Ultimate Dual Fan for Pi3

A Former User

Went up $3 on Amazon. Was $12.59, now $15.59. Free shipping. Still not unreasonable.

DD-Indeed

@livefastcyyoung

Well strange thing is that the prices have gone up with chinese sellers. Bigger retailers still have then in in reasonable prices, allthou, they are varying.

Ecks

@dankcushions Actually there is people on YouTube that does temp tests for the Pi 2 and 3 in which has shown temps going over 80

dankcushions

@ecks my mistake, it throttles at 82 degrees C. my point stands - your warranty isn't voided. there's nothing in the manual to say that it does, either.

Ecks

@dankcushions Alot of products now a days doesn't really tell you what can or can not void a warranty from a manual it would mainly tell you on the main site or a dev on twitter will tell you instead because one system i own which is the Nvidia Shield TV has no where on the system and manual on it being voided if you open the system but on the main site they tell you there.

DD-Indeed

Finally the cooler arrived, I can put it in and do some benchmarks. I will use my current cooler first for the tests, then take it off and try again and then with the dual fan cooler.

dankcushions

@ecks said in Ultimate Dual Fan for Pi3:

@dankcushions Alot of products now a days doesn't really tell you what can or can not void a warranty from a manual it would mainly tell you on the main site or a dev on twitter will tell you instead

ok, so where on the raspberry site (or raspberry dev twitter) does it say that over 80 degrees will void your warranty?

DD-Indeed

Man, this is so cute, I put a lighter next to it to give some scale :D

alt text

The double sided tape that came with it was actually thermal tape with sticky sides, so it could actually be good. But nevertheless, I'll do the tests with other things first.

spruce_m00se

i like the non standard size lighter for scale :)

DD-Indeed

@spruce_m00se

The blades themselves are roughly the size of certain coins, so they're really small. I have some pessimistic feelings about this, since it is very small, but we know more later today. :)

spruce_m00se

it will probably sound like you have a swarm of angry mosquitos inside your case aswell :)

DD-Indeed

@spruce_m00se

That's also something I'm afraid. But, I can always use the 3.3V line instead of that 5V, to make them run a bit more quieter.

spruce_m00se

@dd-indeed a bit, I tried that with one, and it drove me MAD!

spruce_m00se

@dd-indeed I actually bought a cheapo android based KODI box because the PI got so hot running KODI and the fan was annoying me too much, so 30 USD for a kodi box and it seems more responsive anyway and has an IR remote, so happy days, stuck it on a laptop cooling pad and its all gravy

DD-Indeed

Now I actually ran into pretty foolish trouble: I can't find a single benchmark test, that could be used via command line of the Retropie. Only ones I've found require OS to boot.

Edit: Nevermind, found one simple Linux test to be used with this.

DD-Indeed

Raspberry Pi 3 - Ultimate Dual Fan-cooler Tests

Overclock settings:

arm_freq=1300
gpu_freq=525
core_freq=250
sdram_freq=500
over_voltage=4
v3d_freq=500
force_turbo=1
avoid_pwm_pll=1
disable_splash=1

Commands used to do CPU tests:

vcgencmd measure_temp
sysbench --test=cpu --num-threads=8 --cpu-max-prime=50000 run

I ran the test multiple times until I reached the max temp for each setup.

Temp results:

With my original customized chipset cooler with 40 mm 5V fan = around 50 C

Without cooler = Over 80 C, thermal throttling warning visible

Ultimate Dual Fan-cooler with included thermal adhesive tape = 77 C + thermal throttling

Ultmate Dual Fan-cooler with copper plates and thermal paste = 70 C, no thermal throttling thou.

So yea, this device is piece of crap with the thermal tape, absolutely useless. And, when you apply copper plates with thermal paste, the results improve slightly.
As a ''straight-out-of-the-box-and-bolt-on-device'', it's not good, nowhere near of good, and even with improvements, it's still crap.

Conclusion: I'm not recommending this at all. Waste of money and time.

PiBoy

The dual fan test results are almost identical to what I got when I attached a fan with a heatsink to the CPU. Mine idled around 42 and topped out at 68 (with overclocking). However, I used the adhesive that came with the heatsink, which may have been that 9448 tape stuff, non conductive.

I am not convinced the SMC needs to be there, that second fan might not be doing anything.

spruce_m00se

@dd-indeed how do you run this test? I may run it on mine with the big cooler and see what happens

spruce_m00se

well i just did some heat tests,
27.5 ambient
51-52 on idle no fan
38 idle with fan
53 q3 playing with fan
67 q3 playing q3 no fan
Crashed.

from playing q3 with no fan in the past it used to give the temp warning sign after about ten mins, but this time it crashed before it got that far

DD-Indeed

@spruce_m00se

And you used stock clock settings for the Raspberry ?

Here's the test I tried:

(https://askubuntu.com/questions/634513/cpu-benchmarking-utility-for-linux)

''Actually there is a a tool named as sysbench.

You can install it with sudo apt-get install sysbench

To CPU benchmarking you can do like:

sysbench --test=cpu --cpu-max-prime=20000 run

where 20000 is like max event count.

--cpu-max-prime=20000 is optional, the default being 10000. I suggest to keep the default and fiddle with --max-requests instead (which is the number of operations performed)

Also note that by default this is a single-threaded test. For testing N cores you can use --num-threads=N (N = Any number you desire, I used 8 to properly stress the system. -DD), and compare results using the per-request statistics output.''

So, I used this command in my tests:

sysbench --test=cpu --num-threads=8 --cpu-max-prime=50000 run

And immediately after that, I checked the temps with this command:

vcgencmd measure_temp

I'm not expert on Linux commands, so maybe someone could create simple script from these, that could be easily accessed and used for these tests.