed into the back of the unit providing no effective thermal coupling to the case. With the power supply removed, the unit operated quite happily without cutting out. I considered remounting the supply in the lamp with adequate thermal paste but if it failed, the difficulty of lowering the mast was too onerous to contemplate so instead, I mounted the power supply inside the shed and ran the direct current output up the mast and wired it directly to the element.
During the process, I ascertained the EMF radiation causing the noise in the AM radio band originates in the actual LED element and no amount of capacitor decoupling will eliminate it. By reading the current through and the voltage drop across the LED element I also calculated the power consumption to be closer to 30 watts than 50. For a while I toyed with the idea of building a power supply capable of delivering sufficient voltage and current to lift the consumption to 50 watts but finally concluded that the unit was bright enough for my purpose and stayed with the original power supply.
I would buy another one but would do the investigatory (and modification) process before I took the trouble to mount it up in the air. I would suggest you avoid buying such a unit if you do not have the DIY capability necessary to do the testing and any necessary modification.
