How Long Can Teensy 4.1 Run Continuously Without Breaking

Status
Not open for further replies.

grinch

Well-known member
Hi, I am wondering how long the Teensy 4.1 can run a program continuously (no powering off). I have a project that I am looking at installing for a long period of time and am wondering what to expect and what kind of maintenance I might need to do to maintain longevity.

I understand the main issue in these scenarios is usually thermal runaway, which doesn't seem like it will be much of an issue here since my Teensy doesn't get hot when running. I'm wondering if there is anything else that can cause the Teensy to break, or any other limit on running a program continuously.

Assuming the device is properly cooled, could a Teensy run for several months straight? Years? Decades? Centuries? What is the upward limit?
 
i've ran it for several years in automotive field. thermal runaway is not the issue of stability, bad programming is. also don't try sourcing power from gpios, and running an inline resistor to protect the gpio is a plus. assuming your code is stable, it'll run 24/7 without any hiccups.
 
Gotcha, that's what I was thinking. My code is stable and I also would include a watchdog in a case like this, since a momentary interruption is acceptable compared to getting stuck in a hung state.

in the very long term, like decades or centuries, is there anything physical that could break down? Mostly just curious about how this process would work.
 
in the very long term, like decades or centuries, is there anything physical that could break down? Mostly just curious about how this process would work.
If you are not deploying it in coastal areas that gets flooded within the next century.
 
Flash memory will lose integrity over years. Guaranteed are 10 to 20 years usually, but nobody can actually tell. A solution could be a refreshing rewrite after a few years, but this contributes to wear as well.
This time will decrease rapidly in situations where radiation or extreme temperatures come into play, as is even noise on power supplies and signal cross coupling, which naturally happens in a circuit.
So if you plan to send Teensy 4.1 to space, do not make just a good shielding, make an excellent shielding. ;-) And add a an excellent power supply. And thermal magement, to keep the Teensy above the allotropic transformation temperature of the solder (kick-in at -30°C, up to 13.2°C), but below the maximum running temperature of the MCU.
Whisker prevention. Ageing solder grows whiskers in changing temperatures and mechanical stress.
 
speaking of whisker growth - do your homework before your last cleaning of the board,
some components of some cleansers/solvents that get left behind will actually accelerate
whisker formation and i would also say do not spray on an anti-fungal coating - i have
seen anti-fungal coatings that once they get old fungus loves to feed and grow on them,
and the corpses of the dead fungus are fairly conductive. metal migration is proprtional
to current so keep clock speed and io current as low as possible. last hurdle is temperature
- cold to moderate is best - you may need to control with insulation an/or playing with
outer layer/coating ir emissivity vs visible absorption. i agree with Deleted User that 10-
20 years is doable.
 
2020-12-15 22_34_40-Start.png

Here is a appnote re:i.MXRT1060 Product Lifetime Usage Estimates
https://www.nxp.com/docs/en/nxp/application-notes/AN12253.pdf

So, try to keep the temperature low.
But, of course, the i.mx is not the only chip/part. And the other factors mentioned in the posts above play an important role, too.





 
Also make sure that your software doesn't rely on comparisons with millis() for timing. The millis() count will overflow in 49.7 days. Of course, micros() overflows even more quickly.
 
Flash memory may be a factor for extremely long life. This is the only info I could find about the flash chip we're using.

flash_retention.png

I've heard that flash memory retention depends heavily on temperature. But again, if Winbond publishes any more info, I don't have it. This vague 20 year estimate is all I've seen.
 
What about the crystal? It's oscillation is due to electrostriction, kind of mechanical, so what about defects due to to a spot of impurity, or after the impact of a particle (radiation) or mechanical or thermal stress? I think this could limit the runtime as well, be it longer than the flash, depending on what happened during manufacture, which may offset reliability a lot.

Semiconductors wear as well. Here is an interesting article. https://spectrum.ieee.org/semiconductors/processors/transistor-aging
 
Status
Not open for further replies.
Back
Top