3
\$\begingroup\$

I've been researching but found it difficult to find a clear answer on this. I need to add ESD protection to my circuit, in this Texas Instruments video it talks about combining a TVS diode with a FBC (ferrite bead/capacitor) filter to maximize ESD protection, which to me seems like the best approach. I am wondering about placement and values for the filter. I will be powering a 3.3V ADC and MCU from a 5V AC adapter as shown in the schematic. Would the best placement of the FBC filter be right at the start of the circuit at the output of the AC adapter? I saw mention of using them at other places in the circuit and I think I saw using multiple FBC ESD filters mentioned. To me it seems like just one at the start of the circuit should work well but that's just my unknowing guess.

I am also looking for general values for the bead and capacitor. Most beads I looked at have a resistance of about 20-200 ohms at 100MHz. What size bead and capacitor would be ideal for ESD transient suppression? Should the capacitor be smaller value, or larger which could also double as a bulk supply decoupler for the circuit? Does the capacitor type matter? I've seen many mentions of the filter used but values have not been described. Thanks very much for any help.

enter image description here

\$\endgroup\$
0
\$\begingroup\$

What you have shown is reasonable. The TVS wants to have a low-impedance path to the earth or frame ground to be most effective.

As far as the bead, choose a larger size (like 805) to handle the current. You can experiment with the value; 33 ohms or so for a feed line is a good starting point.

Add some high-frequency / smaller value caps to the input as well, either side of the bead, to form a pi filter. Choose values that won’t cause anti-resonance. 4.7uf/1uf/0.22uf is ok (spaced ~ 5x apart).

A couple of resources: Murata app note for filtering https://www.murata.com/~/media/webrenewal/support/library/catalog/products/emc/emifil/c39e.ashx

K-Sim for simulating cap resonance interaction: http://ksim.kemet.com/

\$\endgroup\$
10
  • \$\begingroup\$ Thanks for that info, I am still researching. Do you mean I should I be trying to avoid creating a filter that will resonate at a frequency that is also being used in my circuit? Like if I were using a 1MHz boost converter, I would want to avoid a filter with a resonating frequency of 1MHz? Or do you mean something else? Could you please explain how those values wouldn't cause anti-resonance? And what is meant by values "spaced ~ 5x apart"? Thanks. \$\endgroup\$ – wdbwdb1 Jul 13 '19 at 21:44
  • \$\begingroup\$ Capacitors are not ideal: they have a series inductance and resistance. These elements determine the self-resonant frequency of the cap. Because of this cap non-ideal behavior, designers will use several different cap values to improve the bypass performance range. But this needs to be done in a way that the resonances don’t interact and introduce ‘peaks’ - anti-resonances - in the overall response. Using values that are no more than 5x away from each other avoids this. Try the K-sim tool to see this effect. \$\endgroup\$ – hacktastical Jul 13 '19 at 21:53
  • \$\begingroup\$ That said, anti-resonance for normal cap values is something that happens in the 30~100MHz or more. Another thing to understand about bypassing is that noise sources have harmonics, and these things are what normally give you trouble. Your circuit is on the low power side so it will not be as much of a problem. \$\endgroup\$ – hacktastical Jul 14 '19 at 4:12
  • \$\begingroup\$ So I think I'm taking another route. I'm getting about 20mV of ripple off of the AC adapter I'm using, so I think I can kill two birds with one stone in squashing the ripple and protecting against ESD spikes by replacing the ferrite bead with a beefy inductor with capacitor to make an LC filter. I believe you were definitely right about using the pi filter for the ferrite. But for the LC, I used a 220uH inductor with a 1000uF capacitor and the ripple was smashed right down and I doubt there's any way an ESD spike would make it through that. I saw mentioned somewhere that an inductor would work \$\endgroup\$ – wdbwdb1 Jul 17 '19 at 21:55
  • \$\begingroup\$ Also the only input I'm protecting is the DC jack and I just need flat DC (there was no ripple when viewed on a scope), so the brute force attack on ESD with a beefy LC filter sounds like a good fit to me. It seems most of the specialized ESD stuff out there is designed for compatibility with USB, HDMI, Ethernet, etc, where frequency is a very important factor. I just need the DC jack protected and flat DC. What do you think? \$\endgroup\$ – wdbwdb1 Jul 17 '19 at 22:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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