Questions durability Raspberry PI 3

[gamemaniac]

What is the useful life expectancy of the raspberry pi 3? how many years and how many hours of use it can go resist to fail?

[MarkHaysHarris777]

What is the useful life expectancy of the raspberry pi 3? how many years and how many hours of use it can go resist to fail?

That is a trick question. Some PI engineer will have a mean-time-to-failure; but, the real question is what is the mttf for a PI without heatsinks vs one with heatsinks? I doubt they went to the expense of having military grade components installed... not for $35... so who knows?

... probably years and years and years/

I suspect the SD card will crap out before the PI does... way before.

[darkbibble]

depends on how your running it. overclocks reduce life expectancy

[Heater]

MarkHaysHarris777

I suspect the SD card will crap out before the PI does... way before.

Yep, just had an SD card catch fire on me. Well almost:

After weeks of running doing nothing but taking camera snapshots and serving them over the web a Pi B+ died. On taking it to the bench and powering it up again smoke started billowing from the SD card.

To my amazement that Pi still works with it's new SD.

I am not the first to report an SD card meltdown here.

[gamemaniac]

It does not have an estimate of hours of use after the pi 3 failure? I will not use overclocked but I'll use emulators (snes, sega genesis, arcade, nes, master system, nintendo 64, game cube, ps1, dreamcast)

[darkbibble]

those emulators will run completely fine for years on a pi3 without stressing it at all

[gamemaniac]

It does not have an estimate of hours of use after the pi 3 failure?

[Heater]

gamemaniac

It does not have an estimate of hours of use after the pi 3 failure?

Why do you ask?

The thing may fail after 5, 10, 15, 20, 25, 30 or more years of use. Who knows?

If you are only playing games on emulators why worry about it?

There will be emulators to run those games on whatever comes in the future.

[davidcoton]

It does not have an estimate of hours of use after the pi 3 failure?

RPF may have an estimate of MTTF, and may even have published it somewhere, but it only has a limited bearing on how long your sample of one will last.
The Pi3 is too new to have real figures for life (and even when they are available, it is not likely to be correlated with usage).
All that could be known now is the percentage that have failed so far, possibly with reasons (do you consider the number of Pi3s that failed because they fell down a volcano crevasse to be relevant to your case? I hope not, because I don't know the answer.)

[Heater]

The Pi is a consumer product. As such around here, Scandinavia, it is subject to a two year warranty. That's not to say it won't fail in that time but the store you buy it from has to replace it or refund.

All bets are off for purchases over the internet.

Really what is this? A Printed circuit board with chips on it can be expected to run for ages. If it does not it's cheap enough not to care about anyway.

[Gavinmc42]

Have an old model A that has been running since Nov 2012.

[gamemaniac]

it can run for 20, 30 years on 16 hours daily?

[RaTTuS]

switch on and off may shorten it's life - keep it on
YMMV

Code: Select all

 uprecords
     #               Uptime | System                                     Boot up
----------------------------+---------------------------------------------------
     1   335 days, 17:46:09 | Linux 3.18.7+             Tue Mar 31 17:05:46 2015
     2   130 days, 15:49:43 | Linux 3.6.11+             Wed Dec 18 19:52:36 2013
     3   112 days, 08:14:35 | Linux 3.6.11+             Wed Aug 28 12:33:35 2013
     4    91 days, 05:20:26 | Linux 3.12.24+            Mon Jul 21 10:54:30 2014
     5    77 days, 00:36:37 | Linux 3.12.24+            Mon Oct 20 16:15:16 2014
     6    71 days, 22:59:51 | Linux 3.6.11+             Tue Mar 12 08:37:44 2013
     7    69 days, 20:43:58 | Linux 3.12.35+            Mon Jan  5 15:52:24 2015
->   8    66 days, 05:39:52 | Linux 4.1.13+             Tue Mar  1 09:54:39 2016
     9    43 days, 23:35:53 | Linux 3.10.25+            Mon Apr 28 12:44:21 2014
    10    39 days, 21:42:05 | Linux 3.12.20+            Wed Jun 11 12:20:40 2014
----------------------------+---------------------------------------------------
1up in     3 days, 15:04:07 | at                        Tue May 10 07:38:36 2016
no1 in   269 days, 12:06:18 | at                        Tue Jan 31 03:40:47 2017
    up   1366 days, 22:57:5 | since                     Tue Jul 17 12:46:07 2012
  down    22 days, 04:50:28 | since                     Tue Jul 17 12:46:07 2012
   %up               98.402 | since                     Tue Jul 17 12:46:07 2012

[davidcoton]

it can run for 20, 30 years on 16 hours daily?

It could.
Or it could die tomorrow. You need to buy a crystal ball.

