Eliminating DC Resistively Coupled Noise: A Signal and Power Integrity Tutorial

Figure 8. The measured voltage noise on the victim
trace, on the other side of the ground plane gap,
showing no resistively coupled cross talk on the
order of 10 uV, the noise floor of the measurement.

The following is excerpted from Professor Eric Bogatin's article in the Signal Integrity Journal, The Case for Split Ground Planes. Reprinted by permission of Signal Integrity Journal.

This section continues from the discussion on Inductively Coupled Noise and Resistively Coupled Noise.

. . .

When we cut a gap in the return plane, there will be no DC current flow across the gap. There will be magnetic field coupling across the gap which is why we still see significant mutual inductance coupling between the aggressor and victim across the gap. The gap has only a small impact on this noise.

However, we would expect there would be no resistively coupled noise on the victim trace on the other side of the ground plane gap. In Figure 8, the resistively coupled noise is measured with the same scale and averaging as the noise on the victim line with no gap. The noise floor of this measurement is about 10 uV. To this level, there is no measurable resistively coupled noise, a significant reduction. 

Return Current at Low Frequency: A Signal and Power Integrity Tutorial

Figure 3. Specially configured coax cable with
the front and back of the shield shorted together.

The following is excerpted from Professor Eric Bogatin's article in the Signal Integrity Journal, The Case for Split Ground Planes. Reprinted by permission of Signal Integrity Journal.

This section continues the discussion in Signal Return Paths of the equation,

Z = R + j𝛚L

Where:

Z is the loop impedance of the current loop path, 

R is the series resistance of the loop and 

L is the loop inductance of the path.

. . .

At low frequency, when the loop impedance is dominated by the R term, the current distribution in the return plane is NOT driven by the loop impedance, it is driven by the loop resistance. In the signal path, the current will spread out uniformly as any filament path in the signal conductor will have roughly the same resistance.

But the current filaments in the return path with the lowest R will be those which are shortest. This means that return currents will take the shortest paths, independent of the signal paths. As frequency increases, the return current will redistribute to transition from the path of lowest R to the path of lowest L. 

Signal Return Paths: A Signal and Power Integrity Tutorial

Figure 1. Current distribution in the signal and
return conductors at three different frequencies.
The current redistribution at higher frequency
is driven by the currents taking filament paths
with the lowest loop inductance. 

The following is excerpted from Professor Eric Bogatin's article in the Signal Integrity Journal, The Case for Split Ground Planes. Reprinted by permission of Signal Integrity Journal.

. . .

Why Continuous Return Path Planes

The first step in engineering interconnects to reduce noise is to provide a continuous, low impedance return path to control the impedance, which controls reflection noise, and reduce the cross talk between signals that also share the same return conductor. 

A wide, continuous ground plane adjacent to a signal trace will be the lowest cross talk configuration. Anything other than a wide plane means more cross talk between signal paths sharing this return conductor. This means, never add a split or gap in the return path. You would run the risk of a signal trace inadvertently crossing this discontinuity.

If a signal crosses over a split ground plane, there are two effects which compound each other. Crossing a split creates a higher impedance path for return currents that must cross the split and forces return currents from multiple signals to overlap through the same, higher impedance, common path. 

Your Ground Bounce Questions Answered

Figure 1. Line set to "quiet low" shows ground
bounce occurring as I/O driver switches.

During an October 2020 webinar, Don’t Let Ground Bounce RuinYour Day, Dr. Eric Bogatin was asked several questions regarding his topic of presentation. Here are his answers.

Q: From what frequency should we consider ground bounce to be a problem?

A: Ground bounce is really due to a dI/dt. Generally, it becomes a problem with rise times shorter than 100 ns. The bandwidth of this is about 3.5 MHz. This means ground bounce can be an issue at relatively low frequency.

How to Connect the Returns to an Unshielded Twisted Pair

Figure 1. Three options for connecting a TDR to a UTP.

During a recent webinar, Dr. Eric Bogatin was asked several questions about how to measure unshielded twisted pairs (UTPs), which are differential pairs with no return path. Here is his answer to one of the questions. For more information, check out the whole webinar on Differential Pairs with No Return Paths.

Q: For the UTP measurement, do you need to connect the SMA connector shield grounds together?

A: Yes. It’s very important to make sure the grounds are connected at the cable connectors.

There are really three ways the connections can be made from a differential TDR to the UTP cable. These three options are shown clockwise from left to right in Figure 1.

How Equalization Works in 10x Passive Probes

Figure 1: The adjustable equalization circuit on the oscilloscope
end of the coaxial cable compensates for the 10x passive
probe's inherent low-pass filter characteristics

We've been discussing 10x passive probes and their inner workings; our last post covered all the ways in which a 10x passive probe is apt to be a liability. They'd be basically unusable for any measurements at all but for one attribute: their equalization circuit (Figure 1). Without it, the 10x passive probe makes a pretty good low-pass filter, but the equalization circuit counters that with a high-pass filter to balance things out.

The Causes of Ground Bounce and How To Avoid It

Figure 1: This cross section of a 100-MHz microstrip
transmission line shows us how a return path should look

We've been discussing the topic of ground bounce on digital I/O lines as well as an effective way to diagnose and analyze it. We've also run through a detailed example of how to measure it using a quiet-low I/O driver as a sense line. Now, let's take a step back and examine the root causes of ground bounce, and also discuss some best design practices for avoiding it altogether.

Transmission Lines (Part IV): More Essential Principles

Figure 1: The return current in a transmission line
is as important as the signal current

In our continuing survey of the topic of transmission lines, we'd begun in our last post to cover some essential principles that govern their behavior. We're not quite done with those, so in this post we'll discuss more principles that should be part of the foundation of how you think about interconnects.

Transmission Lines (Part III): Essential Principles

Figure 1: All interconnects are transmission lines with a signal
path and a return path (not ground)

Now that we've covered some of the principles and assumptions that underlie transmission lines in two prior posts, we can now directly address the topic with some essential principles that you need to understand.