ฝัง
- เผยแพร่เมื่อ 18 ก.ย. 2019
- It is well known that the Raspberry Pi 4 gets quite hot and reduces its processing speed if it has to work hard. In one of my Quickies, I tested the passive FLIRC case. In the comments, viewers pointed me to two Chinese cooling solutions and Cristopher Bartlett from Explaining computers tested a few others.
Time for a new Quickie with a summary of all these coolers. And some ideas for optimizations.
I am a proud Patreon of GreatScott!, Electroboom, Electronoobs, and others.
Links:
Chinese Coolers: s.click.aliexpress.com/e/dTspeDEg or bit.ly/2V8ExY2 and bit.ly/2V4DYyd
60mm Fan: s.click.aliexpress.com/e/tiy0uCjW
Videos:
Explaining Computers: • Raspberry Pi 4: More A...
Temperature controlled Fan 1: • #131 Pimp my Raspberry...
Temperature Controlled Fan 2: • #180 Stop this Nagging...
Tips & Tricks: • #231 7 Tricks on how t...
Crimpers: • #282 Crimpers: IWISS I...
#Raspberry Pi #Electronics
Supporting Material and Blog Page: www.sensorsiot.org
Github: www.github.com/sensorsiot
My Patreon Page: / andreasspiess
My Bitcoin address: 19FSmqbBzb5zsYB1d8Bq4KbxVmezToDNTV
If you want to support the channel, please use the links below to start your shopping. No additional charges for you, but I get a commission (of your purchases the next 24 hours) to buy new stuff for the channel
For Banggood bit.ly/2jAQEf4
For AliExpress: bit.ly/2B0yTLL
For ebay.com: ebay.to/2DuYXBp
profile.php?...
/ spiessa
www.instructables.com/member/...
Please do not try to Email me or invite me on LinkedIn. These communication channels are reserved for my primary job - วิทยาศาสตร์และเทคโนโลยี
Fantastic review/addition to Christophers test
Thanks for sharing your results :-)
You are welcome!
Thank you, Andreas 🙏 I ‘ m very happy you did this video. I was looking for some good Case for my P4 and this info looks great
that fan on a screw was so much better than it had any right to :D
I started with the big heatsink in between but found out, that its effect was minimal.
That's the magic with bigger fans, they can spin at lower speeds while being quieter and moving more air than the smaller coolers. And in my experience cheap small fans don't last as long as cheap bigger fans.
Cheap and useful ! Got to love the Flirc for style and functionality !!
Proof that KISS works. I wish this video had come out before I spent twice this on some fan shims.
Wonder if it would perform better than the ice tower with one of those small adhesive heatsinks on the CPU.
I’m simply thrilled that you took Christoffers video and build on top of his work.
Good that you pointed out that in most cases a RPi only runs under full load for seconds.
I’ve watched your video’s for a long time now and I just want to say that you should be very proud of what you have achieved!
I hope you let Misses Spiess read this.
Thank you for your nice words! She read it!
Thanks from Brazil for all information that you share with us Andreas! I really enjoy your videos. Keep up the good work.
I will keep it up!
Yes, another very helpful test. I liked the fact that you extended Chris's work
I think I will be going with the Flirc case. As well as the issues you identified, fans also use energy and for me one of the attractions of the Pi is it's low power consumption.
The Flirc is around my main Pi on in the lab, too.
I would've loved to see the ICE Tower low-profile cooler as well, great video also!
I am not sure it existed back then.
thank you for this short informative video, i have been looking for something like this for a long time.
You are welcome!
This video is a jewel. I subbed; thank you !
Welcome aboard the channel!
I use the FLIRC case like you recommended and i am still happy with it. Looks very nice and do it`s job.
Thank you for sharing your experience!
I'm leaning towards it as well. Has it ever throttled? Is it too hot to hold in your hand?
@@mplovecraft throttled? nope, not till now.
too hot to hold? it depends on your personal hands - 60°C is not nice for a longer time, but do not make any harm and it's ok to put it from place A to place B
@@mplovecraft compiling Kernel 5.3 ~ 1h (98-100%) max temp is 62°C
@@dl8cy Thank you! Ordered one now.
Great job, nice to know that just adding a fan will keep me on the road till I can get a permanent solution. :)
It looks like the heatsinks do not help a lot if you have a fan close to the chips
Andreas, vielen Dank für deine immer informativen Videos. Im Bereich Raspberry Pi lande ich grundsätzlich bei dir. Grüße aus Deutschland und weiter so!
Vielen Dank. Ich habe ja noch ein paar andere interessante Themen...
Just so happened that I got my Flirc case + first Pi 4 in the mail today. As you say, it'll mostly sit there being an IoT server doing very little most of the time, and the Flirc case looks really good too.
Same here :-)
It looks good. But the underside with the screws gives off an unpleasant plastic smell when heated by the CPU. Do you have the same problem?
