PCF8574 defective ?

[rnb]

Hello,

I'm using PCF8574 since more than 10 years to sense dry contacts (PCF is used in input mode). The contacts are located between 1 and 4 meters away form the board, and its action just consist to pull the input (P0-P7) to Vss. The environment is not a a lab but it is not too much "noisy".

This has been working very well so far until last months where I start to got strange behaviors:

As long a s the contact is very close to the PCF8574 (and cable length is short < 0.2m), its still working as designed.

If the cable length increase, at some stage (not always the same distance) the PCF8574 suddenly "collapse" Vcc, like if there is a shortcut inside (and the chips gets very hot).

Trying to understand, I did several experiments and conclude that the problem concerns a specific lot number:

               TH1780.102

               KnG14283

I do not know if and where I can retrieve data from this batch...

Can someone help

Thank's

Rene

[TomE]

> "Like there is a short inside and .. gets very hot".

That looks like "CMOS Latchup". That happens if a voltage is applied to a pin that is higher than VCC or lower than VSS. Doing that turns on "Parasitic Devices" inside the chip.

You can learn all about that here:

https://en.wikipedia.org/wiki/Latch-up

Why does it happen at certain cable lengths? Probably because that length of cable is acting as an efficient radio aerial. And maybe someone has a new mobile phone, or some new WiFi or Bluetooth gear in the vicinity.

You have to add protection to inputs like this. Resistors, capacitors, chokes (inductors), zeners. Just adding a cap would probably fix your case. You can't just connect unprotected silicon to "the outside world".

I remember hearing a story/joke about a technician working on a computer system, and having it partly disassembled on the bench in front of him. Then his phone rang with the boss asking "have you fixed it yet?". He answered "It was working until you rang, but now it is dead".

Tom