[SOLVED] Where does signal's return current flow in PCB?

[ridgemao1983]

A lot of books say, high-speed signals tend to return directly underneath the signal trace on the ground plane.

But I don't understand that, when a high-speed signal travels from the driver to the load, then its return current comes back to the driver on the GND plane, how does the return current flow through the driver to form a closed current loop?

It is easy to understand that if the driver is a battery, the return current can flow through the negative pin to the positive pin of the battery.

But if the driver is an IC's pin, how to form a closed current loop? Does the return current on the GND layer "jump" to its original driver on the signal layer? If it is true, how does it happen?

Thanks.

 

[cyberrat]

Its Magic!

 

[s_cihan_tek]

Hi,

The return signal automatically flows through the GND plane directly below the signal trace, because by doing so the loop inductance of the circuit is minimized.

Currents on the PCB follow the path of least inductance. The only way for that to happen is if the return current travels directly under the signal trace on its way back to the source, as long as no discontinuities exist in the return plane.

And as for your last question, think of the driver pin of the IC as the positive terminal of a battery, and the ground pin of the IC as the negative terminal of the battery.
The return current does not "jump" to its driver.

 

[ridgemao1983]

think of the driver pin of the IC as the positive terminal of a battery, and the ground pin of the IC as the negative terminal of the battery.
The return current does not "jump" to its driver.

I don't think so. Here I drew a simple circuit schematic. The driver is a NOT gate and the loader is a NOT gate too. The current flows from the point Q to the point A on top layer, then travels to point G2 through the loader IC, then returns to point G1 on the GND layer, then "jump" to point Q through the driver IC.

I don't think any current can jump from point G1 to point Q, because a MOSFET is not a battery, the current can't flow from the lower voltage point to the higher voltage point.

Any suggestions? thanks.

 

[btbass]

The current flows from Vdd and returns to Vss.
Point Q is just a point in the circuit.

 

[the boo]

Hi everybody!

I think here's the best place where I could ask my question.
I saw everywhere that "in high-speed signals tend to return directly underneath the signal trace on the ground plane"

But what about low frequency signals? Does the return current flows on the entire ground plane? In that case is the mesured impedance between two random points of the ground plane is the same (we could take just the ground plane with no parts connected to it)?

Thank you guy!

 

[lurchman]

Guy,

The answer to both high frequency and low frequency is "it follows the path of least resistance".

The question then remains what is "resistance" to the current.

From the DC perspective you could divide any conductor into domains and work out what it' DC resistance is and then calculate the path from all the other current sources and sinks on the board.

Not difficult just very labourious and very error prone.

However when you start talking about the AC resistance or impeadence life becomes a little more interesting. The following is a bit of an arm waving explination as the reality is lots of maths

Any movment of charge (ie current) creates a magnetic field and in so doing also induces a current in opposition untill at some point a steady state is achived (even if acompanied by a loud bang . One curious aspect of this is that is also frequency dependant which results in "skin effect" where the current might only flow in the top few microns of the conductor and not at all inside it.

It is however not just the magnetic field to consider there is also the electrical field as well, and just to be anoying unlike the magnetic field that occurs only with the passage of charge, the electrical field can and does quite happily exisit without the movment of charge.

Now interestingly every time a charge moves current flows and a magnetic field is created, however the movment of charge also creates a "hole" behind it in the electrical field into which other charges will flow, (none of which happens instantaniously).

However the intensity of the fields are dependant not just on other charges or the currents that arise from their movment but the physical properties of components around the conductor in which the charge moves...

Now there is another issue things like charge only move at the speed of light as a maximum. Which means that when a charge moves it is moving based on out of date information because other charges in other places are moving as a consiquence.

One consiquence is of this is as current moves down a signal trace it effectivly leaves a path or channel behind it that any other adjacent charge will get draged into to balance the change in the fields hence a current is induced under the signal trace.

Thus at all frequencies comparable to the board size you get some interesting effects in that a charge has moved in one direction due to the influance of an "old" movment of charge at some point inbetween they meet and effectivly pass over each other like waves in a pond, and under certain conditions stability is achived and you get "standing waves".

All of which tends to make answering the question a little difficult as anything more than "the path of least resistance

 

[marce]

DC FOLLOWS THE PATH OF LEAST RESISTANCE, AS THE SIGNAL FREQUENCY INCREASES IT START TO FOLLOW THE PATH OF LEAST INDUCTANCE. SO BY approx 1MHz NEARLY ALL THE CURRENT IS FLOWING UNDER THE SIGNAL. (See first link below)
Signals travel as TEM waves (generaly), these travel down the wave guide (track) at some fraction of the speed of light (NEVER AT THE SPEED OF LIGHT ON A PCB) dependant on the dielectric and the track structure (microstrip or stripline). Holes are created when electrons move from one atom's valance shell to another, electrons travel at about 2.5mm a second, not very fast.
I would ignore skin effect and AC resistance in this context at the moment, they will only confuse issues, AC resistance is not often encounted unless you do high power, high speed.
I would look round the sites at the bottom, they will provide a lot of information from the acknowledged experts in this field.
And again I would like to iterate that the path of least resistance will only apply at DC to a few hertz.

