0
\$\begingroup\$

I am trying to design a LC lowpass π-type filter for power filtering. The power being filtered in question is from a SMPS(LRS-75-24), which has large amounts of HF noise which cannot be filtered out by the LDO(a LM1084). However, several sources state a ferrite bead should be used in place of the inductor L due to its increasing resistance at higher frequencies.

How do I calculate the cutoff frequency \$F_{c}\$ given that there is no inductance listed as a property of ferrite beads? Inductance is needed to calculate

$$ F_{c}=\frac{1}{2\pi\sqrt{LC}} $$

where L and C are the inductance and capacitance of the elements in a π filter.

Additionally, what is the frequency range of the HF noise from the SMPS?

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
3
  • \$\begingroup\$ "what is the frequency range of the HF noise from the SMPS?" -- that's an excellent question; what have you measured? (There is no "standard noise source", SMPS are far too diverse.) Can you measure it? -- if you don't have an EMC measurement setup, how else could you measure it: oscilloscope, radio receiver, etc.? Even the target circuit -- do you know you need a filter at all, what effect is noise having on your circuit? What about common mode noise, have you verified that a DM filter will solve the concern? \$\endgroup\$ Commented Mar 3 at 9:05
  • \$\begingroup\$ The switching noise is usually in the 100 MHz ballpark and up. I.e. at least an order of magnitude higher than ripple harmonics. And it is both in common and diff. mode. If you filter only diff. mode, you can get common-to-differential mode conversion after the filter again. I.e. you should filter switch noise both in common-mode and in differential-mode. \$\endgroup\$
    – tobalt
    Commented Mar 3 at 9:56
  • \$\begingroup\$ @TimWilliams I do have an oscilloscope but not the SMPS, shipping costs will go through the roof if i order them separately. \$\endgroup\$
    – lemon
    Commented Mar 3 at 20:15

1 Answer 1

0
\$\begingroup\$

If you select your ferrite bead from a top tier supplier (Murata, Taiyo Yuden, or TDK, for example), you should be able to get a SPICE model for the specific device.

The behavior is more complicated than a simple inductance, so you'll need to just simulate the filter design (also including relevant capacitor parasitics, and source and load impedance) to get the response at the frequencies that are important to you.

what is the frequency range of the HF noise from the SMPS?

It depends on unpredicatable factors like layout parasitics, but can easily go into the gigahertz.

\$\endgroup\$
3
  • \$\begingroup\$ I don't think any of them provide full nonlinear models though, or at least without licensing or encryption. I've created a few, but nothing that would really be applicable to power filtering (most saturate at ~100mA ballpark). \$\endgroup\$ Commented Mar 3 at 20:23
  • \$\begingroup\$ @TimWilliams, and even without saturation, the models are only for typical performance, not worst case. But they're the best models you're going to get without having a fancy measurement setup to build your own from. (Does Modelithics do ferrite beads?) \$\endgroup\$
    – The Photon
    Commented Mar 3 at 22:43
  • \$\begingroup\$ Well, the models I made are fitted to datasheet curves -- when provided, which is rare enough, sadly. Measurement isn't particularly tricky, but if one lacks a network analyzer (preferably VNA, but SNA will also do) or other tools (sig gen + scope + bias tee?) in the relevant range, confirmation would indeed be difficult. On the upside: since we're talking power filtering, the cutoff can be quite low (MHz tops), where signal generators and scopes are likely cheap, plentiful, and sufficient. \$\endgroup\$ Commented Mar 3 at 23:25

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.