Common mode filter question

Thread Starter

MisterNo

Joined Jun 28, 2010
26
I'm designing a schematic with isolated 12v power supply that handles analog/digital isolated signals comming from external wires to the PCB.
No I'm trying to understand what is going on if I put a common mode filter that must be referenced to the earth ground as on this image http://my.execpc.com/~endlr/a_B2.gif , that is done with C1 and C2.
As i understand all high frequency content that is common mode noise will be shunted to the earth ground, but does this mean that 12V isolated power supply line now is not not isolated or that just DC part is isolated while AC part that is mostly noise in RF range is not?
 

marshallf3

Joined Jul 26, 2010
2,358
Your drawing shows what they usually use on the AC input side of anything mains powered, it rejects noise coming from the source power and also helps prevent noise from whatever it's running from injecting its noise back into the AC mains.

They mainly work by having what's going through one of the inductor coils canceled out on the other leg.

They're of little to no use except in AC circuits.
 

Thread Starter

MisterNo

Joined Jun 28, 2010
26
Your drawing shows what they usually use on the AC input side of anything mains powered, it rejects noise coming from the source power and also helps prevent noise from whatever it's running from injecting its noise back into the AC mains.

They mainly work by having what's going through one of the inductor coils canceled out on the other leg.

They're of little to no use except in AC circuits.
Ok, but what if I have a signal lines (cables) that are coming from other board to my board. Both boards have floating power supply, but cables are in high RF environment so a lot of EMI will be induced and some of it will be common mode noise. In that case will it be beneficial to implement common mode filter? It is just that this filter has to shunt all RF frequencies to the earth ground and I can't understand how that reflects to rest of the circuitry that is not referenced to the earth ground.
 
Last edited:

marshallf3

Joined Jul 26, 2010
2,358
They will help but certainly not eliminate crosstalk. If you know what frequency(s) you're going after, tuning the Pi filters will make a huge difference.
 

Thread Starter

MisterNo

Joined Jun 28, 2010
26
Yes, I didn't say that I'll use Pi filters as well. It is just that I had troubles understanding how common mode filter works on isolated(from ground) signal and power lines
 

Thread Starter

MisterNo

Joined Jun 28, 2010
26
Yes, google has a tons of links but I didn't find an answer I'm searching for. I had visited before 2 of 3 pages that you sent me, thank you! In all of them there is a lack of good mathematical/physical explanations and coilcraft documents is not beyond simple school examples.
For example in the coilcraft document,common mode filters are referenced to the ground. There is no word about how common mode choke works on isolated lines without caps or resistors to the ground,will it work or how good will it be? I guess it must, since there must be stray capacitances between isolated lines and the ground, but no detailed study, beside simple descriptive theories elsewhere.

In the third link they are talking all about ferrite beads, and along entire document they didn't say that they work since one of the power lines is grounded and that forms single pole LP filter to the ground.

I'm not looking about descriptive theories. I would like to see how for common mode choke physically/mathematically works on isolates power lines without any resistance/capacitances towards ground or if not then why.
I agree that there is tons of the links and that is the problem since they don't tell much so its hard to find the exact answers. I don't know what happen to all great EMI books from decades ago, filled with math and physics.
 

Kermit2

Joined Feb 5, 2010
4,162
Differential signals are commonly used in environments with high induced noise on the signal lines.

The signals are amplified by a differential amplifier(opamp) and since any signal COMMON to both lines is ignored, the noise problem is eliminated.

Another method is to use shielded cables,but you need a ground reference for that to work.
 

Thread Starter

MisterNo

Joined Jun 28, 2010
26
Thanks,but I haven't asked that.
My question was, single ended isolated from ground signals in RF environments and common mode filtering implementation does it need ground path to work?
 
Last edited:

Thread Starter

MisterNo

Joined Jun 28, 2010
26
Then how would you explain it?
Common mode noise currents flow in the same direction along the wires. In common mode choke, wires are wrapped on the same ferrite core so that both produce the magnetic flux that sum up not cancelling. Magnetic flux changes will induce loop currents in ferrite material that is loose and energy will be dissipated. In order for this to work there must be return path for the common mode current that is usually capacitive coupling to the ground.
 

marshallf3

Joined Jul 26, 2010
2,358
The ones I'm used to are wired/wound such that if a large fast input pulse comes in on the "hot" wire an equal and opposing pulse will appear in the return path to the neutral.
 

