1
\$\begingroup\$

I tried to find an answer online, but didn't find anything that would make my doubt go away. The thing is: when you want to protect any input, that is referenced to ground, you use ground-referenced diodes, varistors, capacitors etc. But what about an input, that is not referenced to your system's ground? Let me show you my schematic: optical input schematic

It's a normal, iptoisolated input, that should work with about 5-24 V (AC RMS/DC). The resistors are normal ones, so are ESD sensitive, the LED is probably sensitive too. The optoisolator claims to be able to endure HBM 8000V/MM2000V ESD, to I guess it's ok with ESD across it's diode.

My goal: I want to make inputs, that are reasonably rugged against outside conditions and will pass future certification testing (EMC tests, I guess). For all the rest on the device, EMC protection is not that big of a mystery - I use filters, TVSs, and maybe MOV on 24 V power input.

As you can see, I added a DZ1 TVS diode, to protect resistors and the LED from ESD/surge/transients. But the thing is: the protection is not ground referenced, therefore it's hard for me to predict, how exactly will it function.

Could you help me figure out how to do it properly?

\$\endgroup\$

2 Answers 2

1
\$\begingroup\$

Generally looks good from electrical protection perspective, but the thing that is potentially missing for EMC compliance are common mode chokes on the input wires. Ideally they should be placed right against the external connector, or even potentially integrated with the connector (if you use something like an RJ45).

Whether they are needed depends on the rest of your circuitry, but if you need to use a screened enclosure to get compliance, then without common mode chokes, you may get pickup and reradiation of noise along the input lines.

\$\endgroup\$
5
  • \$\begingroup\$ I'm not too worried about common mode here; the isolator capacitance is quite small, so there is little risk of CM-DM mode conversion due to it, for example. Likewise, there's very little a CMC will do, without also adding capacitance to GND / across the isolation barrier. These are, of course, the key words one should evaluate when determining suitability for any given level of immunity; above some point, eventually filtering and shielding will become necessary. \$\endgroup\$ Sep 29, 2023 at 15:40
  • \$\begingroup\$ I'm thinking less of capacitive coupling, more of radiated EMI from the rest of the circuit, being picked up by the input wiring. Depends on how noisy the rest of your circuit is, and how tight/short you keep the input traces. Perhaps use chip LED (with light pipes) and place the resistors and isolator right next to the connector. You could even move the LEDs to the output side to allow you to really limit the amount of exposed input wiring. \$\endgroup\$
    – colintd
    Sep 29, 2023 at 15:48
  • \$\begingroup\$ Ah, for emissions? There's always layout details of course; one could artificially increase CM capacitance by routing the traces over ground plane for example, and perhaps that plane is noisy as you suggest. Ersatz couplers, or fiber optics, can indeed be used for arbitrary isolation voltages of course; the most extreme case probably being the megavolts used in HVDC link converters, with solar panels for power and fiber optics for pulse coupling; or HV physics apparatus. \$\endgroup\$ Sep 29, 2023 at 15:51
  • \$\begingroup\$ Agree that straight isolation is good, and that fiber can be used to really push isolation. I've used fiber in the past for coupling a photo-multiplier detector (low signal levels) to a hydrogen thyratron (high voltage and massive current pulses). I just speak from past painful experience, that if you are looking from strict EMC compliance, and have noisy circuitry inside a shielded box, galvanically isolated but unfiltered wires can still cause you issues due to reradiation (optically isolated RS232 lines in my case)... \$\endgroup\$
    – colintd
    Sep 29, 2023 at 16:03
  • 1
    \$\begingroup\$ Thank you for your input. For now, I don't think these inputs will cause problems. I deleted ground planes below and around these connectors, the opto is quite close to the connectors (around 2 cm), but I may make those connections even shorter. I am more worried about power input, but I used a PI filter there, so should be ok. And if I someday use shielded eclosure, I will use chokes on inputs too. \$\endgroup\$
    – Maks
    Sep 30, 2023 at 18:05
1
\$\begingroup\$

the protection is not ground referenced, therefore it's hard for me to predict, how exactly will it function. Could you help me figure out how to do it properly?

If your input connector that feeds the opto LEDs is for use by someone else, you should assume that they do connect it to ground/earth. Now the problem is resolved i.e. the worst case scenario for ESD is to equipment and electronics that is ground/earth referenced. Sure, floating equipment can be vulnerable via parasitic capacitance to ground but, the worst case scenario is when it's galvanically tied to ground/earth.

\$\endgroup\$
7
  • \$\begingroup\$ You mean that earth referenced electronics get hit the hardest by the ESD? I want my device to be earth referenced, because of the communication ports (Ethernet and RS485). "you should assume that they do connect it to ground/earth" - is it a wishfull thinking, or a legit assumption, than an engineer should make? Asuming that user connects J1 pin 2 to earth potential, but he the connection is established by a long cable, does this help at all? \$\endgroup\$
    – Maks
    Sep 29, 2023 at 15:37
  • \$\begingroup\$ (1) Yes, that's what I said (2) if you are not in control of that port then you can't insist it is earth referenced (3) Not wishful thinking but a statement of the worst case scenario regarding ESD (4) you can't and shouldn't assume a long cable <-- ESD isn't much energy for a TVS to deal with @Maks \$\endgroup\$
    – Andy aka
    Sep 29, 2023 at 16:09
  • \$\begingroup\$ Ok, thanks Andy, after few reads, I start to get what you mean. I can be a bit slow to understand, as this matter is quite complicated and I'm polish, so not at all a native speaker. So grounded terminal is the worst case scenario for ESD going across terminal 1 and 2, so that's athe only one I have to consider. I dindn't understand at first,partly because I also thought about ESD going across the photocoupler, to the microcontroller side. I assume I don't have to think about it? Could you elaborate a bit about the cable lenght? \$\endgroup\$
    – Maks
    Sep 29, 2023 at 17:46
  • \$\begingroup\$ @Maks I don't know anything about the cable other than zero length being the most onerous for ESD. Please link the data sheet for the opto. \$\endgroup\$
    – Andy aka
    Sep 29, 2023 at 18:11
  • \$\begingroup\$ OK I took a look and it mentions the HBM rating at 8 kV but doesn't not explicitly say how this was performed so I would be wary about using that type. Also, if you do use it, you may want to put a cut-out slot in the PCB underneath the opto to add extra creepage distance. \$\endgroup\$
    – Andy aka
    Sep 29, 2023 at 18:31

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.