3
\$\begingroup\$

In my application I have interfaced ESP8266 wifi chip with ATTiny828 micro-controller. We have developed our custom module with these two chips on it. The problem we are facing is that the micro-controller hangs sometimes when switching off the module. The button is the momentary type "push to On" switch and power ON/OFF functionality is implemented with the following circuit : power ON/OFF circuit when the switch is pressed npn transistor turns ON which in turn makes PMOS conduct current and makes the entire unit ON. As soon as the micro-controller gets started it puts ON/OFF pin in permanently HIGH state (to keep the unit powered on after the switch has been released). "SENSE" pin of the micro-controller senses the state of the switch (Pin change interrupts are active on this pin). So if the switch is pressed for more than a pre-programmed span of time, then micro-controller will make ON/OFF pin zero which will turn off the module.

we have taken care of debouncing in the software by adding delay of 100 ms. here is our code

// PW pin is same as "SENSE" pin
ISR(PCINT0_vect) {
    static uint8_t tmr_pw_start = 0; 
    static uint8_t PW_prev = 1, GP2_prev = 1;

    uint8_t const PW = READ(PW);
    uint8_t const GP2 = READ(GPIO2);

    if (PW != PW_prev) {
        if (PW) {               // Start measuring button press time
            tmr_pw_start = 1;
            ovf0 = 0;
        } else if(tmr_pw_start){ 
            if (ovf0 > 60) {                            
                WRITE(On_off_pin, LOW);     // POWER OFF
                _delay_ms(100); 
            }            
            tmr_pw_start = 0;
        }
    }

    if (GP2 != GP2_prev) {
        TOGGLE(BLUE_LED);
    }
    PW_prev = PW;
    GP2_prev = GP2;
}

ISR(TIMER0_OVF_vect, ISR_NOBLOCK) {
    ovf0++;
    if(!(ovf0 % 16))    TOGGLE(GREEN_LED);
}

We keep timer 0 always running and its number of overflows are used to measure elapsed time. it takes approx. 32 ms for one overflow to occur.

Other than this UART interrupts are also active in which it just takes the data and assigns it to a global variable.

The problem is this : By pressing the switch ON/OFF button for more than 2 seconds and then releasing it should switch off the unit which it does most of the time. But in 1 out of 15 cases micro-controller hangs while switching it off. It becomes unresponsive to button press and LEDs stop blinking. Or sometimes it becomes unresponsive with GREEN led still blinking.

Can anyone guess what can be the problem ?

\$\endgroup\$
5
  • \$\begingroup\$ no idea, but change 100ms to 10ms and see if 1/15 changes probability... most momentary switches are <1ms unless really flakey \$\endgroup\$
    – D.A.S.
    Commented Oct 18, 2016 at 22:26
  • 2
    \$\begingroup\$ I can guess because I have seen this problem before with other processors. The problem you are experiencing can be caused by any number of issues. Most common is the way the voltage to the processor decays after you depower the regulator. It must be a razor sharp cutoff, or the processor will "misbehave" on the way down and set the ON/OFF signal back to ON. There are a number of other issues which I couldn't begin to guess without seeing your complete schematic. Start by checking out "snap-on-snap-off" voltage regulators (if they are still made by National Semi, or their heirs). \$\endgroup\$
    – FiddyOhm
    Commented Oct 18, 2016 at 22:29
  • 1
    \$\begingroup\$ What's the voltage of the ATTiny828 and ESP8266? Post-regulator, I mean. Is that 3.3V? Also, it reads like you are focused on the ATTiny and not the ESP8266. Have you tried to just pull the ESP8266 off of the circuit and see about the ATTiny functionality without it present? \$\endgroup\$
    – jonk
    Commented Oct 18, 2016 at 22:40
  • 2
    \$\begingroup\$ Why put the ATtiny behind your power switch at all? In power down mode, it typically needs 100nA. \$\endgroup\$
    – Janka
    Commented Oct 18, 2016 at 22:41
  • \$\begingroup\$ What is the type of ovf0? Is there a chance that it overflows if you are pressing the button too long? If so, that would cause the micro not to stop on the release, as the condition (ovf0 > 60) would be wrong. \$\endgroup\$
    – Edesign
    Commented Oct 19, 2016 at 8:34

1 Answer 1

6
\$\begingroup\$

When you hold the button down and the micro drives the ON/OFF line low, there's a low-impedance path directly from the battery, through the I/O protection diodes on the SENSE pin of the micro, to its power bus — this is a very bad thing. There's another path through your external diode and the ON/OFF pin, too.

At the very least, you should have a resistor in series with each of the I/O lines to limit the current that can flow under this condition. But the micro may still be able to keep running on this low amount of current, so you'll need to augment your design in other ways, too.

The most direct solution would probably be to put the micro on the "always powered" side of the switch, and simply put it into a low-power "sleep" mode when the system is supposed to be off. This would actually simplify the external circuitry required.


EDIT: Actually, after I thought about it a bit more, I realized that you don't turn off the transistors until after the button is released, so the problem described above is only an issue if the button bounces on release — but if it does, the system will immediately power up again, or get hung in a half-on condition.

Is it a snap-action button, or something "mushier"?

\$\endgroup\$
1
  • 1
    \$\begingroup\$ That's where I was headed. Good! \$\endgroup\$
    – jonk
    Commented Oct 18, 2016 at 22:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.