Thanks again Andreas. I have couple closed cases, and couple open cases, like you are showing, where the fan blows air directly on two heatsinks and air can go everywhere, since there are no obstructions. On one I use large LCD display and second is for project I'm playing with. The enclosure 1 I have with heatsinks is plastic and it keeps everything cool - using it for 3D printer. The second is printed by me and also has fan and heatsinks but it is much larger case that is also a support for project that I was working on. For some reason, I don't overwork my PI's much, so they stay cool. But Andreas sometimes makes projects that would melt my plastic case - ha ha ha
Thank you for sharing your experience! Only the Pi4 gets a little too hot for me without cooling under normal load. The others were ok for me. This is also why I do not advocate these high-end tower coolers.
@@AndreasSpiess Thanks for your input Andreas!
i was just about to get another ice tower, but i agree that the FLIRC case looks great.
Most useful, Many thanks!!
You are welcome!
I recently used a "chinese passive" heat sink onto which I screwed two 30 mm fans @ 3.3 V (the screws fit the gaps in the heatsink.
It runs at about 40°C and barely gets hotter under load.
Since I tested on Ubuntu 20.04LTS which has a newer version of sysbench (1.0.x vs 0.4.2 for Raspberry Pi) which is a lot more efficient I wrote my little script using bits and pieces from the internet.
The calculation takes about 4.5 min and I get impressive results (test run in that order after one another):
1. Passive + 2x 30 mm fans @ 3.3 V: 37.2 -> 40.2°C
2. Passive + 1x 30 mm fan @3.3 V: 37.9 -> 42.8°C
3. Passive only: 38.6 -> 47.5°C
i ran a longer test (~85 min) and the pi with the 2 fans configuration heated up by less than 2°C, definitely a good and quiet solution.
I wonder what would the results be if you screwed or fixed a large fan on top of a flirc case?
I also have an icetower for another and it is quite amazing (Pi running at 32°C with fan @3.3 V while playing videos) but nowhere near as compact as setup # 1.
Two years ago I made myself a simplistic addon-board that connects to the 40-pin GPIO-header and which allows me to easily use either a regular 5V DC-fan or a 5V PWM-fan and controlling its speed from the PWM-pin on the GPIO-header. With a passthrough-40-pin header on it, I can still access and use all the GPIO-pins and power-pins as usual as well.
Never actually ended up using it, though, because I came to the conclusion that passive heatsinks are more than good enough.
Sometimes the way is the goal and we learn something even if we do not use the product afterward...
Such a useful video! Thank you very much!
Glad it was helpful!
I just got my Pi4 with 4 GB RAM and the passive heatsink case. I noticed that the case does not touch the components correctly, so I used 2 instead of 1 cooling pads for the contact areas...and it works fantastic! Watching a 2 hour movie on 1080p results in just 54°C. @Andreas: maybe retest this, because a passive case is just sooo much nicer for longevity and noise :-)
Thank you for sharing your experience. One of mine runs in a Flirc case currently at 49 degrees and the other in the passive Chinese heatsink at 54 degrees. But without a lot of load. I added genuine 3M heat transfer tape and ordered thin copper sheets. If they come, maybe I will add one or two th get a better pressure.
Cheapo fan it is! Thank you so much!
You are welcome!
Thanks Andreas, have you seen the Argon ONE case? I would be interested in an objective (measured) comparison of the noise levels of the Argon ONE and the other solutions, I'm prefer silent cooling where feasible.
I think I had it in one of my mailbags
I saw metal heatsink in the preview picture and my brain yelled...WiFI!!! lol
:-)
Thanks man
:-)
Very informative video. Should have watched this first before buying my first Pi. Fortunately, I have the Chinese aluminum case with dual fans.
They are ok, but maybe not always needed. I mainly use fanless cases because my Raspberries are not working hard...
I have owned several Rpi 4 cases. For style and functionality, also comes with a higher price tag, nothing beats the Argon One case. It does actually has a power button built-in. My personal favourite is the Geekwork armor case with dual fans which allows a safe OC to 2.1Ghz for the Rpi 4 8Gb while maintain a temperature under 42 degrees Celsius under heavy load. I also have an Geekpi acrylic case with fan which is rather adequate and cheap but noisy as well. It comes with AC power adapter with a power switch.
Thank you for sharing your experience! I also liked the Argon case.
The Argob One case has an active fan and I completely agree with the sentiment that fans are not the right thing for hardware that should run for years untouched.
Currently using the Ice Tower on a Pi 3B for the Astroberry application. I had to modify the Ice Tower (I did not like the idea of using 3 layers of the heat transfer film) to fit the 3B by taking a copper penny and sanding it smooth on both sides to place between the CPU and heat sink with some thermal compound. Also slightly bent the mounting leg to increase the downforce on the CPU. Mounted the assembly in a 3D printed case found on Thingiverse that I modified to increase the thickness of the case sides. At idle it is steady at 27°C in a room at 20°C. Never goes over 52°C when running intensive applications (such as are possible with a Pi 3B). Very please with the Ice Tower overall.
