ASAP! RPI 3 Heats to much!

[We__Create]

My Raspberry Pi 3 heats to 72.4 degres celsius! I am afraid it will burn
What can I do?
It is running retropie.
It does have a heatsink!

Thanks so much!

[wh7qq]

1. Is it in an enclosure? Open it up.
2. How is your heat sink attached? Some of the heat sink tapes don't conduct very well. Remove the heat sink, carefully scrape the tape off of sink or chip...don't gouge either. Use a thin coat of a good grade of heat sink grease such as arctic silver to replace sink. Absent the adhesive, you probably will need to keep the Pi flat.
3. The Pi will start throttling the cpu at 80 deg, ahead of burning up. How are you measuring temp? In a terminal,

Code: Select all

vcgencmd measure_temp

is good enough. 72.8 is hot but not dangerous...but prolonged running at that high temp will shorten life.
4. Last resort, add a fan.
5. If possible, avoid java script.

[HawaiianPi]

My Raspberry Pi 3 heats to 72.4 degres celsius! I am afraid it will burn
What can I do?
It is running retropie.
It does have a heatsink!

Thanks so much!

72.4C is not too hot. The Pi3 will start to throttle above 80, but you aren't even close to that.

[DougieLawson]

My Raspberry Pi 3 heats to 72.4 degres celsius! I am afraid it will burn

You're OK, paper doesn't burn until it reaches 451°F (232.8°C)

[HawaiianPi]

...
...but prolonged running at that high temp will shorten life.

Because you say so, or do you have official documentation to back that up?

You're OK, paper doesn't burn until it reaches 451°F (232.8°C)

[rpdom]

...
...but prolonged running at that high temp will shorten life.

Because you say so, or do you have official documentation to back that up?

I think it has been mentioned a few times here that running a Pi overclocked/at a high temperature may reduce the expected lifespan from 35+ years to under 30 years.

[HawaiianPi]

I think it has been mentioned a few times here that running a Pi overclocked/at a high temperature may reduce the expected lifespan from 35+ years to under 30 years.

OH MY GOD!! What will I do after 30 years? Oh wait, I'll probably be dead by then...

[tpylkko]

Because you say so, or do you have official documentation to back that up?

Often with semiconductor electronics one hears the rule of thumb that "every extra 10 degrees centrigrade halves the rated service life". So if a product is designed and quality controlled to have 10000 cycles of life at 25ºC, then going to 35 permanently will cause service life of something like 5000 cycles. This is based on experience in industry but ultimately stems from thermodynamics. So, even without documentation anyone can rightly suspect that excessive temperatures are to be avoided in electronics. But, of course, any rational end user will also realize that most electronics products we use get obsoleted, for other reasons, long long before the end of service life.

[We__Create]

1. Is it in an enclosure? Open it up.
2. How is your heat sink attached? Some of the heat sink tapes don't conduct very well. Remove the heat sink, carefully scrape the tape off of sink or chip...don't gouge either. Use a thin coat of a good grade of heat sink grease such as arctic silver to replace sink. Absent the adhesive, you probably will need to keep the Pi flat.
3. The Pi will start throttling the cpu at 80 deg, ahead of burning up. How are you measuring temp? In a terminal,

Code: Select all

vcgencmd measure_temp

is good enough. 72.8 is hot but not dangerous...but prolonged running at that high temp will shorten life.
4. Last resort, add a fan.
5. If possible, avoid java script.

It is in a closed case. Case has holes though.

The heatsink gets hot af so it should be working.

Yes it's in a terminal. I decided to check when I saw a thermonoter icon at the side if my screen.

Getting a fan

Thanks!

[HawaiianPi]

So, even without documentation anyone can rightly suspect that excessive temperatures are to be avoided in electronics.

And... how do you know that 72.4C is excessive? Do you have the temperature specifications of the BCM2837 SOC?

The Pi3 is designed to start throttling over 80C, and to fully throttle back above 85C. We can assume those temperatures came from from Broadcom, and I would think Broadcom knows what they are doing (they designed and built the SOC after all).

It is in a closed case. Case has holes though.

The heatsink gets hot af so it should be working.

Hot is good, but it doesn't mean it's dissipating enough heat. Most of the heatsinks available for the Raspberry Pi are too small to be of any use at all. All they will do is allow your system to run a tiny bit longer before overheating (assuming you are doing something that will actually cause overheating, which is not that easy to do).

Adding a fan will help, even without a heatsink.

... I decided to check when I saw a thermonoter icon at the side if my screen.

If you saw this icon it means your system was throttling down to prevent overheating, but it had not actually overheated.


If you saw this icon then it was actually overheating and you should definitely do something to fix that before you play your games on it again.

[jamesh]

I believe the chips were characterised at well over 120degC (they still work at that temp). From behavior at those temps, the estimated lifetime is gauged. So temps going from 72 to 80/85 is actually not a huge leap compared with the temps they can sustain, so claiming a halving of lifetime for that 10 degree or so jump is not quite right.

