# Do AC input optocouplers require an ESD protection?

I intend on using a circuit I designed for detecting an AC input line for control purposes. One of the inputs will be accessed by third party users and a cable might be connected with a decent length(>3m). The optocoupler I intend on using(VOL628A) has a diode max reverse voltage of 6V and the intended circuit is shown in the following image:

Since the diodes are connected in antiparallel, I was wondering if I should protect them from ESD. If an ESD voltage level should occur, each diode should clamp and protect the other. I know that the diodes react pretty slowly when it comes to ESD and there is a possibility that they get destroyed, but I was also struggling to find a TVS diode with a clamping voltage bellow 6V.

Therefore my question on whether or not should I use a TVS diode. Maybe you could recommend an alternative solution if TVS wouldn't help in this case.

The forward current for the diodes in the opto is +/- 60 mA maximum. The series impedance is 4x 27 kΩ = 108 kΩ. With half the maximum rated current flowing from a threatening surge, the voltage across 108 kΩ is 3,240 volts. This is a scene setter.

For ESD, the event will be over so quickly that I doubt there would be any problem but, with an indirect lightning surge (as per EN61000-4-5) then you might find that a 4 kV surge (sourced from a 2 Ω source impedance) is long enough in duration to harm the resistors. This in turn might cause one to fail short circuit and then damage the opto.

In short, you need to pick the resistor types carefully and ensure they can survive the peak voltages associated with the surge. You won't really find that using a TVS across the opto input is going to help much. You might decide to absorb any residual energy that might get to the opto with a 10 nF capacitor across the input terminals. Do the math to see how big this value can be without upsetting 60 Hz operation.

• I had in mind that ESD or surge could damage the resistors so I decided to use High Power - Anti-Surge designated resistors. I have not really had experience with such resistors and I don't know if they will work in reality. However, they are my only choice. Jun 19, 2021 at 18:27
• Carbon is supposed to work well for that. You did right spreading multiples out in series. I'd spread them between both lines but I don't think it makes a practical difference here. Jun 19, 2021 at 18:35
• In fact, I had the exact idea on spreading them between lines :D However, I wanted to use the neutral line trace on the bottom layer and the PCB will be assembled on top only. I didn't wanna overcomplicate things with neutral line going between layers only to pass through the two resistors. Jun 19, 2021 at 18:41
• @PhillDonn it sounds like you might have got the resistor type right i.e. you are using the right sort of words to imply competence but, the devil's in the detail. If you are looking to detect either 230 volts AC or (maybe ) a 5 volts p-p signal why not consider back-to-back series zener diodes - they will unambiguously block 5 volts of either polarity if chosen to be 5.1 volt types or higher. Jun 19, 2021 at 18:43
• @Andy aka If you suggest using the zener diodes for ESD protection I can definitely say It crossed my mind but I wasn't sure if the zener diodes would react any faster than the optocoupler diodes. And if not, it would only introduce additional costs to my board. Do you think the zeners would improve the situation? Jun 19, 2021 at 18:51

Your D1 and D2 are slow diodes and will not react sufficiently fast for ESD speeds. You need TVS diodes for that.

But honestly, you probably don't need to take any special measures with such a ridiculously high series resistance, the system parasitic capacitance can probably absorb it/

You could add a small parallel capacitance to absorb it better if you want. At 60Hz your system is is so slow that you could calculate an effective capacitance for ESD that is still small enough not to affect your phase response.

The human body model uses 100pF for the capacitance of a human body. With your current resistors you could have 1nF capacitance, which is ten times that and still have a cutoff frequency of 1592Hz. Far higher than your line's 60Hz which minimizes phase delay for your zero-cross detection. That alone will cut the ESD voltage by 1/10th and that's not even considering the energy that your resistors will dissipate yet.

• I won't be doing zero-cross detection, hence the low pass filter on the output which will be connected directly to MCU. I guess I can go even higher with the capacitor if necessary. Jun 19, 2021 at 18:29
• Oh, I see in your post. It's just to detect AC power? Go nuts then. 100nF. Jun 19, 2021 at 18:33
• It is for digital input, indeed. I need 230vac in/0V out and 0vac in/3.3 out. I just wanted to clarify that I don't need to detect the ac line whenever it crosses zero, I just need to detect whether the ac voltage is present or not, hence I used low pass filter to eliminate the 100Hz on the output and obtain DC levels. Jun 19, 2021 at 18:35
• You might want a comparator to avoid slow edge funny business. Jun 19, 2021 at 18:41
• The output will go directly into MCU. I was thinking on avoiding slow edges by filtering the input with a small delay from software. I don't really need fast reaction to the inputs, so the delay will not present a problem to the application. Jun 19, 2021 at 18:58