fully encapsulated arduino failing

So i made a circuit that is fully encapsulated in 2 substances in order to make it water proof. one is a conformal coating the other is simple silicone rubber. Basically the device is activated using a hall effect sensor and activates lights, after that it goes into deep sleep mode. This was designed to used in all sorts of wet environments such as salt water and chlorinated pools so we tested in weird all sorts of substances. Salt water, distilled water, chlorinated water, tap water, etc

In doing several tests we found that if placed in a weak acid(vinegar) after about 20 minutes the device fails. Units only encapsulated in the conformal coating do not fail, but ones that are further encapsulated in silicone do. additionally once removed from the solution the device will reset(presumably the watchdog timer), and then never fails again.

While in this in the failed state it appears the device is consuming a large amount of current(100mA). We are very confused why this is seems to only occur once. We are assuming that is has something to do with charge building up on the silicone and causing the processor to go into an inconsistent state.

Anyone have any ideas?

#include "PinChangeInt.h"
#include "LowPower.h"
const int delayTimeMs = 10000/8;
const int hallSensor = 5;
const int LED_ARRAY_LENGTH = 5;
const int leds[] = {6,7,8,9,10};

void lowPowerOn(){
  int ledIndex =0;
  detachPinChangeInterrupt(hallSensor);
  for(long i = 0; i < delayTimeMs*4; i++){
     pinMode(leds[ledIndex],OUTPUT);
     digitalWrite(leds[ledIndex], HIGH);
     delayMicroseconds(250);
     digitalWrite(leds[ledIndex], LOW);
     pinMode(leds[ledIndex],INPUT);
     ledIndex++; 
     if(ledIndex == LED_ARRAY_LENGTH ){
       ledIndex = 0;
     }   
  }
  attachPinChangeInterrupt(hallSensor, lowPowerOn, FALLING);
  LowPower.powerDown(SLEEP_FOREVER, ADC_OFF, BOD_OFF);  
}
void setup() {
  //set all LED pines to output mode
  for(int i = 0; i<LED_ARRAY_LENGTH;i++){
    pinMode(leds[i], OUTPUT);
  }
  //set up the Hall Effect sensor  
  pinMode(hallSensor, INPUT_PULLUP);
  //Indicate to user that the device is active
  //by testing each LED
  for(int i = 0; i<LED_ARRAY_LENGTH;i++){
    digitalWrite(leds[i], HIGH);
    delay(50);
    digitalWrite(leds[i], LOW);
    delay(50);
  }
  //Now set the pins to input mode to 
  //reduce power consumption
  for(int i = 0; i<LED_ARRAY_LENGTH;i++){
    pinMode(leds[i], INPUT);
  }
  attachPinChangeInterrupt(hallSensor, lowPowerOn, FALLING);
  LowPower.powerDown(SLEEP_FOREVER, ADC_OFF, BOD_OFF);  
}


void loop() {
      
}

circuit attached

What kind of silicone did you use?

Dow-Corning T-2 Silicone Rubber

Hi, Does the T-2 flow like cream or is it thick like the bathroom/roof sealer. Are there any components or PCB edges that protrude out of the silicon rubber.

At work we have to addextra protection to HighVoltageAC-DC boards, along with conformal, which flows into all spaces we use a filler that is in a standard sealer cartridge but runs like PVA glue. It works very well.

I will post its name tomorrow when I get to work.

Tom.... :)

Many silcone formulations are corrosive (acidic) and meant for bathrooms not electronics. You need one that doesn't generate glacial acetic acid when it cures!

https://en.wikipedia.org/wiki/RTV_silicone#Advantages_and_disadvantages

@MarkT its methlyation not acetic, we sent it away to a company for precision coating.

@TomGeorge

Does the T-2 flow like cream or is it thick like the bathroom/roof sealer.

its about the consistency of wood glue only completely transparent

Are there any components or PCB edges that protrude out of the silicon rubber.

not at all, the conformal handles that and then additionally the silicone completely covers everthing

I am guessing that the acetic acid reacts with the silicone to create something that eats away the conformal coating. Try getting an isolated piece of the conformal coating - put a dab on a piece of wax paper or something and perform 3 experiments: 1. put it in the acetic acid by itself 2. put it in the acetic acid with some of the silicone 3. put it in DI water

see what happens. I would guess that it will dissolve or craze in the acetic acid + silicone, but not the others.

@keithRB the silicone is completely inert chemically in terms of acids in general, both coatings were choosing specifically for their non reactive properties. Also, it discounts the fact that the devices work again shortly after removal from the solution and then never fail afterward.

The acetic acid would not allow charge build up since it is a conductor, so that is off the table.

Reflecting on this further it seems likely that the silicone is highly permeable to protons so any acid will migrate through it.

My other thought is that the silicone doesn’t bond at all the to conformal coating so that water
and hydrogen ions wet the interface readily, forming a groundplane with lots of capacitance to
all the parts of the circuit.

My other other thought is why not use a standard potting compound sold for circuit encapsulation?

Do you use stranded wire for connections? Stranded wire is not water proof. Dwight

@MarkT

My other thought is that the silicone doesn't bond at all the to conformal coating so that water and hydrogen ions wet the interface readily, forming a groundplane with lots of capacitance to all the parts of the circuit.

This is actually what we are thinking. It also somewhat explains why once reset it never fails again. Once the silicone is saturated with charge and the device resets it appears to be stable. Just a theory

My other other thought is why not use a standard potting compound sold for circuit encapsulation?

Our first layer of coating is a standard one(parylene) which is extremely common for coating circuits, tough as hell, and has all the properties we desire. The silicone rubber is used to smooth out the surface because the parylene thickness is 50 microns leaving lots of jagged edges and such, silicone was chosen for shock absorption, transparency, and cost.

@MarkT presuming we needed to use these materials, is there anything we could do to the circuit or the code that could mitigate the effect of the capacitance building on the circuit? We have already engaged all the internal pull ups on floating pins and connected AVCC to VCC(which we should have done anyway).

General purpose "mould making" silicone rubber such as RTV-4232-T2 has poor adhesion to almost everything except already cured silicone rubber. It is quite likely that water has wicked down between the rubber and the conformal coating.

In addition, silicone rubber is gas permeable so air including CO2 may be dissolving or evolving from the cured silicone rubber.

If you want to keep water away from the electronics then you will need to use a "high tack" self-healing silicone gel type potting compound. This will still be gas permeable.

To prove/disprove the "charge build-up" theory, you can get an ESD meter electrometer that measures the static charge on things.

It sounds to me like the acetic acid is causing the silicone to swell - it's permeability to carbon dioxide (also acidic) supports this. The swollen silicone increases the pressure on the circuitry inside, and eventually shorts something out.