## pi fan direction airflow question

darkbibble
Posts: 629
Joined: Mon Mar 09, 2015 5:20 pm
Location: corby, england

### pi fan direction airflow question

hi I just got a new little 5v 200ma 3.2cfm fan and cut a hole in the lid of my pibow, the thing is do I need it pulling air out of the case, or do I need it pushing air into the case?
Q; How many Windows users does it take to fix a Linux problem??
A; Whats a Linux problem

MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

### Re: pi fan direction airflow question

darkbibble wrote:hi I just got a new little 5v 200ma 3.2cfm fan and cut a hole in the lid of my pibow, the thing is do I need it pulling air out of the case, or do I need it pushing air into the case?
ha! Do I push or pull?

If the fan is in the top of the case, then evacuate the case by pulling the air out. Why? glad you asked, because you want to minimize the amount of dust in the case, and you want to pull air over the memory chip also. If you pull the air out of the case the new air will come in (mostly) through the hole or slots in the bottom of the case (minimizing dust) and will more evenly cool the entire board.

That said, a fan is generally speaking not necessary. If you have heatsinks on the chips that should be enough (I personally hate noisy little fans). On the other hand, they look kinda cool (pun intended) and they won't hurt anything certainly.

marcus

Rive
Posts: 586
Joined: Sat Mar 26, 2016 5:21 pm
Location: USA

### Re: pi fan direction airflow question

MarkHaysHarris777 wrote:If the fan is in the top of the case, then evacuate the case by pulling the air out.

...that said, a fan is generally speaking not necessary. If you have heatsinks on the chips that should be enough (I personally hate noisy little fans)
I disagree (and especially disagree about the heatsink being enough...if you are in fact referring to an encased Pi3 as there needs to be a way to remove the heat from the heatsink).

I have actually tested this (fan push or pull), and I get better results with push, as demonstrated:

You can see my benchmarks, and cooling tests here (with Stress, Cpuburn-a53, and Linpack):
viewtopic.php?p=921165#p921165
Last edited by Rive on Thu Apr 28, 2016 3:33 pm, edited 3 times in total.
DNPNWO

DirkS
Posts: 10712
Joined: Tue Jun 19, 2012 9:46 pm
Location: Essex, UK

### Re: pi fan direction airflow question

Rive wrote:I have actually tested this (fan push or pull), and I get better results with push, as demonstrated:
Don't rev up the fan too much, otherwise you'll end up with a Pi-copter

MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

### Re: pi fan direction airflow question

DirkS wrote:
Rive wrote:I have actually tested this (fan push or pull), and I get better results with push, as demonstrated:
Don't rev up the fan too much, otherwise you'll end up with a Pi-copter
I think he has to register it with the FAA now; as a drone.

marcus

Rive
Posts: 586
Joined: Sat Mar 26, 2016 5:21 pm
Location: USA

### Re: pi fan direction airflow question

MarkHaysHarris777 wrote:
DirkS wrote:
Rive wrote:I have actually tested this (fan push or pull), and I get better results with push, as demonstrated:
Don't rev up the fan too much, otherwise you'll end up with a Pi-copter
I think he has to register it with the FAA now; as a drone.

Go BIG, or Go Home, right?
DNPNWO

Rive
Posts: 586
Joined: Sat Mar 26, 2016 5:21 pm
Location: USA

### Re: pi fan direction airflow question

Hmmmm

Well, MarkHaysHarris777 is right, and I stand corrected (well at least on "push" vs "pull" debate), but he is still dead wrong on the "no need for a fan" (with a heatsink and in a case) .

