Overheating of a Rpi 4

[tinpanalley]

All I see online is articles about how to stop the thing from overheating. And if it overheats so easily, what's the point of producing it? I mean, maybe this is just the point at which the Pi's small form factor hits a wall and makes it less useful? Is it possible to run a Rpi4 as a media device and have it not overheat?

[drgeoff]

4Bs do not overheat. They will throttle back before that happens.

[tinpanalley]

4Bs do not overheat. They will throttle back before that happens.

Could that cause problems with video playing off a SMB share home network? Or only if they're 2K or higher?

[Imperf3kt]

All I see online is articles about how to stop the thing from overheating.

What you are seeing is the usual nonsense people put out, trying to make money with clicks and hardware sales (fans, heatsink, cases)

The truth is that the Pi does not require cooling unless you're going to be using it to its maximum potential, and for a simple media player, that's doubtful. Even then, it has protection built in to prevent overheating, which will impact performance as a side effect. This is probably what all those articles are actually referring to.

Don't trust everything you read on the internet.

[tinpanalley]

What you are seeing is the usual nonsense people put out, trying to make money with clicks and hardware sales (fans, heatsink, cases)
The truth is that the Pi does not require cooling unless you're going to be using it to its maximum potential, and for a simple media player, that's doubtful.
Don't trust everything you read on the internet.

This is precisely what I thought. Thank you.
I imagine anything more than 4GB for running Kodi for local shares is unnecessary.

[boyoh]

All I see online is articles about how to stop the thing from overheating.

What you are seeing is the usual nonsense people put out, trying to make money with clicks and hardware sales (fans, heatsink, cases)

The truth is that the Pi does not require cooling unless you're going to be using it to its maximum potential, and for a simple media player, that's doubtful. Even then, it has protection built in to prevent overheating, which will impact performance as a side effect. This is probably what all those articles are actually referring to.

Don't trust everything you read on the internet.
[/quote



This Post By " Imperf3kt" on the subject of the Raspberry Pi over heating, Is one of the best posts on the subject of Over Heating

The Raspberry Pi if you load it as it should be you don't need , extra cooling ( Heat Sink & Fan makers love the Raspberry Pi )

Regards BoyOh Retired Electrical / Electronics Technician .,

[davidcoton]

The Raspberry Pi if you load it as it should be you don't need , extra cooling

I'm sorry, but in refuting the common claims that extra cooling is essential, you have fallen into the opposite trap and thrown out the baby with the bathwater.

Given that:

• Individual Pi differ in their thermal characteristics (I don't know by how much, but I suspect significantly).
• People often assume that a Pi running a given task is working harder than it really is.
• People have different perceptions of how hot is too hot.

Additional cooling may not be necessary, a lot can be achieved with correct orientation and good natural airflow. Personally I run two Pi4 at near 100% CPU load 24/7. They both need cooling to avoid throttling, but with cooling they stay below my personal target of 65C -- often substantially below, currently 48C and 61C.

[Voltime]

Additional cooling may not be necessary, a lot can be achieved with correct orientation and good natural airflow. Personally I run two Pi4 at near 100% CPU load 24/7. They both need cooling to avoid throttling, but with cooling they stay below my personal target of 65C -- often substantially below, currently 48C and 61C.

Totally agree with that. I recently bought a Rpi4B for SMB and Torrenting and got temperatures between 50°C and 65°C on stock clock and without cooling. I still added a passive cooling Aluminium block on it. Not only for cooling reasons (it's now between 40-55°C with slight overclock) but also to prevent the board from getting dusty with time.

[user12345]

All I see online is articles about how to stop the thing from overheating. And if it overheats so easily, what's the point of producing it? I mean, maybe this is just the point at which the Pi's small form factor hits a wall and makes it less useful? Is it possible to run a Rpi4 as a media device and have it not overheat?

Some Smartphones cant handle full clocking too without throttling.
My PI4 is at 30-35% CPU and 58°C.
But
a) it has passive cooler and additional holes in the case(without drilled holes it was about 65°C)
b) its clocked at 550-600MHZ and gpu at 432MHZ.

If i dont do that it would get to hot and throttle.
I make it too cause of the energy efficiency, PI4 requires more power for the same task than a PI2 or 3.

See:
https://www.raspberrypi.org/blog/therma ... erry-pi-4/

[jahboater]

Additional cooling may not be necessary, a lot can be achieved with correct orientation and good natural airflow.

Agreed!

but with cooling they stay below my personal target of 65C

That's your own private personal target.
There is no point telling people they need extra cooling to run a Pi at about 20C less than the normal firmware target.

[davidcoton]

Additional cooling may not be necessary, a lot can be achieved with correct orientation and good natural airflow.

Agreed!

but with cooling they stay below my personal target of 65C

That's your own private personal target.
There is no point telling people they need extra cooling to run a Pi at about 20C less than the normal firmware target.

There is every point in quoting my example. I need cooling to meet my target with my workload. Others may well have similar goals, for reasons of Pi longevity, touch safety, or whatever. Such goals may even be based on opinion or feeling rather than scientific fact. You cannot arbitrarily say such a goal is wrong, any more than you can say a goal of climbing Mount Everest is wrong.

Note I also said I need cooling to avoid throttling. Clearly a lesser cooling solution might avoid throttling without getting down to 65C, but I cannot achieve the no throttling goal without some cooling. If I was really concerned, I would use active fan control possibly with a higher personal target -- but I don't want a project to build and deploy the necessary switching elements, and the RPT fan with direct GPIO control is too noisy.
As I said very clearly, you may not need cooling to meet your target with your workload. Where you set your temperature target: as "let it throttle", "just avoid throttling," or some lower target, is a matter of choice. So, cooling is sometimes necessary, but not always.

