Power on without reset



  • I'm working on a project that requires me to seal all of the electronics inside an enclosure. It then needs to wait without turning on until a later date. I wanted to accomplish this by putting a magnetic reed switch (normally closed) in line with the battery so that I could just remove a magnet to start the system. I set this all up and it seems to work well to kill power to the system, but when I try to start the system after power has been removed I'm unable to get the LoPy (Lopy 4 with an expansion board) to turn on unless I physically press the reset button.

    Is there any way around this?





  • @elizar No dice with the RC circuit. I'm still doing some investigation, but it doesn't seem like the single rising edge is enough here.



  • @paloma Don't worry. In years of silicon design we never made a chip that would look for both edges. And I don't know anyone who would build such a thing.
    But it's always a good idea - for any digital circuit - to care for a reliable reset signal.

    What happens to a microcontroller core when the the power supply bounces is that parts of the chip start working while others are still in the reset procedure. When you design a chip you normally know nothing about the later board environment. Therefore the circuitry on the chip is made in a way that it will start/operate perfectly under optimal conditions. Here that means: when the power is switched on it jumps from 0V to 5V in zero time and behaves like an ideal voltage source.

    In the real world nearly every controller has a kind of "startup circuitry" (sometimes integrated in the chip) that holds the core in the RESET state for a while until power is up and ready. PCs power supplys even have a separate signal wire for the POWER_GOOD signal.



  • @elizar My only concern is that the Lopy might be looking for both a falling edge and a rising edge to combat false resets. This RC circuit would only produce a rising edge upon plugging in the battery. The easiest solution would be to just test it out though so I'll report back after giving this a shot.



  • @jcaron I haven't been able to find another thread talking about this. If you can remember some search terms that might send me in the right direction I'd really appreciate that.



  • @paloma Exactly. A simple circuit to hold the reset pin LOW for a few milliseconds. A capacitor between RESET and GND and a resistor from RESET to +3V3 (available at pin 26 @ Gpy; please check the pin number for LoPy).

    Start with short times (t=R*C: 1 uF capacitor x 100 kOhm resistor = 100 ms, roughly) because a too long stretched RESET signal may also cause problems.

    If you use the Vin voltage (+5 V) with a large resistor like 1M, there will be a permanent small current of about 0.5 uA into the RESET pin. That will not harm the chip but causes an additional load to your battery.



  • @paloma I understand there’s an issue with the Expansion board 3.0 that prevents booting when running on battery.

    I believe there’s a simple fix involving a connection to make or a resistor to add somewhere. You should find the details in another thread on the forum.



  • @elizar Just to clarify to make sure I'm following, you're suggesting a resistor and capacitor between the reset pin and the Vin pin?



  • @paloma Maybe a simple R-C combination at the reset pin would do the trick to compensate for the bouncing of the magnetic switch?



  • I have a similar issue with the wipy/pysense. Removing the power for a certain amount of time locks up the device in an unresponsive state you can only leave via a reset button press. I'll add a supply supervisor generating a proper reset in case of power fail.
    Many (most?) processors will lock up if power supply startup/shutdown does not follow a defined (read: fast enough and going all the way up to/down to 0 V) slope. I even once had a case where a battery powered device hung up when the battery was removed for like 5 s. The voltage would drop slowly during that time, but not low enough to guarantee a proper restart. In this case, it was not recoverable, even by a power on reset.
    I have not yet tried if the supervisor approach works (still waiting for the part to arrive), it will help but I would not be surprised if in some cases a lockup would occur.



Pycom on Twitter