Test results:

84C no fan/passive airflow
62C Push
60C Pull

Cpuburn-a53

Code: Select all

wget https://raw.githubusercontent.com/ssvb/cpuburn-arm/master/cpuburn-a53.S
gcc -o cpuburn-a53 cpuburn-a53.S
while true; do vcgencmd measure_clock arm; vcgencmd measure_temp; sleep 10; done& ./cpuburn-a53


No Fan/Passive cooling (15min cpuburn-a53) 84C

Code: Select all

pi@raspberrypi:~ $while true; do vcgencmd measure_clock arm; vcgencmd measure_temp; sleep 10; done& ./cpuburn-a53 [1] 1033 frequency(45)=1260132000 temp=36.5'C frequency(45)=1260002000 temp=53.7'C frequency(45)=1260000000 temp=58.0'C frequency(45)=1260000000 temp=62.3'C frequency(45)=1260000000 temp=65.0'C frequency(45)=1260000000 temp=68.2'C frequency(45)=1260000000 temp=70.9'C frequency(45)=1260000000 temp=73.1'C frequency(45)=1260000000 temp=75.2'C frequency(45)=1260000000 temp=77.4'C frequency(45)=1259998000 temp=79.5'C frequency(45)=1176000000 temp=80.6'C frequency(45)=1128002000 temp=81.1'C frequency(45)=1070998000 temp=81.7'C frequency(45)=1026000000 temp=81.7'C frequency(45)=995000000 temp=82.2'C frequency(45)=958000000 temp=82.7'C frequency(45)=940000000 temp=82.7'C frequency(45)=923000000 temp=83.3'C frequency(45)=880000000 temp=83.3'C frequency(45)=876000000 temp=83.8'C frequency(45)=866000000 temp=83.8'C frequency(45)=852000000 temp=83.3'C frequency(45)=837000000 temp=83.8'C frequency(45)=836000000 temp=83.8'C frequency(45)=835000000 temp=83.8'C frequency(45)=841000000 temp=83.8'C frequency(45)=814000000 temp=83.8'C frequency(45)=842000000 temp=83.8'C frequency(45)=820000000 temp=83.8'C frequency(45)=814000000 temp=83.8'C frequency(45)=817000000 temp=83.8'C frequency(45)=820000000 temp=84.4'C frequency(45)=812000000 temp=84.4'C frequency(45)=809000000 temp=83.8'C frequency(45)=798000000 temp=83.8'C frequency(45)=797000000 temp=84.4'C frequency(45)=789000000 temp=84.4'C frequency(45)=785000000 temp=83.8'C frequency(45)=794000000 temp=84.4'C frequency(45)=800000000 temp=84.4'C frequency(45)=779000000 temp=84.4'C frequency(45)=789000000 temp=83.8'C frequency(45)=770000000 temp=84.4'C frequency(45)=786000000 temp=84.4'C frequency(45)=781000000 temp=84.9'C frequency(45)=774000000 temp=84.4'C frequency(45)=772000000 temp=84.4'C frequency(45)=759000000 temp=84.4'C frequency(45)=774000000 temp=84.4'C frequency(45)=754000000 temp=84.4'C frequency(45)=760000000 temp=83.8'C frequency(45)=756000000 temp=83.8'C frequency(45)=754000000 temp=84.4'C frequency(45)=758000000 temp=84.9'C frequency(45)=749000000 temp=83.8'C frequency(45)=753000000 temp=84.9'C frequency(45)=754000000 temp=84.4'C frequency(45)=750000000 temp=84.4'C frequency(45)=741000000 temp=83.8'C frequency(45)=741000000 temp=84.9'C frequency(45)=747000000 temp=84.9'C frequency(45)=738000000 temp=84.9'C frequency(45)=746000000 temp=84.9'C frequency(45)=755998000 temp=84.9'C frequency(45)=758000000 temp=84.9'C frequency(45)=757000000 temp=84.9'C frequency(45)=761000000 temp=84.9'C frequency(45)=775000000 temp=84.4'C frequency(45)=766000000 temp=84.9'C frequency(45)=761000000 temp=84.4'C  Push (15min cpuburn-a53) 62C Code: Select all pi@raspberrypi:~$ while true; do vcgencmd measure_clock arm; vcgencmd measure_temp; sleep 10; done& ./cpuburn-a53
[1] 1036
frequency(45)=1260134000
temp=35.4'C
frequency(45)=1260000000
temp=52.1'C
frequency(45)=1260000000
temp=55.3'C
frequency(45)=1260000000
temp=57.5'C
frequency(45)=1260002000
temp=58.0'C
frequency(45)=1260000000
temp=59.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260002000
temp=60.1'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260002000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260002000
temp=61.8'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260002000
temp=62.3'C
frequency(45)=1260002000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260002000
temp=62.3'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260002000
temp=62.3'C
frequency(45)=1260002000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.8'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1259998000
temp=62.8'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260002000
temp=62.3'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1259998000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C
frequency(45)=1259998000
temp=62.3'C
frequency(45)=1259998000
temp=62.3'C
frequency(45)=1259998000
temp=62.3'C
frequency(45)=1260002000
temp=62.3'C
frequency(45)=1260000000
temp=62.3'C



