Electronics > Projects, Designs, and Technical Stuff

Approach to design a multi-stage RLC low-pass filter?

(1/12) > >>

kreyszig:
Hi,

So if I want to decide on a multi-stage RLC low-pass filter for the input of an SMPS (to filter its input, but also to prevent noise from the SMPS to back feed into the source), what is a good approach? I have a hard time to wrap my head around this. Some assumptions can be made about the voltage of the source at rest, the inductance of the wires to the SMPS, the frequency of the SMPS, the current drawn from it (hence its average resistance). Where do we go from there?

For a single-stage low-pass filter I guess it is more straightforward, but for I have a harder time with a multi-stage one. For example, if the impedance of a second stage is large compared to the first stage, then the second stage can be approximated as being independent from the first stage, but is it necessarily something we want to do or just a short cut to make it easier to understand? Also the fact that there is some optimisation to do for the impact of the circuit both for the SMPS and the source confuses me.

Also, if I had an expression for the source Thevenin's equivalent circuit and also for the complex impendance of the load, is there a method to optimize for an n-stage filter?

Thanks!

jonpaul:
mains EMI filters are very well documented.

Stock filters, CM, DM chokes,  X, Y caps.

What's the PSU Vin, Vout, P ?

Jon

kreyszig:

--- Quote from: jonpaul on March 06, 2023, 09:33:52 pm ---mains EMI filters are very well documented.

Stock filters, CM, DM chokes,  X, Y caps.

What's the PSU Vin, Vout, P ?

Jon

--- End quote ---

Vin is 5-15V DC (~12V more likely), Vout is 5V, Pout is 3W, using MP4576 converter (using LPD8035V-333 fly-back for dual outputs) at 300 kHz, forced PWM. Power using about 1 m of cable. I want to prevent prevent oscillations, constrain EMI, limit current draw at startup. There was some off-topic discussion about it here: https://www.eevblog.com/forum/projects/surge-protection-grounding-and-fail-safe-biasing-for-isolated-rs-485/ . What I would really like to learn is the full thought process when deciding on a number of stages. designing the filter and picking components. There seem to be so many possible configurations including using TVS, MOVs, ferrite beads, etc.

Thanks!

Benta:
If input noise is a prime consideration, use a SEPIC converter instead. Much more flexible and in your case no more complicated. At 3 W it's a perfect choice.
https://en.wikipedia.org/wiki/Single-ended_primary-inductor_converter

kreyszig:

--- Quote from: Benta on March 06, 2023, 10:25:27 pm ---If input noise is a prime consideration, use a SEPIC converter instead. Much more flexible and in your case no more complicated. At 3 W it's a perfect choice.
https://en.wikipedia.org/wiki/Single-ended_primary-inductor_converter

--- End quote ---

Thanks. I need one dual output, one isolated and one non-isolated, this is why I went with a buck converter+ fly-back ("Fly-buck") design. I am fairly confident that this is a good design for my application. Designing filtering around it (input + outputs) is more what I am looking for. There are two proposed filter configurations in the MP4576 datasheet. As discussed in the other referenced threads it looks like there are issues with the one suggested for EMI filtering.