How does an aggressor raise/drop the voltage of the victim in crosstalk?

Question

I have been trying to understand, intuitively and physically, how crosstalk works. If I have a net that is switching (from either LO to HI or from HI to LO) running adjacent to a static line (LO or HI), I understand that there will be some sort of capacitive coupling.

What I am trying to figure out is how the aggressor line (switching net) is able to raise or drop the voltage of the victim line as it switches. I haven’t been able to intuitively understand this and was wondering if someone could help explain how this is happening (in terms of charges, voltages and physical effects).

Thank you in advance!

Answer 1

Here's a simplified schematic. Drivers don't have zero impedance, and even if they did the circuit board traces act like transmission lines and so would allow for a transient.

The noise source and the receiving lines are connected by a parasitic capacitance (or, sometimes, by parasitic inductance). When the source switches rapidly, C1 injects some current into the receiving line -- this works against the various impedances between that line and its drive, and make it bounce.

^{simulate this circuit – Schematic created using CircuitLab}

Answer 2

I understand that there will be some sort of capacitive coupling.

Yes, but capacitive coupling has implications that are even more essential than the name. Capacitive coupling implies that AC current will flow from the aggressor to the victim - via the parasitic capacitance.

And, whenever conductors are magnetically coupled - think parallel traces nearby on the board, or current loops coupling to each other - there will also be AC current induced in the victim, even if the coupling capacitance was zero.

Transformers work just fine with an electrostatic shield between the windings, after all, where the primary-to-secondary capacitance is approximately zero.

Now, if the victim had zero source impedance, then the DC current would not change the voltage. But no such zero-impedance circuits exist unless you're dealing with superconductors. As soon as the victim's impedance is non-zero, any current - whether DC or AC - will cause a voltage drop in the victim.

Furthermore, even if the victim was a perfect conductor at DC, it will not be a perfect conductor at AC, since at AC it's the mere geometry of the conductor that implies non-zero inductance. So, in all cases, AC currents will cause voltage drops across inductances in any conductor.

So, the physical explanation to all this is: practical circuits have non-zero coupling between all nodes, whether capacitive, magnetic, or both, and they have non-zero impedances. Thus currents flow from any node to any other node - we can name them aggressor and victim, but the nodes of course don't care about any of it. What matters, then, is limiting the magnitude of this coupling to keep the circuit functional.

Answer 3

^{simulate this circuit – Schematic created using CircuitLab}

The crosstalk cannot be properly explained without the complete network. The above circuit shows the complete (simplified) network diagram of all the elements involved.

To analyze, employ Superposition. To do this the effect of each source is analyzed separately with the other source set to zero as shown below.

^{simulate this circuit}

To further simplify to clarify the question combine several elements into single impedance elements and redraw with the intention of revealing V2x due to V1.

^{simulate this circuit}

So you can see that the crosstalk from NETA to NETB via Cc is a simple voltage divider. To complete the superposition set V1 to zero and apply V2. Calculate V2x due to V2 then add the previous result for the complete voltage at V2x.