Electric pulse from multimeter while testing circuits

[jerryk]

I have built a small circuit to interface with my old security system at my home.  The circuit interfaces with the GPIO pins on a raspberry pi and via python code texts my phone when a switch in a closed loop on the alarm system is opened.

It works most of the time correctly with the following exceptions:

1.  At random intervals from 12 to maybe 18 hrs apart all eight zones will trigger at the exact same time and sending me eight false alarm texts.
2.  When I try to test each zone voltage on the breadboard as soon as my negative probe touches the ground on the pi all zones will trigger in the same manner above.

I have captured the problem in #2 above on the oscilloscope on two of the zones as shown below.

Further investigation reveals - The electric pulse can be triggered anytime I touch a multimeter lead to the pi(pin9) ground.  You then have to wait for a short period of time and then this event can be repeated.  You can even touch a thin welding rod to the pi ground and the exact same thing occurs.  This leads me to think there is a difference of electric potential building in the pi and it discharges on any conductive object with enough of a difference in potential.  Each time this happens all zones are triggered and I get eight unwanted texts.  I can control this but that leaves problem #1 above.  I think #1 and #2 are in effect the same thing or related. This occasional random triggering of all zones make this project not viable.

I'm not sure if there is an issue with the grounding of my raspberry pi internally or is my circuit design begging for this problem.

I have also attached a schematic of board I made and the actual breadboard mockup.

Any help solving this mystery will be greatly appreciated.

Thanks in advance - Jerry

[danadak]

You are in a breadboard which is a tough environment to start with.

I only see one cap, looks like a tant, on supply rail of breadboard. I would
add another cap close to the pins on the UP or control logic parts. Two
caps, a tant, try 10 uF to start, and a .01 ceramic disk. Each part or adjacent
parts, say every two.

Touching pins and triggering logic sounds like ESD issues. Is the ground on the board
close to source or long lead to source = lots of L = transients.

Pins that interface to outside world should have a low Z interface to minimize C
noise coupling. Thats counter intuitive as it means more Pdiss, the input must source
needed current to meet logic levels/nosie margin, but stray hi z coupling is a issue
especially in CMOS type interfaces. Also pins should have diode protection to clamp
any signal that wants to violate Vdd >= Vin >= Vss.


Regards, Dana.

[danadak]

If you have a DSO set up trigger for two cases, one for V > Vdd + .7,
and one for Vss - .7, and see if you can get any triggers. Use one shot
mode. If you do then you have coupling of signals outside supply rails
of CMOS parts and need the diode protection on pins.


Also set it up for infinite persistence, auto trigger, and look at supply rails
for pk-pk noise. See if there are any issues there.


Regards, Dana.

[jerryk]

Thanks Dana for the reply.  Im just a bigginer and am doing my best to apply your advice to what's happening in my circuit.  I do agree that it must be ESD related and want to avoid damaging the Rasberry Pi as part of the learning curve.  I think there is static around and how it randomly finds it's way down the 3.3v supply rail is still a mystery to me.  The pi is on a very short ribbon cable so I don't think lead length is in play.

As far as the caps go, there are three.  One on the 3.3 supply rail and one on the in and out of the linear voltage regular that provides the 9v ref voltage for the LM339.  I assume you are talking about adding more caps to the 3.3v supply rail closer the input gpio pins as in one on every other pin?  But I'm not sure I fully understand what you are suggesting.

With regards to diode protection of the gpio input pins I have attached a schematic of what I think you are suggesting.

Again I appreciate the advice and am leaning as a go.

Jerry