How signals travel, with pretty pictures:
http://www.x2y.com/filters/TechDay0...log_Designs_Demand_GoodPCBLayouts _JohnWu.pdf

Propagation speed of signals:
http://www.ultracad.com/mentor/microstrip propagation.pdf
Transmission Line Rules of Thumb

Electron Speed:
**broken link removed**

SITES for FURTHER EDUCATION:
Signal Consulting, Inc. - Dr. Howard Johnson
beTheSignal.com
Welcome
Speeding Edge consultants specialize in high-speed PCB and system design disciplines
Tech Tips

 

[lurchman]

Marce,

You have raised an issue with my comment that the signal "follows the path of least resistance" I put the expression in double quotes to indicate that it was being used in the common human sense, not the specific electrical sense. I further confirmed this by asking the quaetion what "resistance" ment before going into very generalised (hence the comment "arm waving") explanation.

With regards this it also needs to be clarified that even what some consider as "DC" is effected by "AC" effects when some other physical or electrical change in the general locality occures. A simple example to be considered is an H configuratin of resistors or a Whestone Bridge with thw top open and having two DC supplies. If you adjust the arms to ballance tha galvo then (in theory) all the current flows into one arm flows down that arm of the H to the common return of the battery. However if one of the battery voltages decreases then the galvo will go out of balance and current will flow in from the other arm of the H even though it's battery voltage has not changed, nore have the values of resistance around the circuit. So the "DC" arguments should realy be considered only as "steady state" arguments ("dear ***" pendantry can be long winded

With regards the skin effect it is actually fairly easily measurable at around 60KHz where it's about 0.25mm and shrinks rapidly so that by the time you get to the Long Waves (200KHz) it is distinctly noticable as is another problem in wound components such as inductors of the "Proximity Effect". So much so that early radio engineers developed a special wire that consisted of multiple fine wires wwrapped in a fine cotton insulator and these wires were interwoven in a special way to overcome the skin effect. You can see this "Litz wire" not just in the inductors in old "long wave" valve radios but also in some modern home appliances such as induction hobs for cooking, some HF lighting systems and likewise in some induction battery charging systems used in sealed objects such as electric tooth brushes and yes Litz wire still popes up in some radio circuits (have a look at the coil around some ferrite rod antennas) but not that often as there are easier ways of getting the required inductance with considerably less turns of wire.

As for the tardy nature of electrons yes as individuals they move very slowly in human terms, however so do billiard balls in a tube (a well respected physics analagy). But as it's quick to see if you push a billiard ball into the full tube one appears imediatly at the other end. It is the same with electrons in a wire however there is a limit on the immediacy and that is the speed of light which is the maximum speed anything can go at in the normal physics models you get taught at school etc.

And this is the problem as was once pointed out to me the teaching of physics involves telling a succession of lies each a little bit more accurate than the last lie but never getting to the truth. That is we do not understand why the electrons move as they do but we have some very fine mathmatical models that match what the electrons do behave on mass. Sometimes we find our models have holes in (think electrons magicaly moving through insulators due to quantum effects) and have to come up with better models that give us great movment in understanding but often in technology as well (ie many if not most semiconductors actually work by quantum effects not clasical physics effects).

The trick is to explain things in ways most people can get an understanding of based on their current understanding which is why I made the comment about armwaving and mathmatics. I could have just given links to the papers but how many of the people reading this thread are going to go through them sufficiently to understand them in a meaningful way.

 

[marce]

I agree with your views but I think the simpler explanations with pretty pictures are more acceptable for PCB designers, also for High Speed design and related EMC and signal integrity it is benefitial for us to think of the return current following under the route. Skin effect can be easily worked out by using the Saturn PCB toolkit (its easier than working out the maths, it does it for you). Again it is somthing that will not often concern PCB designers that often in general design work, and again I wouldn't like to complicate matters by looking at AC resistance where it can vay with frequency due to skin effect, as it adds further complications for people getting to grips with the basic things high speed design brings to designers.
As to understanding the flow of signals, it is becoming a necessity as frequency of signals increases, and in my view alwyas has been.Have Fun
Marc

 

[thega]

Very useful link, thanks marce