I’m using the latest MSO Logic. We’re seeing unexpected signal transitions in the 46ns range. Could this be noise? Setup?
It happens across all channels:
As I zoom in it’s across all channels, but not all simultaneously:
Do I need to enable the glitch filter?
This has happened to me several times, with different Logic hardware and software versions. Every single time it came down to my own mistake: primarily making sure that there is a very solid ground connection between the logic analyzer and my circuit. Fixing the ground solved my ‘noise’ glitches every time. Maybe something similar is happening in your case.
It is best to figure out the noise cause and fix directly, if possible. As @Brian said, good probe grounding is a likely cause. Did you connect the ground wire on each probe? Are they all connected to a good ground reference on the target board? Are all the probe connections good & tight, too?
You can use the analog channels to see what the noise looks like, and check if it correlates to a specific cause. Higher current loads switching on/off? Any power transients or other load changes?
On the MSO channels, the digital thresholds are programmable, so you can also tweak the logic voltage levels to something higher/lower based on the noise pulses.
If the glitches are still too strong, but short enough – then the glitch filtering should clean up the display. However, I’d try better grounding and tweaking the logic level voltage thresholds first.
Finally, if the noise is actually coming from the hardware and not from bad probe connections or instrumentation settings, then the hardware design might need to be reviewed & updated as well.
@kurt_shetler Two more thoughts:
-
In at least two cases for me, my “super short glitch on every channel at the same time” was due to not having local grounds. I had the logic analyzer’s ground tied in to the board where the microcontroller was, but then there was a cable, with an LCD at the end of the cable, and I had the analyzer’s wires tied in to the LCD end of the cable - but with no ground at that point. (Ground is ground, right? Why would it matter?) Bringing a couple of the LA’s grounds to the end of the cable make the glitches disappear.
-
When debugging ‘strange’ things like this, I tend to turn off all digital channels, and use just a bare minimum analog channels with the sample rate maxed out. Look at what the glitches look like. Do they look like a normal digital logic pulse? Do they have the same max and min voltages as you expect digital signals to have? In one case for me, I had great ground connections to the LA, but when I looked at the glitches in the analog domain at 500MS/s, I saw that it was a very strange shape and that it’s maximum voltage barely crossed the LA’s threshold (for a digital ‘high’). This led me to figuring out that the problem was ground bounce from turning on a high current part of the project. Adding bulk caps solved the ground bounce issue and made the glitches go away.
Earlier in my career, that’s how I understood it – grounding was a simple binary switch: it was either connected or it was not. When I was learning more about LCR meters and why 4 lead/wire measurements were needed for better precision vs. two leads on simple multimeters, I understood more about how real world circuit connections really don’t behave like nice ideal models. Every wire or trace is really a transmission line, and no electrical path or circuit connection is truly ‘frictionless’ (i.e., not without some inductance, capacitance, and/or resistance characteristics). The relevance of these often ignored characteristics depends on the type of signal and how precisely you want to measure it. Simple DC voltages or slow speed single-ended ‘digital’ communication signals often tolerate ignoring them, as their effects aren’t usually significant enough to affect most basic measurements.
However, high-speed communications, differential and/or low-voltage signals can be more sensitive to these characteristics and using ‘bad’ probing techniques will likely have more noise, or could even disturb the signals being measured and affect the device/circuit under test. The probes themselves will introduce inductive, capacitive, and resistive loads into the circuit being measured and the effect of the extra loading may be consequential. That’s why high-end probes exist for high-speed (high-frequency) applications, sensitive differential measurements, and/or high-precision signal integrity analysis, where minimizing circuit loading and preserving the original signal characteristics is critical. Active probes, low-capacitance passive probes, and high-bandwidth differential probes are designed with ultra-high input impedance and very low input capacitance to mitigate these issues.
Saleae has an article about probe setup and grounding techniques that can help avoid measurement problems:
… an older post about probing techniques with analog probes and looking for maximum resolution / signal quality:
… and finally, another post about noise while probing: