How necessary is extra cooling for the RPi4 now?

[elias4444]

I know when the RPi4 first came out it needed some extra cooling. I got great results with the Flirc passively cooled case. I'm buying another RPi4 now and was wondering how necessary another Flirc case would be (I prefer the silence of passive cooling).

Now that they've updated the firmware and cooled things down a bit, is a stick-on heatsink in a ventilated plastic case enough? I do run some CPU intensive tasks from time to time, so I wanted to make sure I get whatever I might need.

[W. H. Heydt]

/shrug It depends on what you're doing with the Pi4B. The one I have set up for my grandson has a fan shim installed, but that was more because I had one to hand than anything else. The other Pi4Bs I have in use mostly have RTCs installed because there will be times when they won't have an internet connection. The RTCs preculde fans. If they develop thermal issues, the first change will be to remove the case tops. The second will be a nearby USB powered fan.

[bjtheone]

I run a 4B 8 GB as a desktop. With a Fan Shim set to trigger at 65 C, it will cycle on and off if I am streaming video or browsing very actively. Otherwise it stays off. This is in a very open (but horizontal orientation) case in an air conditioned office that sits around 24 C. I doubt that a non artificial load will get it to throttle in an open case, with just passive cooling. In a more closed in case I suspect it might throttle under heavy streaming load.

[alphanumeric]

IMHO It's debatable whether its necessary or just desirable. I'm of the opinion that cooler is better so I have been using the Pimoroni Fan Shim on mine. Worked very well for me, you can set the on off temps, RGB LED to show current temps, and a button that can be used to shut down the Pi. My only complaint would be the friction fit "shim" bit. It's caused issues for some. https://shop.pimoroni.com/products/fan-shim
I just recently bought a passive heatsink case that I'm liking a lot.
https://shop.pimoroni.com/products/alu ... 235997267
Not all that Hat friendly though. I have a Blinkt on a Booster header showing CPU temps. Under moderate load I'm seeing ~60c. And its about 28c in that room right now. The Fan shim is ~50c under the same conditions.

[jlgreer1]

I am an RPi original adopter. I had the first RPi in the first PiBow case (the one with the hearts etched on the top.)

As performance density increases in an electronic device, cooling becomes an issue. The RPi 4 have reached that level. As with all of my Pis, my 4B - 4 Gb is in a PiBow case. It is the first with a fanshim. I live in SE Texas where it is in the upper 80s (F) even at night. Our AC struggles to keep the house in the upper 70s during the day.

I have to run the fanshim whenever I am using the RPi. the CPU temp generally stays in the upper 30s to lower 40s (C). I have experimented with the passive heat sink cases. What a disaster. With significant use, the temperature spiked freezing the RPi. I have purchased an Argon One. It looks promising and has a very nice feature set. I may get around to giving it a try. I am just uncomfortable when the CPU temps move into the 50s and above.

Cooling has been an issue with personal computers since its early days. I had a MacBook Pro that lost its smoke earlier this year. An excellent company in Huntsville, Al that does component level repair tried to replace the errant GPU. Too much heat damage to the motherboard. They said it was common with that particular model and year. I replaced it with a Lenovo.

I will probably continue to use cases that are mostly open with some sort of fan cooling like the Pibow and fanshim. Bitscope has some interesting Blades in the works. It might be fun to have a rack of Pis on the back of my desk. Each Pi in the Blade with its own cooling.

I hope to have an 8 Gb RPi shortly. My current projects include Arm 64 Assembly.

Jeff

PS. I continue to be amazed at the great works the RPi Foundation continues to do. I have never had an RPi fail. I usually keep one of each model for my collection. Many have been given to students. Even with students' rough handling the RPis rarely fail.

[Heater]

I know when the RPi4 first came out it needed some extra cooling.

No you don't. It did not.

My Pi 4 have never had heat sinks or fans. They work fine.

Of course if you are loading it with work that causes it to throttle the CPU and you want some performance back, then stick a heat sink on it, use a fan, put it in a case that also sinks heat.

There is so much talk about this here about this. It's not a big deal.

[alphanumeric]

I know when the RPi4 first came out it needed some extra cooling.

No you don't. It did not.

My Pi 4 have never had heat sinks or fans. They work fine.

Of course if you are loading it with work that causes it to throttle the CPU and you want some performance back, then stick a heat sink on it, use a fan, put it in a case that also sinks heat.

There is so much talk about this here about this. It's not a big deal.

My first Pi 4B 4 gig, would throttle with no heatsink or cooling fan. No damage was done, but it hurt its performance. And it happened relatively quickly as soon as I put any kind of load on the SOC. I was not alone in quickly adding a heatsink, and then a fan to stop it from throttling.
That being said, it wasn't long before the Pi foundation released updates to reduce the operating temperature and maintain performance.
IMHO, there was a time when it was very desirable to add some type of cooling to your Pi 4B if you didn't want to take a performance hit because it got hot under load.

[Heater]

Exactly. If you feel the need for speed put a heat sink and fan or whatever on it.

Actually my understanding of physics suggests that if you need all four cores maxed out to 100% then no tweaking of firmware/driver etc software is going to stop it burning power and making heat.

I guess I just don't see why this is such a big and endless topic for discussion. Just do it.

[Imperf3kt]

If it helps at all, I live in Australia, where the temperatures are regularly above 100°F. (The last two summers have reached peak temperatures of 116°F)
My Pi4b is merely inside a flirc case and doesn't reach throttling temperatures even when artificially stressing it with something like CPU burn.
Another Pi4b sits in open air, propped on its edge instead of flat against the desk and it never gets above 68°C unless I intentionally try to max it out and stress it hard.

[LTolledo]

my answer to this type of question is:

what is your preferred SoC operating temp at highest ambient temp ever recorded in room?

if current SoC temp > preferred SoC temp, then YES
if current SoC temp <= preferred SoC temp, then NO

my preferred SoC operating temp: <45C
highest temp ever recorded in room: 35C

therefore: its a YES for me.

[Imperf3kt]

my answer to this type of question is:

what is your preferred SoC operating temp at highest ambient temp ever recorded in room?

if current SoC temp > preferred SoC temp, then YES
if current SoC temp <= preferred SoC temp, then NO

my preferred SoC operating temp: <45C
highest temp ever recorded in room: 35C

therefore: its a YES for me.

It's not bad, but there's a problem with that, it's addressing personal preference, not "necessities" as was asked.

For example, is it necessary for me to do the speed limit on the highway?
If current speed > speed limit then YES
If current speed < speed limit then NO

Clearly, this is not the correct way to look at the issue. There's a difference between what one needs, and what one wants.

The Pi4b only "needs" to be cooled at 80°C (which is achieved via throttling, or user intervention via a heatsink or fan)

All other requirements are "wants" and only confuse new users.

[pidd]

Whilst there was the big temperature drop with one of the updates, some of the recent updates have sneaked the temperature up a few degrees again.

I am fairly convinced that some Pi4-4's run hotter than others, my two have flirc case on with a little bit of extra heatsinking and in normal use (eg using this forum) the temperature is around +22C from ambient. Others get similar performance from considerably less heatsinking, yet I've also checked the flirc case temperatures and they are very close to the cpu temp.

You can see my temps on a currently very underutilised Pi4-4 + flirc + extra in the contact link on the right. Ambient in the room is around 24C to 26C at the moment.

[HermannSW]

If you don't imprison your Pi4B in a case, there is no need for cooling for my Pi4Bs because I run them hanging with air from all sides.
Even cpuburn does not bring temperature above 82°C where throttling kicks in:
https://github.com/ssvb/cpuburn-arm/blo ... burn-a53.S
Without throttling there is no difference for me whether Pi4B runs at 50°C or say 80°C.
Only difference is that you have to be careful when touching it at 80°C.

[davidcoton]

But if you run your Pi4B with a heavy processing load, you may need cooling to avoid throttling. By design, the Pi is limited to 85C, but that is at the expense of performance. Cooling will allow more work throughput before it reaches 80C and begins throttling. (Someone will correct me if I've got the temps for start throttling/hard throttling wrong, but the principle is there.)

