Will external clock signal overshoot affect micro operation?

[Circlotron]

My board has a 4MHz oscillator module that feeds into a 74AC04 inverter that then feeds the inputs of four additional inverters in the same package. The four outputs then go off to various places on the board. The clock pcb tracks are not terminated at either end with a resistor. The track I have measured the far-end signal is about 70mm long. The board had a poured ground plane on both sides. I used a spring thing on the probe for a very short ground connection ~10mm to the ground plane.

As you can see, the nominally 5V signal overshoots about 1.5V and undershoots about 1V. Not very pretty. Would this kind of clock signal cause a micro to misbehave sometimes? I'm trying to track down a fault that only happens sometimes on only a very few boards. Micro is NXP 9S08PA16.

[RoGeorge]

Could be an overcompensated probe, too, was the probe compensation adjusted recently?

Cut the trace near the clock pin (at the signal source side) and insert a series resistor of 20...200 ohms.  That should reduce or eliminate the ringing.  Then observe on a known bad board if the error still happens when the ringing is not there any more.

Also, make sure you have good decoupling 100nF or so, close to the supply pins (near each chip if possible, or as often as possible).  For debugging purposes, try to solder a 10-100nF decoupling cap straight to the power pins of each IC.

[AndyC_772]

Overshoot beyond the supply rails might cause damage - check the absolute max rating for your microcontroller in the data sheet. Usually it's something like VCC+0.7V on the positive end, and VSS-0.7V on the negative.

https://cawteengineering.com/design-focus-high-speed-digital-design-and-termination/

[RoGeorge]

Another thing, it starts ringing before an edge is about to happen, that's usually caused by sinx/x digital artifacts.  Ringing can not happen before an edge, but only after an edge (causality).

Disable sin(x)/x from the oscilloscope and se if the ringings are still displayed (my Rigol can disable sin(x)/x approximation from the 'Acquire' menu), switching the display from vector mode to dots mode might help too (to see how much of the trace is fake reconstruction by the sinx/x algorithm).

[JPortici]

Overshoot beyond the supply rails might cause damage - check the absolute max rating for your microcontroller in the data sheet. Usually it's something like VCC+0.7V on the positive end, and VSS-0.7V on the negative.

https://cawteengineering.com/design-focus-high-speed-digital-design-and-termination/

But then we should see the clamping caused by the ESD diodes. I see no clamping, so it could still be an artifact

[David Hess]

There is nothing to worry about here.  There is a good chance that what you are seeing is an artifact of your oscilloscope probe and how you are using it.  The suggestions to add a little bit of series termination and decoupling at the driver is a good one if you are concerned.

Another thing, it starts ringing before an edge is about to happen, that's usually caused by sinx/x digital artifacts.  Ringing can not happen before an edge, but only after an edge (causality).

High order filters which lack linear phase delay can cause ringing before the edge.  An obscure example of this is any oscilloscope delay line; the reason it is not normally visible is because a phase compensation network is included to remove it.  It does show up however on fast oscilloscopes like the Tektronix 2465.  That is not going to apply here however.

[jonpaul]

Yes overshoot or undershoot can disturb ICs, CPUs, MCUs.

Your Probing is in question.

Suggest   NO probe, make a Zo probe into 50 Ohms.

See Tektronix Circuit Concerpts Oscilloscope Probe Circuits.

https://w140.com/tekwiki/images/6/62/062-1146-00.pdf

DUT>>450 or 4950 R>>RG174/U coax>>BNC>>Scope 50 Ohm.

Commectins at end of R, coax very short, < 1 cm

Should see tru signal if scope is any good


Jon

[Kleinstein]

The overshoot before the actual transition may indicate a general supply/ground problem, e.g. already 1 inverter earlier in the chain.
Ground pour often make a poor ground.
As a test one could probe the ground at different places.

[2N3055]

That short ground clip cable on probe rings like crazy... If you have use probe tip spring for ground and like Kleinstein said, probe around for best return path place..

[Kleinstein]

It is not so much about finding a good place for the return path, more about checking the qualtiy of the ground in general. So one fixed ground lead and probing different points on the ground to get an idea on how much ground bounce is present. With only the unwanted ripple it is less critical to have a small loop for this test. A clean signal stays clean even with a long ground lead.  If ground at different places is significant different there is more of a problem than just the clock overshoot.

Strong overshoot would cause current through chip internal diodes and this can impede the function of the µC, but also other chips. With the µC this would normally be first (rel. low curent) effects on analog functions / additional leakage and worst case a latch up. A latch up of any chip would likely effect the supply in general. So it could be a different chip that is more susceptible to the ringing.

Strong ringing  often also causes EMI problems.

[2N3055]

It is not so much about finding a good place for the return path, more about checking the qualtiy of the ground in general. So one fixed ground lead and probing different points on the ground to get an idea on how much ground bounce is present. With only the unwanted ripple it is less critical to have a small loop for this test. A clean signal stays clean even with a long ground lead.  If ground at different places is significant different there is more of a problem than just the clock overshoot.

Strong overshoot would cause current through chip internal diodes and this can impede the function of the µC, but also other chips. With the µC this would normally be first (rel. low curent) effects on analog functions / additional leakage and worst case a latch up. A latch up of any chip would likely effect the supply in general. So it could be a different chip that is more susceptible to the ringing.

Strong ringing  often also causes EMI problems.

For overshoots that will IC see, only reference is the actual -Vdd pins(s) of the IC itself.

And with fast edge and long ground lead scope probe will definitely ring like crazy. Try it. It is a resonant circuit.
It is damped, low Q, for sure, with resistors and dissipative cable but will ring.
And there is another problem, inductance of the lead will induce high frequency peaking too.

Attached images :
1. probe with grounding wire and crocodile clip
2. probe with coaxial adapter