I'm designing a power amplifier and I want to sense power-drops early, in order to properly shut down the class-D amplifiers, SMPS's and save status in the microcontroller, before brownout occurs. In order to do this, I'd like to rely on AC mains sensing with an AC-input opto-coupler.

I could do this in a very simple way, by adding appropriate series resistor(s) to the LED's, but with about 2mA that consumes about 0.5W permanently. Hence I'd rather rely on a capacitive voltage drop.

The 'problem'

I'd like to protect this sensing circuit from the nasties that come from the mains power line. So, I've looked at the surge behavior of several components involved:

  • opto-coupler (SFH6206): 2,5A surge, check
  • X2 safety series capacitor (TDK B3293 47nF): OK for series operation, this is however the problematic part for surges.
  • series resistor (Panasonic ERJP08 ): 10k resistor should be able to endure 160V during surges.
  • TVS diode across opto-coupler and series resistor (Littlefuse 8.0SMDJ100CA): can absorb enough power and clamps to appropriate voltages with regard to the other components.

My questions

In normal operation, the below schematic, using the components enlisted above, behaves just fine and will be reliable. I am however hesitant when it comes to power line surges.

First of all, what exactly is a power line surge for this type of domestic, consumer type device for indoor use. I've seen most references suggesting a 8/20µs peak with a top of 6kV, I've however also seen documents suggesting a 0.5µS rise and 6kV peak, with 100kHz decaying oscillation waveform, with a 60% decay between the first and every following positive or negative peak. I'd like some guidance from knowledgeable people here.

It also seems that any thermal fuse or circuit breaker will react much slower (in about 10ms), so any surge damage will occur before mains power is cut off. Correct?

Last problem, assuming I have to consider the 8/20µS 6kV peak, how will this affect the X2 capacitor. It is certainly out of specs. But looking at X2 safety capacitors, pulse types are not recommended for series operation and types for series operation do not withstand surges very well.

Below is the simulated schematic.

enter image description here

During the surge, I observe the following peak values in the simulated circuit:

  • D3 clamps at 144V/18A, peak power of 2.5kW, this is OK.
  • R1 clamps at 141V/15mA, this is OK
  • C1 and C2 conduct 18A while rising up to 2.9kV with a peak power of 49kW, this is problematic, I think.

So my additional question is: can this be improved upon or should I accept that the capacitor will suffer and self-heal? And if it does, will it be fit for further use? It's capacitance will drop, current through the opto-coupler will be affected, but by which order of magnitude can I expect degradation?

As a last note, I can imagine that a series protection device can protect the X2 capacitors. But which type of device should that be, it should be able to absorb enough energy, it should act fast and it should have low resistance in normal operation.

  • \$\begingroup\$ How will a bi-directional opto-coupler help? The SFH6206 is not a device of this type. \$\endgroup\$
    – Andy aka
    Commented Jan 30, 2021 at 14:55
  • \$\begingroup\$ With bi-directional, I mean AC input, it conducts both positive and negative current at the input. It is however irrelevant for this question, I believe. \$\endgroup\$
    – gommer
    Commented Jan 30, 2021 at 15:19
  • 1
    \$\begingroup\$ Well, if you want folk to be unreasonably steered left and right when reading the question, leave it as it is. \$\endgroup\$
    – Andy aka
    Commented Jan 30, 2021 at 15:20
  • \$\begingroup\$ Pardon me? If you didn't understand me yet, this question is about the surges, about the X2 capacitor behavior and surge protection strategies. The opto-coupler is part of the design, but irrelevant with regard to the essentails of this question, I believe. Corrected nevertheless, for people who get easily distracted. \$\endgroup\$
    – gommer
    Commented Jan 30, 2021 at 15:29
  • \$\begingroup\$ I got distracted for sure, that's why I mentioned it. \$\endgroup\$
    – Andy aka
    Commented Jan 30, 2021 at 15:30

1 Answer 1


Well, I'll answer my own question, partially at least.

After a lot of searching it seems reasonable to assume that a residential AC mains peak can safely be assumed to be a 8/20µS peak with 6000V peak. It is also reasonable to assume the current peak will be a maximum of 6000A.

With regard to the additional protection device, I'm still not aware of series devices that could serve the purpose. An alternative would otherwise be a MOV at the input instead of a TVS across opto-coupler LED and resistor. Clamping voltage will be higher (worse dynamic behavior) but in return they will allow much higher energy absorption.

All in all, I will abandon this whole approach in favor of another energy efficient technique that I found on this very platform (but found through Google, not with the SE search function): Efficient method for isolated detection of AC voltage without a stepdown transformer Shame on the question poster for not accepting the excellent answer found there.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.