The lowest impedance (inductance) signal return path for high-frequency current is in a plane directly adjacent to the signal trace. What one wants is the smallest signal current loop area possible. In order to evaluate a PCB layout/stackup and determine if this objective is achieved or not, one must first know exactly how the actual return current flows on the PCB.
Many designers/engineers, however, are confused about how and where digital logic return currents flow, and what is the source of the digital logic current. In my EMC Seminars, I receive many questions relating to this subject, typical question are:
- Is it better to have a digital logic trace adjacent to a ground plane, or adjacent to a power plane?
- What about a trace between a power and ground plane?
- Would it be better to have the trace between two ground planes, or possibly two power planes?
To answer these questions we must know two things:
- What is the source of the current?
- What is the path taken by the current in returning to the source?
First let me state that the logic gate is not the source of the current, the logic gate only acts as a switch. Only the transient (switching) current is important, and the transient current is not dependent on the existence of the load at the end of the line. The source of the current is either:
- The driver’s decoupling capacitor, or
- The parasitic signal trace to ground/power plane capacitance in parallel with the load capacitance
The return current path is a function of:
- The trace configuration, microstrip or stripline
- The adjacent plane or planes (power and/or ground), and
- The logic transition (low-to-high, or high-to-low)
There are ten possible combinations of the above.
To answer the question “what is the return current path?”, I created the following Power Point presentation. Click here to view the presentation.
From the ten examples presented in the above presentation, it can be concluded that it makes no difference whatsoever if the adjacent plane, or planes, are ground or power. In all cases the current returns directly to the source through a small loop. In none of the cases does the current have to go out of its way, or flow through a large loop, in order to return to the source.
Therefore, the answer to the original three questions is that it does not matter if the return path is the ground plane or the power plane, all configurations are equally acceptable.
We can, however, conclude that stripline is always better than microstrip since two current loops exist, and in one loop the current flow is clockwise and in the other loop the current flow is counterclockwise. Therefore, the radiation from the two loops tend to cancel each other.
With respect to the question of what is the source of the current, this is summarized in the table on slide 13 of the presentation.
Note: That the power plane discussed in the above presentation is the same voltage as that supplying the driver. The case of referencing a power plane not the same as the driver voltage is a different subject–possibly the topic of a future blog.
Reference: Ott, H. W., “Electromagnetic Compatibility Engineering,” Section 10.7, Digital Logic Current Flow, John Wiley & Sons, 2009.