Pull (15min cpuburn-a53) 60C

Code: Select all

pi@raspberrypi:~ \$ while true; do vcgencmd measure_clock arm; vcgencmd measure_temp; sleep 10; done& ./cpuburn-a53
[1] 1054
frequency(45)=1260132000
temp=32.2'C
frequency(45)=1260000000
temp=49.4'C
frequency(45)=1260000000
temp=52.6'C
frequency(45)=1260000000
temp=54.8'C
frequency(45)=1260000000
temp=55.8'C
frequency(45)=1260000000
temp=56.9'C
frequency(45)=1260000000
temp=58.0'C
frequency(45)=1260000000
temp=58.0'C
frequency(45)=1260000000
temp=59.1'C
frequency(45)=1260000000
temp=58.5'C
frequency(45)=1260000000
temp=59.6'C
frequency(45)=1259998000
temp=59.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1259998000
temp=60.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260002000
temp=60.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1259998000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260002000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260002000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260002000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260002000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1259998000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260002000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260002000
temp=61.8'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260002000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.8'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1259998000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260002000
temp=61.2'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=61.2'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.1'C
frequency(45)=1260000000
temp=60.7'C
frequency(45)=1260000000
temp=60.7'C

DNPNWO

MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

### Re: pi fan direction airflow question

Rive wrote:Hmmmm

Well, MarkHaysHarris777 is right, and I stand corrected (well at least on "push" vs "pull" debate), but he is still dead wrong on the "no need for a fan" (with a heatsink and in a case) .
Keep in mind I meant in general, and I also meant in my own experience. In other words, I can see from your example tests that you have shown a need, and you have shown some evidence; what I'm saying is that in general the system will generally speaking not be loaded that way in most cases and will not require the extreme fan solution you have provided; although, it does look like your setup requires it!

In my own experience small heatsinks (all three main chips) are sufficient to keep the system from throttling while running medium loads normal processing most of the time (normal operations, compiling, number crunching, photo or stream editing, &c). I bought the fan, and its sitting here if I need it, but I have not needed it yet... just saying...
marcus

Heater
Posts: 18862
Joined: Tue Jul 17, 2012 3:02 pm

### Re: pi fan direction airflow question

fans suck. Excuse the pun. Noisy, unreliable, mechanical stuff.

Hard drives too.

If my Pi needs a fan I don't need the Pi.

Luckily this is not the case so far.
Memory in C++ is a leaky abstraction .

Rive
Posts: 586
Joined: Sat Mar 26, 2016 5:21 pm
Location: USA

### Re: pi fan direction airflow question

MarkHaysHarris777 wrote:
Rive wrote:Hmmmm

Well, MarkHaysHarris777 is right, and I stand corrected (well at least on "push" vs "pull" debate), but he is still dead wrong on the "no need for a fan" (with a heatsink and in a case) .
Keep in mind I meant in general, and I also meant in my own experience. In other words, I can see from your example tests that you have shown a need, and you have shown some evidence; what I'm saying is that in general the system will generally speaking not be loaded that way in most cases and will not require the extreme fan solution you have provided; although, it does look like your setup requires it!

