The complaints that I kept on receiving were undoubtedly because of the initial switch ON surge which kept destroying the 1 watt LEDs connected at the output of the circuit.
The above problem is pretty common with all capacitive type of power supply, and the problems has created a lot of bad reputation to these types of power supplies.
Therefore normally many hobbyists and even engineers opt for lower values capacitors fearing the above consequence in case larger value capacitors are included.
However as far as I think, capacitive transformerless power supplies are superb cheap and compact AC to DC adapter circuits which requires little effort to build.
If the switch ON surge is tackled appropriately, these circuits would become spotless and could be used without the fear of any damage to the output load, especially an LED.
During switch ONs, the capacitor quite acts like a short for a few microseconds until it gets charged and only then it introduces the required reactance to the connected circuit so that the appropriate amount of current only reaches the circuit.
However the initial few micro second short condition across the capacitor inflicts huge surge to the connected vulnerable circuit and is sometimes enough for destroying the accompanied load.
The above situation can be effectively checked if the connected load is inhibited from responding to the initial switch-ON shock, or in other words we can eliminate the initial surge by keeping the load switched OFF until the safe period is reached.
This can be very easily achieved by adding a delay feature to the circuit. And that's exactly what I have included in this proposed surge protected hi-watt LED driver circuit.
The figure shows as usual an input capacitor, followed by a bridge rectifier, until here everything's pretty common capacitive power supply.
The next stage which includes the two 10 K resistors, two capacitors, transistor and the zener diode form the parts of the important delay timer circuit.
When power is switched ON, the two resistors and the capacitors restricts the transistor from conducting until both the capacitors get fully charged and allows the biasing voltage to reach the transistor base, illuminating the connected LED after a delay of about 2 seconds.
The zener is also responsible for prolonging the delay for two seconds.
The 1N4007 diode across one of rhe 10K resistors and the100 K resistor across one of the 470uF capacitors helps the capacitors discharge freely once the power is switched OFF so that the cycle can repeat enforcing the surge protection into action on each occasion.
More number of LEDs may be connected in series for increasing the power output, however the number may not exceed 25 nos.
UPDATE: A more advanced design is discussed in this zero crossing controlled surge free transformerless power supply circuit
The videos below show the LEDs illuminating after about a second on power switch ON.
Complaints From the Readers (Resistors burn, transistor becomes hot)
The above concept looks great but is probably not working well with the proposed high voltage capacitor power supply.
The circuit has to be researched a lot before it becomes completely free from troubles.
The resistors in the above circuit are unable to withstand high current requirements, same is true for the transistor which also becomes quite hot in the process.
Finally we can say that that unless the above concept is thoroughly studied and made compatible with a capacitive transformerless power supply, the circuit cannot be put into practical use.
A New and Probably a Much Robust and Safe Idea
Even though the above concept failed to work it doesn't mean the high voltage capacitive power supplies are completely hopeless.
There's one novel way of tackling the surge issues and making the circuit failproof.
It's by using many 1N4007 diodes in series at the output or in parallel to the connected LEds.
Let's have a look at the circuit:
The above circuit is yet to be tested for many months, so these are still early days, but I don't think the surge from the capacitor will be high enough to blow the 300V, 1 amp rated diodes.
If the diodes remain safe so will the LEDs.
More diodes may be put in series for accommodating more number of LEDs.
The first circuit attempt which seemed to be vulnerable itself to surge causalities can be effectively remedied by replacing the power BJT with an 1 amp mosfet as shown in the following diagram.
The mosfet being a voltage controlled device, here the gate current becomes immaterial and therefore a high value 1M resistor works perfectly, the high value makes sure that the resistor does not heat up or burn during the initial power switch ON. It also facilitates a relatively low value capacitor to be used for the required delay ON surge suppressing feature.
A little investigation revealed that the high voltage transistor in the first diagram is actually not needed, rather it can be replaced with a high current Darlington TIP122 transistor as shown in the following diagram.
The high voltage surge from the capacitor becomes ineffective against the high current specs of the transistor and the LEDs and no damage is caused to them, in fact it forces the high voltage to drop to the specified allowable safe limits of the LEDs and the transistor.
The TIP122 also allows the use of a high value base resistor thereby making it sure that it does not become hot or blow off in the course of time, it also allows the inclusion of a low value capacitor at the base of the transistor for implementing the required delayed switch ON effect.
The above design further improves in terms of safety and surge suppression when used in a common collector mode, as given below: