Optocoupler protection

[cosminnci]

Hello,

I use a transformerless power supply from Microchip AN 954, the purpose of this is to detect on off from an microcontroller.
I have the setup as in attachment
so far from my tests when I get a power fluctuation or a spike the optocoupler output is dimming and finally burns out.
how can I protect the optocoupler from this?, maybe somme shotky diode near the zener?

thank you.
regards,
Cosmin

[SeanB]

Make R2 higher in value, 1k is better, it will dissipate a little more power. R11 can be 1k, this reduces the current through the opto to be lower, making the LED last longer. Instead of the LED on the output, use a 10k resistor and a low current LED, as this is a low transfer ratio unit, so it will have at most a 500uA drive capacity. C5 can be 47uF, and add a VDR across the input, 275VAC will do. This will make the opto last longer.

[Alana]

If it goes bad on brown out it means that phototransistor in the optocoupler has too little light from LED to go into saturation and starts working in linear mode - and this means heating up and eventually damage. VDR or other over voltage protection tricks wont work here.
Minimizing current via phototransistor may do the work. Try this [pic]:

[cosminnci]

The calculations for the values are from this form:
http://www.daycounter.com/Circuits/Transformerless-Power-Supplies/Transformerless-Power-Supplies.phtml

with R1 220 and C1 0.33uF the circuit produces 5.1v, 12.9mA and R1 dissipates 0.168W
the 4n25 is calculated to be powerd with 1.5v 10mA and the R11 value of 390 would limit the current, the opto max is 60mA

replacing with 1K the R2 would dissipate 0.68W and produce 12.1mA, not much change, the power is calculated to be dissipated trough R1 and R11 in either combination.
during normal operation all look well but during a spike the led from the opto gets damaged the part after is not too relevant the other led is for local test and can be ignored.
I will try a varistor.
does the zener diode actualy work in this circuit? do I need it?

 

[SeanB]

The zener is needed, it limits the forward current applied to the LED. You can place a 1n4001 in parallel with it as well, as some zener diodes are composed of a single zener diode in series with some forward biased regular diodes.

[cosminnci]

I tested today and managed to damage the zener, it started conducting, but the opto was fine.
I tested extreme conditions that should never occur (shorted the power resistor).
will test the 1n4001 diode solution but I hope I won`t need it.

thank you.
regards,
Cosmin