Thank you for sharing your experience!
I have recreated these tests for normal uses of the Pi 4B. The Argon One case came out the very best.
The Argon also has some other nice features...
I made very good experiences with passive heatsinks glued onto the processor with a very thin layer of thermal glue (used for power leds and IC heatsinks) instead of using one of these thick removable thermal pads. The heat conduction is better with the thermal glue since you only need a very thin layer. It also secures the heatsink much better then the cheap glue pads, i had these fall off on occasion.
You are right, this is the best solution if you do not need to bridge any distances. Similar material is also used for PC coolers.
I was able to use a regular 40pin cable with the FLIRC case after shaving off some plastik on either end of the 40pin connector. I hope they correct the issue wither later cases.
I am not sure it will be easy for them as the holes on the RPI are quite close to the pins. Anyway, you did the right thing.
Very interesting!
Thank you!
Thanks!
You are welcome!
That HUGE one Chris tested was very very promising but sheesh, it was just a miniaturized Personal Computer HSF combo. You know if this keeps up, as it has been, the SOC will lose one of its reasons of being as it was to be smaller and able to fit in a pocket so a full fledged computer in your pocket. Pretty hard to do that with a HUGE HSF like the one Chris showed.
It's OK to consider that huge one if you are in the States - I'm in the UK and couldn't find one being offered by a national retailer for retail purchase - perhaps they will emerge in time ! Fanless devices have always been so pleasant to work with - the 1970s/1980s X terminals, AMD Duron processors which were so slow that they didn't need one, early Intel Pentiums which would keep going irrespective of whether or not the fans worked, thin clients, and some ITX boards. Being deafened by a fan is indeed an annoyance - I second your posting !
@Phil Dodd Oh, I like it and more will be made as China loves to steal IP and offer everything BUT my point was when the next Pi comes out will it be even hotter and demand even a larger HSF combo?
I really would love to go fanless, a dream of mine for 20 years now, but it is getting crazier and crazier out there. As we see air has to be moved on even the Ice or it takes a substantial hit. Fanless for less dust, less noise, and I hate the sound of my fans revving up and down as the temp on the CPU goes up in my current PC. This is why SOC was a diamond because the marketing hype was to have all of the power of a PC without the need for active cooling.
I do know what the issue is and that is silicon because we really need a new substrate material as we are hitting the manufacturing limits for PC and still 95-135w TDP on them. I think in 3, or 4, generations we will not be able to go smaller as the manufacturing process has its limits AND electricity will jump over the internal traces so there is a limit. At first Intel was saying their goal was to have thousands of cores so each function had its own core (I am sure that article is still on the net somewhere) but it didn't take too long to realize more cores means more heat and after a certain level you no longer receive any gains and actually slow down.
I think all of this is coming to head, finally, to force us forward or to stay right at the brick wall and we are close to that brick wall now. Shoot, a computer made a decade ago run most software just fine (a few YT videos proving it) so all of this advancement hasn't really advanced us much as far as the CPU goes.
@@thebeststooge Indeed it will, and that will be deemed as "progress" ! Chris said as much on his channel when he talked about the massive heatsink, that it is the way he envisages SBCs going - a continuation ( in spirit ) of Intel's "Moore's Law" ( as you imply )? There always is a tinkering with the balance between centralised and distributed processing - maybe a massive heatsink with a tiny board attached on the top of it has gone a bit too far one way ?
@Phil Dodd Well, we shall see but, as I mentioned, Silicon needs to be replaced and about a decade ago I read where scientists at Intel, and IBM, were working on a replacement but then it went quiet.
@@thebeststooge Now THAT'S interesting ! Perhaps they hit upon the "heat limitations" ? We're told too that the limitations are being reached on our current generation of materials - solar panels overheat in the sun, and that a more efficient material than lithium is needed to improve batteries for electric vehicles, not to mention new materials needed soon to fly people to Mars... We hear mumblings along the lines of "carbon nanotubes" and that type of area - where will the research restart ? We're told that AI requires a step up in raw power every time that it is to make a significant advance, so maybe that will be a driving force in computing power ?
I actually use a 80mm fan pushing air from above, pretty much like your cheap solution, and I find this to be about the best when it comes to price tag compared to effect. The only real difference in my setup is, that I use a 12v fan system for my RPi cluster, since this provides greater efficiency for the fan motors and is very silent, since the rpm of the fans is still very low. In fact I cannot hear the fans themself, but the air they blow, when I get very close with my ears.
Good point. I did not want to have an extra PS.
