Why does my circuit keep burning out Raspberry Pi pins?

Question

I have a reed switch (actually several reed switches) that came pre-installed in my home as alarm sensors. When the 2 parts are together, the circuit is closed, when the 2 parts are apart, the circuit is open. The sensors are in various parts of the home, so the leads can be varying size. They run through my attic, and they may be routed near appliances, ducting, and potentially 110V wiring. The cable itself seems to be CAT5 wire where 1 pair is used for each sensor. In some areas the sensors are actually in series to create a "zone" that would trigger if any sensor was opened.

I have these switches wired to my raspberry pi through the GPIO pins. One side is wired to a GND pin, the other side is wired to a GPIO pin. I do not have any external resistors or power sources in this circuit.

I initialize all of the GPIO pins with the code:

GPIO.setup(sensor,GPIO.IN, pull_up_down=GPIO.PUD_UP)

Then I monitor the state of the pins with the code:

current_state = GPIO.input(sensor)

The circuit works perfectly for several months, but then 1 of 2 things will happen:

1. The pin will begin to trigger false alarms, often several false alarms in succession followed by hours or days with no false alarms.

2. The pin will become stuck either high or low.

I am familiar with basic electrical engineering principles, and I cannot understand why this simple circuit seems to be destroying my raspberry pi pins. I have used (and damaged) several Raspberry Pi's and the result is always one of the 2 above.

I specifically wired them in this manner so that the circuit was connected to the ground pin and not the 3.3V pin on the Pi. My thought was that this would prevent any unintentional current draw. I've read about using external resistors, but I don't believe it is necessary in this circuit.

Any ideas what might be going wrong? I am much less worried about the false alarms and more worried about fixing the permanent damage I seem to be causing to the GPIO pins.

Answer 1

Do you use shielded or unshielded cables ? how long are the cables ? do they run in their pipes, or along AC mains or other cables ?

I bet the usual problem : long cables act as antennas, catching spurious spikes from every near RF emitter: other cablings, electrical motors, microwaves hovens, etc...

You have to protect your micro with some buffers, passive or active circuit, in front of the pins of your micro. This circuits clamps the spikes.

The inputs pins read many repeated events... because mechanical switches make many tiny clicks every time their contacts move. Avoiding these events is called debouncing, it is possible to do in hardware or software.

Look on the web, there are many solutions and suggestions about clamping and debouncing, like this http://www.kevinmfodor.com/home/My-Blog/microcontrollerinputprotectiontechniques

Answer 2

From your comments (which should be part of the question) it appears you have attached long wires to the Pi and run these in a hostile environment.

It is hardly surprising that these act as an antenna, and pick up interference, which can cause false triggering and possible damage.

There are established techniques, which engineers build into circuits to prevent this.

I have written a number of answers to similar questions e.g. Pi 3 reboots when GPIO 13 is given 3V (you can find others by searching this site for interference).

Basically you should use a low impedance circuit - a pullup of no more than 3.3kΩ but preferably 330Ω. Include a series resistor (~10% of the pullup) and a capacitor across the GPIO input.

All switches bounce - reed switches are particularly prone to this - use debounce protection.

Take measures to reduce interference e.g. twisted pair cable which should NOT be run in parallel with power cabling.

There are more extreme measures, but the above should suffice.