Reasons for glitches observed on PCB from physical touch of floating enclosure

Question

I have a circuit on a PCB on which I’m observing “analogue glitches” when making physical contact with the enclosure.

This enclosure is from ABS and there is a circular connector on one side for supplying USB power on the enclosure connected to the internal circuit board. The ABS enclosure is otherwise “floating," isolated and not generally connected to anything (unless USB power is connected to the circular connector.)

This physical contact does not seem to generate an electrostatic discharge as one might when walking across the room and then grounding to some earthing point. The circular connector shell on the enclosure is metallic and is isolated from the PCB through the ABS material. The internal contacts of the circular connector do lead to the internal PCB through about 5cm of wiring but they are also isolated of course from the circular connector shell. No explicit effort was made to connect the connector shell to the PCB ground etc. so the shell is essentially "floating."

Touching the circular connector shell with a metallic object such as a small metal ruler or screw driver, or simply a USB cable (the metallic part, without insertion) may generate this glitch that is observed on some of the I/O lines (some of which lead to CMOS level inputs, eg D-F/F etc.) It does not seem to be of an ESD nature, but somehow is observable simply by tapping the connector shell in that the D-F/F clock is sequenced on this clock-line generating an unwanted toggle. It is not always reproducible, and there seems to be a requirement for some delay between subsequent contacts before the glitch becomes evident (almost like a charging phenomenon.) It is quite strange to say the least.

What is the nature of this induced signal “passed on” to the PCB in the enclosure? Inductive, capacitive, radiative, etc.? How does this signal makes it through and what would be done to filter it out?

There seem to already be 5V TVS diodes on the connector input pins for ESD bypass in case of an ESD event, but in this case even if the tiny ESD is hugely attenuated to the PCB ground when it occurs (even though I’m not sensing this,) there is still the possibility of a residual that may be the cause of the glitches that are in the range of about x2 logic-levels, that is a few volts (positive and / or negative) but < 12V. The oscilloscope measurement level may not be the most accurate given the ground clip inductance etc, but the observed glitch event being correlated to the physical touch definitely is confirmed on visually on the oscilloscope.

Now I have heard some suggestions that series resistances, and or RC low-pass filters on more susceptible signal lines may greatly improve or attenuate this phenomenon, but otherwise is there a more direct way to prevent this on the enclosure itself? I am not seeing how the ABS with GOhms of material resistance can even conduct the smallest of currents, and why the connector shell might make this more probable even though the inside of the connector is also isolated from the shell by at least 100s of MOhms.

Answer 1

So for those that may be interested, I looked closer at this issue with a PCB we received. It appears the glitch induced was working its way to the input of a Schmitt trigger gate with a too high a pull-up for its input impedance, giving it quite a high sensitivity to perturbations, and propagating the glitch to the D-f/f clock ... in brief, having turned the weak pull-up into a strong pull-up and including a simple RC filer on the input pin as an added precaution it looks all good and dandy, no false triggering regardless of how much I try ... nevertheless, thanks to all those with the added comments....

Answer 2

To add to your answer. You seem to think that ESD was the cause of this glitch, which I find unlikely (since ESD is highly transient and should be absorbed effectively by your TVS diodes). You're probably seeing near field effects which, as you said, are injecting enough noise to cause unwanted gate transitions on your schmitt trigger. It's easy to see near field noise in action; simply grab the tip of your oscilloscope with your hand. If you're near a source of 60hz noise (fluorescent light? grab a laptop charger?) you can see the noise go up to the order of 10s of volts.