[gamemaniac]

raspberry uses various components (chips, capacitors) they have quality and durability?

[Heater]

Yes.

Next question...

[Rive]

The Pi is a consumer product. As such around here, Scandinavia, it is subject to a two year warranty. That's not to say it won't fail in that time but the store you buy it from has to replace it or refund.

All bets are off for purchases over the internet.

Not like that here in the US... store returns vary from around 15-90 days, and warranties are usually handled by the manufacturer (not store) for consumer goods, regardless of where it was purchased (store or online)... so that would be element14 usually.

The only exclusion to this warranty is resale by unauthorized sellers (e.g., ebay), and warranty usually only is valid for the original purchaser. Hence warranties require legitimate proof of purchase from an authorized seller in the form of a sales receipt.

[gamemaniac]

Quais sao os tipos de capacitores usados no pi3 e sua vida util?

[jahboater]

it can run for 20, 30 years on 16 hours daily?

I saw a figure of 40 years for the Pi1.

[W. H. Heydt]

raspberry uses various components (chips, capacitors) they have quality and durability?

Yes, and there are established formulae to calculate the chance of failure from the chances for each each component. Basically, you multiply the probability of all the components together to get the probability of continued operation of the whole system. The failure probability is obtained by subtracting that figure from 1.

To create a simple example... IF you have two components each with a 50% failure rate over some time period, the the chance of the combination failing in that time is 75%.

[ejolson]

raspberry uses various components (chips, capacitors) they have quality and durability?

Yes, and there are established formulae to calculate the chance of failure from the chances for each each component. Basically, you multiply the probability of all the components together to get the probability of continued operation of the whole system.

Multiplying the probabilities assumes that the reliability of the components are independent of the way they were connected together. For example, placing a Pi inside a plastic case would likely increase its durability beyond what would be calculated as the product of the individual reliabilities. Connecting it inside a much more durable blender, might have the opposite effect.

Similar to the blender, connecting a 64-bit CPU to a power supply without a heat sink may change the combined reliability figures for the whole in a negative way.

For the use case of a child playing with a Pi to learn computer science, the manner in which sdcards and GPIO wires are connected and the level of expert guidance dramatically affects how soon before the Pi breaks. Compared to toys such as footballs and Barbie dolls, a bare circuit board is not very durable.

[Heater]

Clearly estimated MTBF failure rates of individual components can only a be applicable to carefully controlled usage and environments. Not to mention the ways they are connected together.

Throw a hundred or so such components together, put the whole assembly in a teenagers bedroom, it's amazing the thing works for five minutes.

Same applies to any industrial/commercial usage.

The Pi is designed as a "consumer" product. Don't expect any more than that.

[W. H. Heydt]

raspberry uses various components (chips, capacitors) they have quality and durability?

Yes, and there are established formulae to calculate the chance of failure from the chances for each each component. Basically, you multiply the probability of all the components together to get the probability of continued operation of the whole system.

Multiplying the probabilities assumes that the reliability of the components are independent of the way they were connected together. For example, placing a Pi inside a plastic case would likely increase its durability beyond what would be calculated as the product of the individual reliabilities. Connecting it inside a much more durable blender, might have the opposite effect.

I gave a simple formula (note my use of "formulae"). In part because the Pi doesn't have redundant parts (that takes a more complex formula). I would think that putting a Pi in *any* case would reduce the expected MTBF simply by restricting airflow and making it run (even if marginally) warmer. These sorts of calculations don't generally concern themselves with any sort of abuse to the device.

Back when this subject first came up--about 4 years ago--the expected life time of the SoC under normal operation was given as 40 years. I have personally run an original Pi B pretty much continuously for nearly 4 years with no problems. At this point, I think it is reasonable to expect any Pi go long enough without failure that it will be replaced as hopelessly obsolete....even for a Pi.

[Heater]

Your simple formula is just fine.

As long as all components are always operated within the envelope that the formula applies to.

We have no publicly available information about the Pi circuitry to be sure that is so. Never mind the MTBF figures of the parts themselves.

So it's all speculation.

Of course with 8 million Pi out in the wild, with some crowd feed back we could soon get an estimate of actual failure rates.

I fully expect any Pi I have, operated sensibly, to out live me. So what the heck?

[stderr]

Yes, and there are established formulae to calculate the chance of failure from the chances for each each component. Basically, you multiply the probability of all the components together to get the probability of continued operation of the whole system.

Multiplying the probabilities assumes that the reliability of the components are independent of the way they were connected together. For example, placing a Pi inside a plastic case would likely increase its durability beyond what would be calculated as the product of the individual reliabilities.

But don't worry, we don't actually know what the reliability of the individual components are either, unless they are hard drives, in which case the failure rate over the span is 1, especially if they are made by Kalok.