Adjuster

Joined Dec 26, 2010
2,148
The principle is that a real-world circuit will present some common-mode impedance to ground at its input, whether by parasitic capacitances or otherwise. This impedance will of course vary according to the circumstances. If the common-mode impedance were truly infinite, a common-mode choke could produce no attenuation, but then there would also be no effect from common-mode interference in a practical system.

If the circuit being driven had no coupling whatever returning it to ground, its circuitry would simply follow the common-mode voltages without any interfering input voltage being generated. Only when there is some reference to ground can the presence of the common-mode signal be detected, and outside of a purely theoretical situation such reference requires a coupling impedance.

This effect can sometimes be observed where an isolated battery operated amplifier (etc.) operates satisfactorily until connected to another grounded device, when interference appears. The input contained a significant common-mode interfering signal, but this was not apparent until the ground connection was added.

In an unbalanced situation (e.g. coax. passed through a bead, or wrapped round a toroid).the "cold" side of the connection typically returns to the common line of the circuit being driven. This will usually have significant coupling to ground, often by deliberate connection. Intentionally isolated circuits may have little coupling to ground at low to moderate frequencies, but at sufficiently high frequencies parasitic capacitance tends to become apparent.

Evaluating the actual attenuation achieved will require knowledge of the source and terminating impedances, as well as details of the inductor itself.
 

Thread Starter

MisterNo

Joined Jun 28, 2010
26
@Adjuster, excellent!! That is what I was thinking, but was unsure since I couldn't find confirmation elsewhere and you expresed it out nicelly!
My situation is that I have 2 pcb's. One PCB has H bridge that switches +-200V/us when needed, and it is beeing controled by another board that has micro controller. Everything is isolated, but as you said in the real word both boards must be coupled to the ground by parasitic capacitances what makes the closed loop for common mode induced noise/current or in other words makes common mode noise to exist.
My situation is that switching of +-200V/us even with small parasitic capacitances ~10pf will induce huge common mode noise current as C*dV/dt that flows to another board and it is in 1-10 Mhz range. If i insert common mode filter in between the boards that MUST be referenced to the earth ground, and if one of the boards has tendeny to float away, filter will bring back its potential in respect to the earthback to where is was before by discharging parasitic capacitor that couples that board to the earth. Potential of another board in respect to the earth, that is voltage across coupling parasitic capacitor, will remain the same, so no common mode current will flow to that board since it is shunted to the earth by common mode filter.
Is this all correct to say, nothing to add?


The principle is that a real-world circuit will present some common-mode impedance to ground at its input, whether by parasitic capacitances or otherwise. This impedance will of course vary according to the circumstances. If the common-mode impedance were truly infinite, a common-mode choke could produce no attenuation, but then there would also be no effect from common-mode interference in a practical system.

If the circuit being driven had no coupling whatever returning it to ground, its circuitry would simply follow the common-mode voltages without any interfering input voltage being generated. Only when there is some reference to ground can the presence of the common-mode signal be detected, and outside of a purely theoretical situation such reference requires a coupling impedance.

This effect can sometimes be observed where an isolated battery operated amplifier (etc.) operates satisfactorily until connected to another grounded device, when interference appears. The input contained a significant common-mode interfering signal, but this was not apparent until the ground connection was added.

In an unbalanced situation (e.g. coax. passed through a bead, or wrapped round a toroid).the "cold" side of the connection typically returns to the common line of the circuit being driven. This will usually have significant coupling to ground, often by deliberate connection. Intentionally isolated circuits may have little coupling to ground at low to moderate frequencies, but at sufficiently high frequencies parasitic capacitance tends to become apparent.

Evaluating the actual attenuation achieved will require knowledge of the source and terminating impedances, as well as details of the inductor itself.
 
Last edited:

marshallf3

Joined Jul 26, 2010
2,358
Well, I was just thinking about the ones designed for a 60 Hz line voltage input. You guys are dealing with RF and I've never used one in that situation.
 
Top