Millis goes back to zero after 50 days, what happens to delay()??

I am going to use Arduino to make a pump protection, such that the pump does not burn itself up if the water is frozen, the pump clocked etc.

This will be done by measuring the current, issuing short delays, measuring again, another delay. From these measurements I will compute the current power supplied to the pump, and if it exceeds safe levels cut the supply voltage.

Now, the documentation tells me that the millis function goes back to zero about every 50 days. What happens to a delay that is issued before the function millis goes to zero again, but which would have expired after the reset. It seems such a delay would never terminate.

Very much interested in light on this issue! (Rather than having to analyze why pumps blew up every now and then).

Is it possible to reset the millis timer through code (say once a month)?

Thanks, peter

Nothing. Delay() works fine even between rollover.

Arrch: Nothing. Delay() works fine even between rollover.

Unless you try to delay for 52 days ;)

You can check if current millis is smaller than on the previous occasion you checked.

majenko:

Arrch: Nothing. Delay() works fine even between rollover.

Unless you try to delay for 52 days ;)

It will delay for 2 days, so It would still work, just not as intended. :P

majenko:

Arrch: Nothing. Delay() works fine even between rollover.

Unless you try to delay for 52 days ;)

In which case it would be better to turn off the Arduino and use a calendar instead :)

UKHeliBob:

majenko:

Arrch: Nothing. Delay() works fine even between rollover.

Unless you try to delay for 52 days ;)

In which case it would be better to turn off the Arduino and use a calendar instead :)

But the 9V PP3 battery in my calendar runs out before then...

No one sees to have asked why the OP is using “delay()”.
So, here goes: Why?

PeterLq: Now, the documentation tells me that the millis function goes back to zero about every 50 days. What happens to a delay that is issued before the function millis goes to zero again, but which would have expired after the reset. It seems such a delay would never terminate.

The correct technique is to note the current time (in an unsigned long) eg. when you turn the motor on.

eg.

unsigned long motorOnTime;

...

motorOnTime = millis ();

Then you check, using subtraction, if that time is up. That works, even after millis rolls over. Don't attempt to reset millis.

eg.

  if ( (millis () - motorOnTime) >= timeToRun)
    {
    // turn motor off
    }

In this case you set up timeToRun to be the time you want it on (in milliseconds). If you do this in the main loop you are not "blocking" (with delay) and can thus do other things, like check other motors, switch presses, etc.