In my own experience small heatsinks (all three main chips) are sufficient to keep the system from throttling while running medium loads normal processing most of the time (normal operations, compiling, number crunching, photo or stream editing, &c). I bought the fan, and its sitting here if I need it, but I have not needed it yet... just saying...
Do you consider floating point calculations normal processing? On a pi3 wouldn't that cause you to overheat & throttle?
In computing, floating point is the formulaic representation that approximates a real number so as to support a trade-off between range and precision. A number is, in general, represented approximately to a fixed number of significant digits (the significand) and scaled using an exponent; the base for the scaling is normally two, ten, or sixteen. A number that can be represented exactly is of the following form:

\text{significand} \times \text{base}^\text{exponent},
where significand ∈ Z, base ∈ N, and exponent ∈ Z.

For example:

1.2345 = \underbrace{12345}_\text{significand} \times \underbrace{10}_\text{base}\!\!\!\!\!\!^{\overbrace{-4}^\text{exponent}}
The term floating point refers to the fact that a number's radix point (decimal point, or, more commonly in computers, binary point) can "float"; that is, it can be placed anywhere relative to the significant digits of the number. This position is indicated as the exponent component, and thus the floating-point representation can be thought of as a kind of scientific notation.

A floating-point system can be used to represent, with a fixed number of digits, numbers of different orders of magnitude: e.g. the distance between galaxies or the diameter of an atomic nucleus can be expressed with the same unit of length. The result of this dynamic range is that the numbers that can be represented are not uniformly spaced; the difference between two consecutive representable numbers grows with the chosen scale.[1]

Over the years, a variety of floating-point representations have been used in computers. However, since the 1990s, the most commonly encountered representation is that defined by the IEEE 754 Standard.

The speed of floating-point operations, commonly measured in terms of FLOPS, is an important characteristic of a computer system, especially for applications that involve intensive mathematical calculations.
Souce

What happens when you run this?

Code: Select all

sudo apt-get install libmpich-dev
wget http://web.eece.maine.edu/~vweaver/junk/pi3_hpl.tar.gz
tar -xvzf pi3_hpl.tar.gz
chmod +x xhpl
./xhpl
DNPNWO

darkbibble
Posts: 629
Joined: Mon Mar 09, 2015 5:20 pm
Location: corby, england

### Re: pi fan direction airflow question

I got it pulling air out the top now, will test tomorrow when I got time to see if I still get yellow square when running quake3

thanks for the advise
Q; How many Windows users does it take to fix a Linux problem??
A; Whats a Linux problem

stderr
Posts: 2178
Joined: Sat Dec 01, 2012 11:29 pm

### Re: pi fan direction airflow question

MarkHaysHarris777 wrote:If the fan is in the top of the case, then evacuate the case by pulling the air out. Why? glad you asked, because you want to minimize the amount of dust in the case, and you want to pull air over the memory chip also. If you pull the air out of the case the new air will come in (mostly) through the hole or slots in the bottom of the case (minimizing dust) and will more evenly cool the entire board.
Most tower type cases blow air out of the back, most all power supplies blow air out the back of the power supply and take air in from inside the system. In this configuration, the hotter air should rise to the location of the power supply and be expelled. The cases often have holes or slits in the front in the bottom region of the case. The problem is that there is often a lot more dust closer to the floor than in the air higher up and people who place their systems on the floor tend to see a lot of dust, especially when the case fans or even just the power supply fan run all the time. Essentially they become your round the clock hoover. Once these front slots are clogged, the next place for the air to come in from used to be the floppy disks and then the CD/DVD devices, which contributed to their doom. People who smoke would probably give it up if they looked inside their computer case. Yuk.

MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

### Re: pi fan direction airflow question

stderr wrote:
MarkHaysHarris777 wrote:If the fan is in the top of the case, then evacuate the case by pulling the air out. ...
Most tower type cases blow air out of the back, most all power supplies blow air out the back of the power supply and take air in from inside the system.
Yup. Those of us with experience with attic fans know this too... all attic fans pull air in through the house windows and blows it out the top/ and its immediately 'feelable'.
stderr wrote: ... People who smoke would probably give it up if they looked inside their computer case. Yuk.
Oh, absolutely. Cigarette smoke filtered through a computer case is the destroyer of worlds...