Will the FLIRC case accommodate any kind of HAT. I want a RTC if possible? Also, you forgot to mention the ambient temperatures. Switzerland is pretty cool. Some of us live with no Air conditioning in much hotter climates?
I do not think a hat will fit. I assume you will also not have any problems if your ambient temperature is 35 degrees. Of course in bright sun it will collapse (also here)
i have noticed the big passive heatsink does not sit nowhere near flush to the cpu. Even with the heat pad stickers installed, it does not seem like the cpu is transfering enough heat out to the heatsink. Essentially the big, passive heatsink should be the best solution, and even better with a small fan attached. Its just flawed by a mis-measurement of the distance between the sink and the cpu's surface. I am going to try to shim a small slice of copper or alluminum in the space in between and see if it changes anything. The heat transfer should improve. sorry for the misspelling. Then i will put one of my old AMD cpu fans on top of the sink.
I have the big black heastsink with 2 fans. The temperature drops down to 50'C which is quite good enough for general usages.
I had it also running for a test. Same experience here for my IOT server
Are you selling the screw and fan option? No doubt this is a precision Swiss screw to give such good results?
Definitively. This is why we protect the "IP" and only use if for us ;-) Maybe we share with you after Brexit!
I have a case from that chinese manufacturer (not the heatsink case btw, it's made out of acrylic and has 2x 20mm fans at 5V)
and it's design looks cool, but isn't cool when it comes to cooling:
it presses air inside, but there is only little way for it to get out (almost nothing)
Thanks for sharing your experience.
I already have the chinese metal heatsink-case with no fan. I think I will replace the thermal pad of the CPU with a better one, or copper with thermal paste, then use a fan that can be toggled with a temperature sensor on the heatsink, as suggested later in the video. I did not expected the ICE Tower to be that good, because I thought it was just a shiny gadget and not an actually efficient solution. Thanks for all the testing.
You are welcome! I do not like fans and am happy it is not needed in my setup.
The control of the fan does not require an external temperature sensor, the Pi knows it's CPU temperature already.
Nor does it require a python script like Pimoroni provide.
Everything is already available as an overlay, gpio-fan that you can configure with high and low threshold or similar concept.
The PoE Hat from the foundation works that way, out of the box, without configuration.
It is actually a good solution for the Pi4 except that it block the GPIO.
Interesting. Do you have a Link to the description?
@Shadan Rikan Thank you. But it does not say that the function is already built-in as David mentions in his comment.
@@AndreasSpiess I think the "built-in" referred to is the temp sensor within the pi. That way an external temp sensor is not required but, as stated, the downside is the need to use a GPIO pin so that the script can switch the fan on.
The other down side to this "overlay" method is that it requires the Pi to be functioning properly to activate the fan. If the script process fails when the GPIO is off then the fan will never turn on. The simple external sensor method will work regardless. It's a softwarehardware control conundrum.
@@havenview Embedded should refer to both the fact that the sensor is in your CPU and that the software control is in Raspbian. I believe that the PoE HAT enable it by having an eeprom on the reserved I2C that contain the overlay description (?).
So the hardware and software should be there, just add the fan on a GPIO.
You can replace the eeprom by software enabling it in config.txt.
And maybe that is better than a Python script?
I inserted disable_poe_fan=1 or disable_poe_fan=0. Both witout any change on GPIO 12 (newest Buster)
Surprised the Pimoroni fan shim turned out so bad. The fan on a screw is ingenious. I bought a fan shim with pimoroni case for one pi 4, also bought a similar case with built in fan for another. Both seem equal, but second half the price.
The shim fans are smaller. Maybe this is the reason for the difference.
Hi have a pi4 Heatsink case with dual fans and watched your video No.180 using a KSD9700 heat sensor switch. Can you please let me know which is the best way to mount the heat switch on the case, I take it that it needs insulation around switch as you vid explains to wire the switch between the cut positive red cable, sorry Im no electrican 😁 Thanks again.
The heat sink has to be where you expect the highest temperature. You also get such switches in plastic and ceramic cases which are insolated.
Does the Pi4 have the pads to add an antenna connector like the 3 has? If so the external antenna could mitigate the only problem I see with the Flirc.
I do not see an obvious place for an IPX connector and I did not find information on the net. There is a solder point which seems to be the antenna output. But before somebody has more info I would not try it.
Those passive Chinese "Armour Heatsink Cases" are known for their big gaps between the heatsink and the chips. The thermal pads sometimes can't even touch them.
I experimented with a *very* thick blob of thermal paste (MX-4 in my case). As a result, at full stress of all four cores I only got 53 °C after 30 minutes (no overclocking).
I'm curious how these values will be when the thermal paste starts drying, though...
BTW: That case get's *hot*. And the underside as well. Perhaps the screws work as a "heatpipe"?
Maybe a simple piece of copper will do it, too...