When used with throttling, the chip should last decades.

[tpylkko]

So, even without documentation anyone can rightly suspect that excessive temperatures are to be avoided in electronics.

And... how do you know that 72.4C is excessive? Do you have the temperature specifications of the BCM2837 SOC?

That question might have made sense had I claimed to know such. But when it is so obviously clear that I didn't, maybe you misunderstood something?

[wh7qq]

If all you are doing is playing games, nobody GAS but if you need to have your RPi running 24/7 doing something serious...well, "...go ahead sucker, make your day (or night)". Testing at 120C or not, somebody in design decided it was a good idea to put on the brakes at 80-85 C. The chip may be ok at 80C but thermal cycling of solder joints... Since I have to live with the consequences of device failure, I tend to be pretty conservative on my home automation devices. OTOH, I run my 3 pretty hot because I have other desktops as backup and if it blows, I am out $40 and some time but nothing else. I just use this for browsing and email.

[jamesh]

If all you are doing is playing games, nobody GAS but if you need to have your RPi running 24/7 doing something serious...well, "...go ahead sucker, make your day (or night)". Testing at 120C or not, somebody in design decided it was a good idea to put on the brakes at 80-85 C. The chip may be ok at 80C but thermal cycling of solder joints... Since I have to live with the consequences of device failure, I tend to be pretty conservative on my home automation devices. OTOH, I run my 3 pretty hot because I have other desktops as backup and if it blows, I am out $40 and some time but nothing else. I just use this for browsing and email.

Testing at high temperatures is used to characterise the chip - to determine how long it will last at lower temperatures. Not sure why 85 degrees was chosen, I suspect it was a good lifetime number and a temperature that wasn't going to burn people. Solder cycling of solder joints, whilst the device is ON, shouldn't be a problem. It's just moving from lets say 60 to 85. The BGA on the SoC is designed for that. As far as I know, solder cycling failure has never been confirmed on the Pi (not to say it hasn't happened), which indicates that it will be exceedingly rare.

[wh7qq]

You are welcome to cycle your system as much as you like...I will continue to avoid it as much as I can. You are also welcome to leave your finger on the 85 C chip surface as long as you like...it will burn you...if you could tolerate it that long. Proteins start coming apart at 44 C. But we are talking about semiconductors and solder joints that lack a liquid cooling system like skin.
Stuff expands and contracts with cycling...the nature of metals, plastics and virtually anything that goes into electronic devices. Differences in rate of thermal expansion for different materials results in mechanical stresses that can cause joint failure. The less that occurs, the fewer failures that occur.

[rpdom]

Stuff expands and contracts with cycling...the nature of metals, plastics and virtually anything that goes into electronic devices. Differences in rate of thermal expansion for different materials results in mechanical stresses that can cause joint failure. The less that occurs, the fewer failures that occur.

Do you think that isn't taken into account when designing and testing parts like these?

[wh7qq]

I am too well aware of the interaction between marketing, manufacturing and design engineering interests and published specs tend to be a compromise between what will run and for how long and what can be built and sold for what price. The Raspberry Pi Foundation which you represent may be lilly pure but theirs are not the only fingers in the Pi (pun intended).

[rpdom]

The Raspberry Pi Foundation which you represent

I do not, and never have claimed to, represent the Raspberry Pi Foundation. I don't know where you got that idea from.

[wh7qq]

The Raspberry Pi Foundation which you represent

I do not, and never have claimed to, represent the Raspberry Pi Foundation. I don't know where you got that idea from.

I am very sorry, you are correct. I got you confused with jamesh who has it in his signature. Kindly accept my apology.

[jamesh]

The Raspberry Pi Foundation which you represent

I do not, and never have claimed to, represent the Raspberry Pi Foundation. I don't know where you got that idea from.

I am very sorry, you are correct. I got you confused with jamesh who has it in his signature. Kindly accept my apology.

In fact, I don't work for them either (yet!). However, for the last 8 years I have worked for semiconductor manufacturers so have some experience in the matter. I am also quite familiar with the process of getting chips on to PCB's.

People have cycled their Pi's on an off for years with no issues with regard to thermal expansion. This is taken in to account during design and manufacture. The BGA connection of the SOC to the PCB is used in billions of products worldwide, not just Pi, it's very reliable, or it would not be used. You are clearly not familiar with the genesis of the Pi, there was no 'marketing' involved at all! There was, of course, design an manufacture, and since they have been making these tings for 4 years oover a number of iterations, have sold 11 million, with a very very low failure rate, I think we can be sure they are well designed and made.

Also, 85 degrees is a temperature at which you cannot keep your finger on the SoC for very low, but as long as you are not stupid or have no sense of feeling, you will remove the finger before burning occurs. Generally people can hold their fingers on about 65-70 degrees, much above that and the reflex is to remove it as quickly as possible! It's a fairly safe compromise.