Common mode chokes for single ended signals

[cupoftea]

Please confirm that all signals that are about to go "off board" onto a cable, are better off having common mode filtration (eg common mode choke) on the board, right by the connector?
Also, that whether a signal is differential, or single ended, has no bearing on whether or not a common mode signal filter is needed, because common mode noise will couple to a cable no matter what type of signal is going to be carried on it? (differential or single ended)

 

[scopeprobe]

You only need common mode filtration if you have a common mode noise on the signal lines. If your not generating it then theres nothing to filter so to answer your question it is completely dependant on you electrical system.

 

[cupoftea]

Thanks, i agree, but if we consider PCBs that contain comms signals on them.....and/or signals with rising or falling egdes, and with SMPS's on them, then generally they are going to be generating common mode noise, and this will result in radiation from cables going off the PCB.
As such...
Please confirm that all signals that are about to go "off board" onto a cable, are better off having common mode filtration (eg common mode choke) on the board, right by the connector?

(as you know, comms signals often have pretty fast rising and falling edges)

 

[betwixt]

Not sure what comms signals you are using but I normally add a clamp on ferrite to the cable. I would be worried that the inductance of most common mode chokes would distort the signals.
Note that for many protocols there are drivers with limited slew rate available to help reducing EMC.

For reference, I have CAT-5 cable carrying 38.4K Bauds RS422 signals running parallel to and about 3m away from an antenna for over 30m and nothing is picked up by the antenna when receiving and the RS422 works fine with 100+ Watts of RF pumped out on to the antenna.

Brian.

 

[KlausST]

Hi,

If you have*
* several signals
* one commin GND signal
--> you can't use common mode chokes

****
If you have
* a GND signal that carries other currents than the "signal's" return,
--> then a common mode choke can not work satisfactory

****
If you have
* several signals
* each with it's own GND (return),
--> then I call it "quasi differential", common mode chokes will work.

***
Mind:
If you want a GND signal to pick up noise with same magnitude (over frequency), the GND needs to have the same impedance (over frequency) than the signal.
Example: if a "signal" is generated by an Opamp with gain of 5,
--> then you may generate a "GND_ref" signal with an identical Opamp and also gain of 5. This way both signals have identical source impedance.
The same applies for the load_impedance.
In most cases you will need an independent low impedance GND signal carrying other currents, maybe shielding.

But in detail it depends on application details.

Klaus

 

[cupoftea]

If you have*
* several signals
* one commin GND signal
--> you can't use common mode chokes

****
If you have
* a GND signal that carries other currents than the "signal's" return,
--> then a common mode choke can not work satisfactory

Thanks, in this case, would it be worth it to add ferrite beads in the 'go' and 'return' of the signal?...because then there will at least be some common mode impedance

 

[KlausST]

Hi,

the common mode impedance will rise. It is a series impedance.
But whether it´s a benefit depends on the "load impedance".

Try to see it as a voltage divider.
When the series impedance is 100 Ohms and the load impedance is 10k Ohms, then the benefit is ignorable 1%.

So you have to calculate with both: source impedance and load impedance.

Klaus

 

[scopeprobe]

Thanks, in this case, would it be worth it to add ferrite beads in the 'go' and 'return' of the signal?...because then there will at least be some common mode impedance

Individual Ferrites in the line form differential filtering as opposed to common mode so won't address common mode noise. What I believe betwixt is referring to is a ferrite clamp which is clamped around all of the the signal coductors which acts as a common mode impedance without introducing significant differential inductance therefore having minimal effect on any high speed data lines.