@@AndreasSpiess I think so. But: Woher nehmen, wenn nicht stehlen? 😉
Hornbach
Your Swiss accent is nice. :)
Since this video came the Argon ONE Pi 4 and Argon ONE Neo.
That is the "problem" of electronics. Very short product life cycles...
Is it actually safe to power inductive loads directly from the GPIOs?
I mean when the fan are starting they make a high curent draw and when stopped it’s usually recommended to use a flyback diod?
dezmob I don’t think it is safe because the GPIO pins aren’t designed to handle the current required. You want to use the 5V (or 3.3V) power pins, not the GPIO pins. If you want to use the GPIO, for example to PWM, you can use a transistor or MOSFET for the switching and use the 5V/3.3V for power.
Also isn’t there any fuses on the 3.3V rail ? I’ve read somewhere that the Pi 3.3 and 5V rail are already close to they max amp rating, so adding another loads on those se rails could be harmful?
It is not about the current. Coils can produce high voltages which easily destroy an MCU. That is the main reason for me. Flyback diodes are not good because we want to have a RF carrier without a lot of harmonics.
Wow...ok...I cut a piece of northbridge heatsink from an old mainboard, and used this piece as a shim between the heatsink and cpu, as I had planned. Huge difference with the heat transfer. Even the bottom plate was hot below the Pi. This little computer runs super toasty. I will definitely need a fan to cool this heatsink.
Good idea!
the 60mm fan on a bolt is genius
Thank you!
Right now my cooling options is an ice tower or chinese passive + fan (5 cm or maybe 6 cm), the chinese passive + fan would be cheaper, and probably more quiet. (And ~45% cheaper.)
But the wifi interference might be a problem, i haven't tested yet, but when i first used the raspberry pi 4 the wifi was not working. (I used a aluminum heatsink case, identical to one made by DFRobot)
Then suddenly it worked, the final OS i used is Ubuntu desktop now Lubuntu, i am planning on installing Lubuntu LTS through the Ubuntu Server LTS ( i tried at first but it didn't work) to an SSD.
Staring at the image in the online store i can see the part that contacts the CPU is quite bulky.
I only use passive cooling for the Pi4 here and it works ok. The Wi-Fi does not need to travel too far because the next AP is 3 meters away (plus one wall).
The Flirc and the chinese passive should, I believe, give very comparable results. They're both just big chunks of Al. The only reasons I can think of for a 12 degree difference in performance are 1) the flirc case maybe has more surface area to radiate heat, and/or 2) for some reason the heat transfer from chips to case for the C.P. was not as good. It seems to me there's some investigation to be done there.
It is well possible that there was a heat transfer problem. I did not investigate. I just used the same my viewers get when they purchase the product. The pads, however were the same for both Chinese products. I should have used my flirc camera. Maybe it would have pointed us in a direction.
@@AndreasSpiess I would think a little as Richard do !
I'm surprised the Chinese Passive doesn't do better with its fins.
I wonder the metal weight of each ones ( FLIRC and the Chinese Passive) ? I'm pretty sure the CP weights at least as much as the FLIRC. So with the fins, it should perform at least the same, if not better !
Andreas could maybe tell us about their weight ?
Did You use the thermal pads included on the C.P.? Also You Will get better results with thermal compound.
I tried now the passive with paste. The result: 69 degrees in the end. The same as with the pads. Maybe the heatsink does not touch the MCU. I cannot test that easily.
@@AndreasSpiess The heatsink is not touching by default (a lot of reviews suggested on amazon for example). If you shine throu, you can see the gap even on RAM and CPU. Please try with a little bit more paste and post your results :)
I LOVE THIS GUY!
:-)
Does a peltier cooler work well for cooling this, or arm based sbc's? I really want to use a tec for an sbc but idk how to determine the wattage it can cool
A peltier element consumes a lot of energy. So I never considered it as a good solution.
@@AndreasSpiess wow, I just looked it up and apparently they're only 10%-15% efficient. That's not good at all, for low power applications like an sbc with a tdp of 10watts or less it would definitely still be worth it to only use a fan and heatsink, sub ambient temperature probably isn't a huge benefit that justifies the power consumption
Thanks for your input
Hi, I am thinking about creating my own nas with this rp4 and maybe this case. Is there any way to attach an internal SSD/HDD to it without using the usb ports? Cheers and thank you!
You will need the "blue" USB port to attach an SSD/HDD. Or you buy some quite expensive hats to attach SATA drives. But I do not know them
🤔We will get one or more mini heatsink for the powered chips? In infrared camera it shows the power chips heats as much as the cpu.
If the heatsink has the same temperature as the CPU it works!
What about the Argon M.2 case? It has passive and active cooling.
I do not know it. Many new cases appeared after I made this video.
The FLIRC looks nice. Would you say that these are powerful enough to use as mini desktop computers?
I doubt it if you compare it with a PC. But it depends a lot on what you want to do. It has nothing to do with the case, only with the Raspberry.