... The tars build up on the components making them 'sticky' and then the dust (rather than blowing through the case) sticks in the tars causing heat buildup; as well, there would be a static situation that would occur where (on power down) the 'dusticles' , if you will , would collapse on the components creating micro shorts. On power-up the machine would die (never run again). I've even seen this [in] larger equipment before we all realized that 'cigarette smoke' was the culprit.

edit: [in] above
Last edited by MarkHaysHarris777 on Fri Apr 29, 2016 10:50 am, edited 1 time in total.
marcus

Netstar
Posts: 3
Joined: Wed Jan 27, 2016 4:14 pm

### Re: pi fan direction airflow question

I think that a heat sink is enough to keep the temperature low, if a Pi is in normal use. Because the processor is not calculating at 100% all the time, like in CPU test, it has time to cool down.

MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

### Re: pi fan direction airflow question

Netstar wrote:I think that a heat sink is enough to keep the temperature low, if a Pi is in normal use. Because the processor is not calculating at 100% all the time, like in CPU test, it has time to cool down.
Yes, usually. My Black Ice cases get warmish; but not hottish... and niether of my PI 3Bs has throttled since installing the three tiny heatsinks. Good 'ol convection is working fine. I should also add, however, that the Black Ice case may be helping out; because, there is a slot in the top of the case over the GPIO pin block (similar to a flew) that allows for 'convection' airflow through the case without a fan. My original idea was to mount the fan in a stiff paper adapter over that slot... so far, it has not been necessary.
marcus

jahboater
Posts: 7708
Joined: Wed Feb 04, 2015 6:38 pm
Location: Wonderful West Dorset

### Re: pi fan direction airflow question

Netstar wrote:I think that a heat sink is enough to keep the temperature low, if a Pi is in normal use. Because the processor is not calculating at 100% all the time, like in CPU test, it has time to cool down.
My Pi3 is held upright, outside of a case, with all three heatsinks. The heatsinks are the 25C/W ones from ModMyPi. Having it upright, with the fins aligned vertically, gives the best convection cooling. This arrangement is entirely sufficient, and avoids throttling, even for very large jobs.

I do have a fan if needed which does have a dramatic effect, but for me, its only useful for running cpuburn and other synthetic tests.

GCC 6 has just been released, so I have been compiling that on my Pi's. It is over 17 million lines of code in 104 thousand source files. Building it keeps all four cores active for over six hours. For me that's a big, but realistic workload and the heatsinks easily cope.

darkbibble
Posts: 629
Joined: Mon Mar 09, 2015 5:20 pm
Location: corby, england

### Re: pi fan direction airflow question

I didn't know I'm supposed to have 3 heatsinks, I only got a 6mm high one for the main chip from pimoroni
.
Q; How many Windows users does it take to fix a Linux problem??
A; Whats a Linux problem

RaTTuS
Posts: 10808
Joined: Tue Nov 29, 2011 11:12 am
Location: North West UK
Contact: Twitter YouTube

### Re: pi fan direction airflow question

on the PI3 only the main SOC chip gets hot enough to want a Heaksink
How To ask Questions :- http://www.catb.org/esr/faqs/smart-questions.html
WARNING - some parts of this post may be erroneous YMMV

1QC43qbL5FySu2Pi51vGqKqxy3UiJgukSX
Covfefe

blc
Posts: 465
Joined: Mon Sep 05, 2011 9:28 am

### Re: pi fan direction airflow question

stderr wrote:
MarkHaysHarris777 wrote:If the fan is in the top of the case, then evacuate the case by pulling the air out. Why? glad you asked, because you want to minimize the amount of dust in the case, and you want to pull air over the memory chip also. If you pull the air out of the case the new air will come in (mostly) through the hole or slots in the bottom of the case (minimizing dust) and will more evenly cool the entire board.
Most tower type cases blow air out of the back, most all power supplies blow air out the back of the power supply and take air in from inside the system. In this configuration, the hotter air should rise to the location of the power supply and be expelled. The cases often have holes or slits in the front in the bottom region of the case. The problem is that there is often a lot more dust closer to the floor than in the air higher up and people who place their systems on the floor tend to see a lot of dust, especially when the case fans or even just the power supply fan run all the time. Essentially they become your round the clock hoover. Once these front slots are clogged, the next place for the air to come in from used to be the floppy disks and then the CD/DVD devices, which contributed to their doom. People who smoke would probably give it up if they looked inside their computer case. Yuk.
Dust filters. Dust filters on your intake fans are a life-saver. Pop them off, give them a clean, and you're good to go. Although they won't save a PC from the effects of 20 cigarettes a day being puffed into it .

