# RF Common-Mode Filter Topology

I am working on a project with two inputs and an output as shown. IN2/OUT are external connections, and IN1 is internal. Based on input from this forum and research I have come up with the following common-mode filter for placement between IN2 and OUT. (L1&L2 are actually the same CM choke, just drawn separate.)

I've calculated and simulated this as shown:

I'm happy with both differential and common-mode response, but I've only simulated the filter in common-mode without any load, as I don't exactly know what to use for a load when simulating CM filters. So my questions:

1) Does this topology look ok with the way the circuit is supposed to function (shunting CM cable noise from IN2 and OUT out of the circuit?

2) When simulating CM filtering, what is usually used as a load? I watched a youtube video where a current source was used to simulate the filter, but I didn't fully understand the reasoning behind it.

3) This may be a brain fart question, but is there any useful significance to simulating filter frequency response without any kind of load, since the source and load will change the frequency response? I was reading a post about PI filter analysis, where differential filter response was said to be meaningless without source and load attached.

• What is your driving impedance and load spectrum? AS shown it is 0 ohms 1kHz with no load. What are your SNR requirements? THe choice of CM Choke is more than just L1, L2 but the imbalance and CMRR Nov 30, 2017 at 23:28
• In differential mode driving is below 100 Ohms and load >10k. In DM I've simulated with cables/loads/sources. In common-mode I don't know how to determine the impedance of source/destination at RF frequencies, or what load impedances to use in simulation (which is why I posted and am hoping for some guidance). I'm trying to prevent RF in the FM range especially, but also beyond, from being picked up on cables and passed through the system. It's basically an interface between balanced audio equipment Nov 30, 2017 at 23:41
• FM RF will be a small capacitance coupling. AC noise may be a weak capacitance as well , then conducted ground noise from remote connections and AC conducted noise currents different at each end. Nov 30, 2017 at 23:55

The issue with CM filters is that both the source impedance and the load impedance are somewhat unknown. All manufacturers give curves for 50:50 ohm, since that is what the VNA gives without extra fixturing.

Good manufactures also give 1:100 and 100:1 ohm curves since they are more representative of the common environment the filter will be used in.

As for the way they are tested the datasheet posted by Tony explain these quite nicely. Your spice only check for one topology.

The real test of the filter is of course in the final equipment with the chassis closed and the power wire in the capacitive clamp (or LISN or whatever your EMC test standard requires)

And, as last question, yes the load is absolutely necessary and influences the filter response.

Consider something like this.

Then when you draw impedance curves of Caps, the Ratio of impedance is your attenuation added to your Zcm and Zdm . You want DM loss to be 0dB and CM loss to be >60dB or what? Surely you have some measurements for long cables.

• Actually the choke I was planning on using is here: product.tdk.com/info/en/catalog/datasheets/…, the 220. I mentioned above that L1&L2 are actually one CM choke, hopefully that was clear. I'm not working to any spec, and I don't have any numbers so I was just trying to get rejection as low as possible in that range, not very precise, I know. Input/output is line-level audio and this setup will be used near an FM transmitter, I could go to the location where it will be used and do some testing, but not exactly sure how to determine what CM ... Dec 1, 2017 at 0:31
• rejection would be required. Dec 1, 2017 at 0:32
• One takes the longest cable used and terminates the impedance at both ends as correctly as possible into a spectrum analyzer or scope. Cable quality varies widely and ground connections matter too. I once used a spectrum analyzer on CATV signals near a train passing and measured significant interference, mostly due to poor grounds on coax. This is a CM noise converted to differential due to impedance imbalance. Dec 1, 2017 at 0:45