@@AndreasSpiess Thanks Andreas. Yep I know the case has nothing to do with it. I'm wanting to find a micro pc for my TV that is not a ludicrous price.
Hi Andreas, I manufacture an all aluminum passive cooling case. I would love to see you review it and would be happy to send you one. I think it can beat all of these.
All my viewers can send me stuff for my mailbag. But usually I do not test it. If you provide your own comparison I for sure can show it.
Just wondering if your cheap fan would have had a better result with say a small plastic or cardboard cylinder directing the flow from the cpu. Unless the direction of flow was downwards towards the cpu.
Maybe you could help with a little air "guidance". Just try it.
Any ideia of the life span of the Ice Tower cooler using It 24/7? Thanks in advance
No idea. I do not like fans, also because only very expensive ones have a long life
@@AndreasSpiess i see... Thanks!!
that chinese passive solution + that 60mm fan, how would that turn out?
I tried it, but it had nearly no effect.
@@AndreasSpiess thanks for letting us know this. maybe put it in video description? :)
@@AndreasSpiess Hi Andreas. Concerning this Chinese Passive solution :
You definitely need to investigate heat transfer between CPU and case. It's not possible that adding a fan on such a big heat sink doesn't make big difference ! Somebody already stated previously in that direction !
I tried now replacing the pad with the paste: Exactly the same result. Maybe the two metals do not touch. Not easy to check.
@@AndreasSpiess Andreas. I bought this fanless chinese:
www.ebay.fr/itm/For-Raspberry-Pi-4B-Protect-Box-CNC-Aluminum-Alloy-Case-Enclosure-w-Cooling-Fan/153604726380?ssPageName=STRK%3AMEBIDX%3AIT&var=453625756727&_trksid=p2057872.m2749.l2649
As you can see on the 9th photo, there are metal "bumps' on each side of the case.
Is yours the same ?
I've choosen that one because it is supposed to take the CPU from both side, l(ike in between 2 fingers). I thought that solution was good because I tried myself to fit a heatsink on the Pi and it was difficult to tightly make contact with CPU/heatsink. I tried to make some kind of braces, but it was bending the entire PCB because, pushing on the CPU, the braces were attached on the side of the PCB. It was mechanically a bad solution despite being thermally quite efficient.
So, I was quite confident that this passive chinese cooler was a good solution.
I haven't received mine yet. You might confirm that this passive chinese tighten the CPU from cpu side AND pcb side when the case is closed. There might also be 2 blue tack pad (1 on each side).
I hope I'm clear enough.
I stuck a heat-sink (55x85x12) that I collected on an old graphic card on top of my Flirc (instead of the black plastic top) with thermal paste and that shaved an extra 4 degrees at full throttle (with sysbench as described above). Easy solution, take a good case (I wanted a case, and not an open solution) and give it a little boost for free (almost).
Thank you for sharing your experience. Seems to work!
Ur the best andreas. Super guet!
Bought the iwiss and engineer. Andreas rules
:-)
would have thought a heat-sink on the controllers and other important parts then a fan on top ?
With a fan, the heatsink had no big effect. This is why I left it out.
I cant fully get my head around on why the big Chinese aluminium blocks are performing so bad. Did u use a decent cooling paste with them?
No, I used the same thermal patch which came with one of the two heatsinks.
@@AndreasSpiess Ah, that explains the bad performance. According to my tests, this tape is as "good" as chewing gum ;)
Andreas when you put it on Facebook, it only played a minute or less. Very straight. From the server fan guy.
I always do it the same way. No idea why it did not work.
I've would have liked a no cooling "control" sample as well and all of these with cpu MHZ. Otherwise, still a great test 4 years later.
I never run a Pi4 without cooling as it approaches 80 degrees in no time...
@@AndreasSpiess Oh, then now i know why my pi 3 is slow in no time ( I assume it requires cooling as well)
@@Doile911 The Pi3 is always slow ;-) I never cooled a Pi3. That only started with the Pi4
OK! My Passive Armor case for 4b (not the universal one) come today... On stress dont pads from 54C... I put a Steel flat and thermal paste o the CPU instead the thermal pad that come with it...
Thank you for sharing your results!
so what type of usage intensity is retropie?
I never used it, so I don’t know.
Welchen Noctua Lüfter hast du am ice tower? Hast du den auf zwei pin selbst umgebaut? Kann der RPI pwm ansteuern?
Ich habe gar keinen Ice Tower :-( Wie gesagt habe ich die Daten von Christopher genommen und mit meinen eigenen Lösungen verglichen.
i had to go with the argon one, mad i bought one before i saw the m.2 version lol
In technology this will always happen, I fear...
Which pins does the fan operate on?
Usually on 5 V or 3.3V is you want it slower
2:10 is 'chinese with fan' the one at 1:36 ?
