Ground plane - tell me more about it

[urlkrueger]

Quite often I will see a statement like "This circuit needs a good ground plane to work well" especially when high speed digital or RF signals are involved.

My understanding of a ground plane is as a 2D electrically conductive sheet at signal ground potential in near proximity to the circuit components and paths and whose size is at least as large as the circuit.

I believe I read somewhere that with PCB's the ground plane makes the signal traces look like transmission lines as opposed to plain wires which improves propagation of signals with less loss and distortion.  This makes some sense to me given my limited understanding of transmission lines.

Another function I can see for a ground plane is to provide more direct low impedance paths for return current.

Anything else?

thanks...
earl...

     

[marshallh]

Google "Maxim mixed-signal layout"

[Rerouter]

When you get your nose truly stuck into the finer points of layout you end up focusing more heavily on the signal return path than the actual signal (or atleast equal consideration)

As you use higher frequency signals a larger portion the return current will want to flow along the path of least impedance vs the path of least resistance, even with picosecond edge speeds there is still a small fraction of the current that wants to travel along the path of least resistance (not much but it exists),

this is why for digital and RF people focus on a nice big ground planes for the path of least impedance, to clarify by path of least impedance i mean that the signal will want to return as close to the path of the original signal as possible, and the better this path is followed the lower the effective impedance of the signal trace,

Any deviation from this route say a break in the ground plane and the return current has to veer off, go around the deviation and return back to the path it was following, this causes an impedance mis-match in the signal which can cause things like reflections and radiated emissions, (think stub antenna's) and breaks into ground loops a little, with decoupling caps being a golden example of this, with most people placing the positive pin close to the cap and leaving the ground plane to deal with the negative connection, when in reality the current will be trying to trace along the current path through your IC.

At the low end living mainly with path of least resistance you tend to find people running dedicated earth traces or star grounds to keep things at the same potential (generally analog) as all traces have resistance and as such all ground traces create a small voltage deviation from the supply ground, equally you may find analog segments isolated on there own little ground plane island, this is generally to keep larger currents that would cause a greater voltage deviation from crossing it,

breaking the ground planes can introduce radiated noise issues if there are any fast edges about, and just because your op amp is controlling something slow will not mean that its output can not move quickly due to a high gain, but this is diverging into slew rate limiting..

Now with most of this stuff there is no exact right and wrong, to throw a spanner in the works of most books, have a think about a high speed analog system, you will want to maintain a path of least impedance, but also keep large currents away from your circuit, which becomes more difficult as floating it on its own little ground plane and having any form of break or deviation in the path of least impedance makes a lovely little RF noise source, this is why on some designs you see little capacitors stitching ground planes together, it provides a lower impedance path for fast signals than otherwise, reducing the voltage developed over the gap and reducing the amplitude of the emitted RFI

[urlkrueger]

Marshallh - thanks for that link. Interesting info there I'd wondered about.

Rerouter -  WOW!  Reminds me of how lightening bolts retrace the same path.


Now all I have to do is figure out how to deal with all the fuzz on my scope from the 1kW AM transmitter 1900 ft away from my lab broadcasting at 1.499 MHZ.   I don't suppose the FCC would consider this harmful interference.

Thanks again...
earl...

[Dave Turner]

Faraday's cage?

[Dago]

I believe I read somewhere that with PCB's the ground plane makes the signal traces look like transmission lines as opposed to plain wires which improves propagation of signals with less loss and distortion.  This makes some sense to me given my limited understanding of transmission lines.

Another function I can see for a ground plane is to provide more direct low impedance paths for return current.

These are basically the same thing. There is no "plain wires", everything conducting current is a transmission line. Just depends on the frequency content of the signal if the transmission line effects effect anything in the circuit.

For decoupling caps the function of the "transmission line" (composed of the trace going to the IC pin and the ground layer/ground track) is to match the very low impedance of the coupling cap to the IC = the transmission line should provide as small of an impedance to couple energy efficiently to the IC.

[jahonen]

The idea behind ground plane is to provide lowest possible impedance between ground points, so voltage between ground points of various components is much more uniform. That reduces common-mode emission problems (by far the most difficult EMC emission problem type).

Another "side effect" is that most of the stored energy between PCB traces and ground is localized between ground plane and the trace (electricity is lazy!), instead of spreading randomly. This makes it radiate much less, having much less crosstalk between adjacent traces and much more tolerant to external interference. As the transverse geometry of trace and associated return (ground plane) becomes constant along the trace length, so does the impedance of the trace.

Regards,
Janne

[coppice]

Faraday's cage?

I think the correct terms are "Faraday cage" or "Faraday's coffin".

[Dave Turner]

or Faraday shield

[urlkrueger]

Ah, some more good comments. 

Funny how this all goes back to that very first lesson in electricity, "Every circuit must provide a complete path for the current to flow."  I tend to get so focused on the bits and pieces in the circuit that do interesting stuff that I forget about the lowly return conductor.

I mentioned the Faraday cage idea my wife and her only comment was that she didn't think covering the walls with wire mesh would go well with the rest of the decor.  I wonder what she'd think of aluminum foil?