Airflow in PCs is generally pretty easy. You want clean cold air being pulled in at the front and then exhausted from the rear, using fans at both ends. It works for data centres (i.e. "hot" and "cold" aisles) and it generally works for home PCs. But then there are cases designed with "positive pressure" airflow - meaning they only have intake fans, no exhausts - and they can deliver cooling just as good as the traditional "push-pull" arrangement.

I think the difference for a Pi will be so small as to make no practical difference. If you're using a fan to keep the temperatures away from the throttling point then the key thing is just to get cold air moving over the heatsink fans. My personal preference would be a fan "pushing" air over the heatsink, since you've probably got a better chance of getting an even distribution of cold air across the heatsink fins. But then we're getting into the realms of fluid dynamics, and my physics is pretty rusty.

Koeshi
Posts: 228
Joined: Sun Mar 20, 2016 11:16 am

### Re: pi fan direction airflow question

darkbibble wrote:I didn't know I'm supposed to have 3 heatsinks, I only got a 6mm high one for the main chip from pimoroni
.
That is the same as I have, I haven't even seen heatsinks for elsewhere on the board from the more reputable dealers. I know it isn't strictly needed but I want to make a portable gaming system which of course is going to be in a case with limited airflow so I worry about heat dispersal.

MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

### Re: pi fan direction airflow question

darkbibble wrote:I didn't know I'm supposed to have 3 heatsinks, I only got a 6mm high one for the main chip from pimoroni
.

I used this heatsink set from amazon. If you order the Black Ice case from C4Labs [he] ships these same heatsinks with the case.

marcus

jahboater
Posts: 7708
Joined: Wed Feb 04, 2015 6:38 pm
Location: Wonderful West Dorset

### Re: pi fan direction airflow question

darkbibble wrote:I didn't know I'm supposed to have 3 heatsinks, I only got a 6mm high one for the main chip from pimoroni
.
You are not.

The SoC definitely needs one.

The sdram chip doesn't produce much heat, but may benefit as the refresh rate is adjusted according to the temperature. It will probably overclock better if cooler. It is normal practice for fast PC memory to have additional cooling.

The ethernet/usb chip doesn’t really need it, but ModMyPi provide a cute little heatsink for it free with the main one! On the first Pi's, thermal images used to show this chip as rather hot, but then the SoC was cool, so the settings would be different. Don’t forget the little chip may have to drive a signal down 100m of cable and control the four usb ports.

Rive
Posts: 586
Joined: Sat Mar 26, 2016 5:21 pm
Location: USA

### Re: pi fan direction airflow question

SoC on stock Pi3
Last edited by Rive on Fri Apr 29, 2016 1:00 pm, edited 1 time in total.
DNPNWO

jahboater
Posts: 7708
Joined: Wed Feb 04, 2015 6:38 pm
Location: Wonderful West Dorset

### Re: pi fan direction airflow question

http://hackaday.com/2012/06/23/checking ... pberry-pi/

Of course these pictures depend on what the chips are actually doing. On the Pi1, the SoC is covered by the memory chip and so probably appears cooler.

On the Pi3 picture there is nothing at all connected to the usb/ethernet chip.

Aydan
Posts: 742
Joined: Fri Apr 13, 2012 11:48 am
Location: Germany, near Lake Constance

### Re: pi fan direction airflow question

If I remember correctly, on the Rev1 PiB was a design error that caused the LAN chip to supply the 1.8V for the cpu, not the actual linear regulator. That's why the LAN chip could get significantly hotter than normal.

Regards
Aydan

Return to “General discussion”