I have a Pi4B running four cores basically flat out 24/7. With a small heatsink and a fan, in a case, it runs at around 60C. But I do think that different Pis inherently run at different temperatures for the same workload and environment -- it's difficult to account for all the data points quoted here otherwise.

[emma1997]

I know when the RPi4 first came out it needed some extra cooling.

No you don't. It did not.

Yes it did. It did too!!!

I was one of the first when Pi4 came out and at that time it ran too hot to touch. Few weeks later with a new upgrade it ran MUCH MUCH cooler. Similar situation with original Pi3 but by then new B+ arrived and the old one was shelved. Not seen any of these actually throttle since.

I have never had a Pi in open air need any cooling no matter what ambient, software, or attachments.

[mahjongg]

Too hot to touch simply means its hotter than the temperature that protein coagulates at, its why you die if your temperature gets too high.Say >42 degrees celsius.

But a raspberry PI SoC is NOT made of proteins and the temperature at which protein solidifies does it no harm at all.

If you would try to touch an Intel CPU while it is running you will get far more "burned" than the simple SoC of a PI that heats up with just a couple of watts, instead of the hundreds of watts a typical PC CPU turns into heat.

There is no way a RPI SoC will show heat damage except if you have damaged it by static electricity or by putting more than 4 volt on a GPIO. If it's working normally it cannot overheat, as it will just throttle to a lower speed, using less power until a safe equilibrium is reached.
The way to test if the SoC is damaged is by not giving it any boot code to run and checking if it heats up anyway, if you can feel any warmth from it powered up without an SD-card, then it is damaged.

[emma1997]

So it's true. In addition to teaching Linux to die hard Winders fans (against their will, incidentally), replacing $1000 PC as desktop, and controller for Mars landing, we can use it to fry an egg for breakfast. Maybe sparrow or at the very least hummingbird.

Also worth mentioning I have a very low pain threshold. lol

[Imperf3kt]

So it's true, we can use it to fry an egg for breakfast.

Fry, no. Slowly cook, yes. It does a much better job heating liquids and melting chocolate.

I still require a CNC to produce a proper cooking surface...

[Barie1s]

It is likely extra cooling will be vital if the Raspberry Pi 4 is utilized inside a case.

[mahjongg]

It is likely extra cooling will be vital if the Raspberry Pi 4 is utilized inside a case.

Why?
The only reason I can see is that a case can act like a hothouse, and its inside can heat up fiercely (especially when parked in the sun. It's why you should not leave chocolate on it. (same as inside a car, instead of a case)

Still I doubt it will get so hot that silicium is damaged, its silicium not chocolate, and the PI will simply shuttle down so that critical temperatures can never be reached, unless you heat its environment up to say 250 degrees celsius, at which the plastic of the connectors will melt. LOL.

But you are right that using the red white official plastic case isn't a good idea if you want to use your PI to the max, in that case you will be better off with a case made from two perspex plates screwed to the PI, or simply place the RPI on its side without a case
.

Or you could drill a few holes in the sides of the official case, that wouldn't hurt, as any airflow no matter how little will have a big effect.

[ganzgustav22]

I tortured the Pi4 at 141 degrees Celsius CPU temp at 1500MHz with cpuburn and it worked fine (but I did not test longterm, just for a day).

viewtopic.php?f=63&t=264786&hilit=heat+torture#p1610708

[mwrich4]

Best, if you have to ask, still the same need for original release (nothing magical happened). If you have issues then rethink design.

[MikeDB]

I tortured the Pi4 at 141 degrees Celsius CPU temp at 1500MHz with cpuburn and it worked fine (but I did not test longterm, just for a day).

Just as well - the MTBF will be about 85 times shorter than running at 60 degC
And you've cooked all the caps. Fortunately no electrolytics on a Pi or they'd have all dried out but X5Rs are only rated to 85degC.

[LTolledo]

I still require a CNC to produce a proper cooking surface...

as egg whites starts to "cook" from 60C.....
and if I find a suitable stainless mini-fry-pan (about dia 40mm, depth 6mm)
I might just try it.....as an experiment.....on this RPi4B-4G......(after I get a newer one and replace the RPi4B-4G on this desktop unit)

but if someone beats me to it (with a video or picture collage), then there's no need to me to pursue this...