Yes. There is also a comment where somebody used a little copper between the CPU and the heatsink instead the pad with good results.
Thank you so much to both yourself and Chris for covering this situation so fully. Morally, the Pi Foundation should be reimbursing you both for all of the research and testing that you have done - either financially or with a few dozen Pi SBCs each ! THANK YOU to both of you !
I am not sure the love me after my rant video when the Pi4 came out ;-)
I replace "which is best" in my use cases for "which is good enough". In my use occasional throttling is ok. I rarely do long high power runs anyway. If you ran very high loads and really need full power, you may like more cooling that suits me.
Too many videos try to overcool like getting cpu from 50C down to 40C or even less. I see no point to cool more when it already is not throttling.
I hate noisy fans, especially the small ones with high pitch. In some cases like having one in a bedroom they definately need to be passive! No failing fans either with passive cooling.
As a rule of thumb, every additional 10°C above room temperature halves the life expectancy of components. So in some cases that extra reduction to 40°C might be worthwhile. But for most people's use, you're correct. We run a very quiet computer in the bedroom, 24/7/365, and it contains large fans with continuously variable speeds, and which only speed up under heavy load. This makes it essentially inaudible when running normally. A fan turning on to full speed and off again, repeatedly, is very annoying.
So "best" for you is "good enough". I hope you find all the information to do this decision.
@@RWBHere Life expectancy is a very good point. I wonder what is life expectancy for a component like RPi cpu? Does it really matter? I have never had any cpu fail on me before the device is retired anyway. That may be because most of my projects mostly idle and do the heavy stuff quite rarely and momentarily.
I've personally had mostly aged capasitors and misused or severly undercooled power devices fail when run out of specified region.
Gruetzi! Love the accent :)
Thanks! 😃
The Flirc looks like a good option for passive cooling, but I wouldn't be able to sit an SSD on top of the case! :)
The case gets also quite hot. This is why I have the SSD beneath.
imma combine hte passive case cooler with the 60mm fan or one of the 100 random old pc fans i have laying arround to overclock the pi to 2.4ghz
Cool!
I've just bought the chinese case with the two fans. I've still to install it, but it seems the sticky pads supplied are not 3M thermally conductive, but are just ordinary double-sided sticky foam pads. My plan is to get the proper stuff, see if that works better.
Try to replace them with thermal paste...
Try to replace the thermal pads with thermal paste
@@PULSTARUnboxing Already ordered, copper shim to take up most of the gap, plus thermal paste.
What size Shim have?
@@PULSTARUnboxing I've ordered 0.5mm copper. Hopefully that will work.
Hallo Andreas, auf meiner qrl musste ich mich mit einem LVDS Signal befassen, um ein Display anzusteuern. Wenn man darüber Informationen hat, könnte man eventuell funktionierende Displays aus dem Müll retten? Gruß aus Lübeck, danke für deine Arbeit!
Die heutigen Geräte sind oft wertlos wenn man nicht eine gute Doku hat. Und die wird leider
nicht von allen Herstellern herausgegeben...
Maybe the flirc case is so massive that it takes time for it to heat up so as long as it heats up it is going to show low temperature, however it may still plateau at a high temperature. Maybe you should let the flirc case run for hours to see at what temp it plateaus.
I do not think so. I tried once double the time and it did not change. The case is not heavy
The FLIRC has enough surface area that it is able to dissipate the heat generated by the Pi quicker than the Pi can generate it. It conducts the heat into the air well because it is aluminum. Have you ever touched aluminum foil shortly after taking it out of an oven?
Today I've noticed one interesting thing about "Chinese passive" case. I have two rpi4 and two identical SSDs, first rpi has acrylic case with fan, second one uses Chinese passive case. When I run RPI+SSD+acrylic case in USB 3.0 all OK. When RPI in Chinese passive case uses USB 3.0 for SSD - wifi stopped working, RPI is not connecting to the router anymore. When I've changed USB port from 3.0 to 2.0 - wifi works. My solution was to use external wifi module, because router does not support 5 Ghz and I need full speed of USB 3.0 SSD. I suppose you know that USB 3.0 interfer Wifi 2.4Ghz. I think Chinese passive case makes this interference worse(or the signal is weak due metal case and even small distortion from USB 3.0 affects wifi connection)
Thank you for sharing. So far, I have not had this problem (I use 5GHz or Ethernet on my Pi "server"). It seems that the metal case shields the Wi-Fi signal which makes it more susceptible to EMI from USB.
did you test with overclock?
No.
@@AndreasSpiess can you make videos testing with overclock please?
I never overclock my machines. I have no need for that. So I have no knowhow to share :-(
If you overclock to 2ghz, what is the best solution then? Sorry for my bad English.
I never overclock my devices, so I do not know.
I got the case with no fan but my bluetooth controller disconnects.
Can this case block the signal?
Yes, it can weaken the signal because it is of metal.