To go a step further, if cooling is necessary to meet your target in your circumstances, you may need a heatsink, or a fan, or both.
Anyone who proclaims one, true, complete, universal answer has either not understood the question or is misguided.

[jahboater]

To go a step further, if cooling is necessary to meet your target in your circumstances, you may need a heatsink, or a fan, or both.

Yes. Actually I agree with most of your post.

What I meant was this:

I keep seeing posts where people say words to the effect "you absolutely need a fan because I personally require a fan to keep my Pi below 45C". (or some other totally arbitrary temperature).
So the beginner goes away thinking they need a fan when its likely that they do not.

Any temperature less than the firmware throttling temperature is purely personal choice, usually based on some gut feeling about longevity.

There is only one hard limit that affects all users equally and that is 80C.
Therefore I am suggesting that recommendations for cooling solutions be based on that figure.

That is, unless the questioner explicitly says, "please suggest a cooling solution to maintain my Pi below XXC".

[davidcoton]

What I meant was this: ....

But that was exactly my point, but avoiding the opposite trap of assuming that 80C is a satisfactory limit.

The educational part of the answer is to ensure that the right question is being asked -- thus avoiding false assumptions in either direction, and any other (conscious or unconscious) bias in our answers. Over-simplification is a common cause of error, especially in this case.

Both "you must have a liquid nitrogen cooling solution" and "you never need cooling because the Pi will protect itself" are so incomplete that they are wrong.(Before anyone comments, yes one will be wrong more often than the other, though both could be correct in some circumstances.)

[jahboater]

What I meant was this: ....

But that was exactly my point, but avoiding the opposite trap of assuming that 80C is a satisfactory limit.

The educational part of the answer is to ensure that the right question is being asked -- thus avoiding false assumptions in either direction, and any other (conscious or unconscious) bias in our answers. Over-simplification is a common cause of error, especially in this case.

Both "you must have a liquid nitrogen cooling solution" and "you never need cooling because the Pi will protect itself" are so incomplete that they are wrong.(Before anyone comments, yes one will be wrong more often than the other, though both could be correct in some circumstances.)

Agreed!

I have been assuming that the 80C threshold is satisfactory simply because it was chosen by the RPF engineers - likely in consultation with the chip manufacturers. Also, it is the one and only temperature, which if exceeded, results in visible effects (a slow down).

For me, I like a lower temperature because it gives a safety margin for hot weather!

[bjtheone]

What I meant was this: ....

But that was exactly my point, but avoiding the opposite trap of assuming that 80C is a satisfactory limit.

The educational part of the answer is to ensure that the right question is being asked -- thus avoiding false assumptions in either direction, and any other (conscious or unconscious) bias in our answers. Over-simplification is a common cause of error, especially in this case.

Both "you must have a liquid nitrogen cooling solution" and "you never need cooling because the Pi will protect itself" are so incomplete that they are wrong.(Before anyone comments, yes one will be wrong more often than the other, though both could be correct in some circumstances.)

Agreed!

I have been assuming that the 80C threshold is satisfactory simply because it was chosen by the RPF engineers - likely in consultation with the chip manufacturers. Also, it is the one and only temperature, which if exceeded, results in visible effects (a slow down).

For me, I like a lower temperature because it gives a safety margin for hot weather!

The other aspect is the impact of higher operating temperatures on long term reliability and MTBF (mean time between failure). Higher temperatures impact both. Realistically most workloads are not sustained such that they would pin the SOC at the throttling temperature indefinitely and I am sure the decision of where to set the throttling point takes this into account. However, lower temperatures are not going to adversely impact operation and as a former Component Reliability Engineer I am happy to add on demand fans to my desktop Pis.

[jahboater]

The other aspect is the impact of higher operating temperatures on long term reliability and MTBF

Yes.
But the expected lifetime is around 30 to 35 years I believe.
Probably outlast me no matter what temperature I run it at

[dinko]

The other aspect is the impact of higher operating temperatures on long term reliability and MTBF (mean time between failure). Higher temperatures impact both. Realistically most workloads are not sustained such that they would pin the SOC at the throttling temperature indefinitely and I am sure the decision of where to set the throttling point takes this into account. However, lower temperatures are not going to adversely impact operation and as a former Component Reliability Engineer I am happy to add on demand fans to my desktop Pis.

This is why I run the "stock" Canakit fan albeit at 3.3V instead of 5V.

My simple testing found that idling is about 13C cooler when running the fan at 3.3V than running without a fan (closed Canakit case with venting). It was about 15C cooler when running at 5V but the noise was annoying.

The 3.3V fan adds 0.05 Watts where as the 5V fan added about 0.1 Watts. I didn't do any testing without heatsinks. For what it's worth, lid vs no lid with no fan is about a 5C difference while idling.

If my Pi ever inexplicably dies, I'd never know if it was due too excessive (avoidable) heat. I feel like a low voltage fan with stock heat sinks is a good compromise; it added less than $5 to the cost and probably drops the temperature at least 15C when idling.

If this fan dies, I'll get a GPIO temperature controlled fan and won't have it spin at all it hits 60C. Yes, that's an arbitrary number but at least it would be silent most of the night and that's the temperature where water starts to scald. So if it's not spinning, it's probably safe open it up and not get burned.

[jahboater]

If my Pi ever inexplicably dies,

If its does, its most likely due to accidentally touching a GPIO pin whilst the Pi is powered on, or something similar.

I have never heard of a Pi failing early because of running a few degrees hotter, they have not been around long enough.

Purveyors of cooling parts will likely say otherwise