Mhm...
Can I boost it back if I get a bluetootheye with usb?
You always can add a BLE dongle if you want.
after watching so many pi4 videos now i am lusting for it. but i don't know what to do with it! with the metal case, there is one more problem. at least at my home. if the earthing is not proper and the ac to usb c adapter is not properly built, +/- 5V sits halfway between 230V AC which makes the metal touch "Electrifying"
Everything has advantages and disadvantages. Here we have a protection circuit for that mandatory in all houses. So you learn your lesson, but you are not killed.
BTW: I will come to Hyderabad in October. You do not live there, by chance?
Adapter business is tricky. i do not have RCCB (residual current circuit breaker or whatever it is called) I think its the right time to install it. Looks like earthing wire is broken somewhere. By the way i got a "FULL 230v" shock last week. i switched off the mcb (they follow german system, on is top position off is bottom position but our switches are the opposite from the British, bottom is on top is off). neutral was connected, at least i thought so. but someone had reversed the live and neutral wires. luckily i had shoes and foam below. so i was saved. but it was the "SHOCK OF MY LIFE" And i learned the lesson, never hold anything naked even if it is too tempting.
no i do not live there at Hyderabad. Most of the IT industry is located there, other than at Bangalore which was the hifi science hub (including defense, nuclear and space tech mostly government controlled). Unfortunately we do not have successful Electronics industry. these are mostly imported. Some of electronics is manufactured. These industries are offshoots from their ancient counterparts. i live 2 hours away (by flight).
By the way, October-December is the best time to visit india. Weather is good. but its fairly large and diverse. Both weather and people along with their thinking varies greatly. We have a popular saying, our dialects change every 5 kilometers. Unfortunately i do not have any connections in hyderabad. We have several festivals in october, so most likely i will go for holidays with family but not sure about the dates. In anycase, if you have time, i can show around my place. Most welcome :) But we don't have much of science here. there were trees, now a lot of concrete.
I was asked to join www.thethingsconference.in as a speaker. So, maybe next time. Good to know that this is a good time to come to India!
regarding weather, i live in northern part of country. so thats the good weather months for us. this year in the rainy season, july-mid september, there were no rains in august but heavy and very heavy rains in september end. its that unpredictable! most of our cities are a mess, and with government's policy of "development" huge population is migrating to cities especially after 1990s. so if a road is designed for 1000 vehicles, there are at least 50,000. Its beyond control. we don't think design. We just copy from others and adopt whatever works!! Designing something for local conditions is not even a consideration its too complicated and costly for Return on Investment.They installed synchronized traffic lights in the cities but no one knows how to operate them. As a result there is huge confusion and traffic police is the happiest of all. They can now focus on doing what they do the best, issue tickets :-) If i were to be a decision maker at such new things i would have wanted to keep things as simple as it can be.. especially with complicated systems with so many components depending on so many external factors. In general, our idea of division of work is to blame the other group for all the shortcomings!!
Anyway, its good that they are now going global. Anyway, our economy is haywire now. Lets see how LoRa is used.
Cue me getting an active cooler with a huge heatsink to run RetroPie
Hey Andreas. Ich brauch mal deinen Rat. Ich möchte mir eine Retropie aufbauen und weis nicht wie hoch die Belastung für die CPU ist. Reicht ein Passivkühler wenn ich ein paar stunden damit zocken will oder sollte er in diesem Fall einen Lüfter besitzen? MfG aus Deutschland
Der Pi4 braucht auf jeden Fall einen Kühler. Ob passiv oder aktiv ist nicht so wichtig wie verschiedene Tests beweisen
Thank you from Bangladesh 🇧🇩
Greetings to Bangladesh!
The Raspberry Pi itself has a built in temperature sensor so no need for an external. I have found the best of all worlds is the Chinese heat sink with the fan. I then use a transistor+resistor to let the Raspberry Pi itself turn the fan on when it goes over 50. The passive cooling is good enough that with my heavy use the fans only ever kick on a few times a week. This saves the bearings in the fans to not have them running when not needed. Plus the power usage with the fans on is very low. Best of all worlds. Here is a pic of the setup: www.whiteshepherd.net/pictures/electronics/8gbserver/1.jpg
Good idea. I once made a similar video and used PWM to steer the fan speed according the temperature (Pimp my Raspberry)
What about 50mm fans.
With/without cheapo heatsinks
There are aluminium cases that Fit 50mm fans with minor modifications {the only mod you need to make is drilling some of the holes bigger}
My pi 2B runs at 28-30C while running Pihole with cheapo heatsinks+ coolermaster 12v 50mm fan {fan runs at 5V}
Without fan the temps are around 38-40C {with some offshoots of 45C}
I do not particularly like a fan if I can use a passive cooling. But for sure fans are better in removing heat.
@@AndreasSpiess if you use a low DB 12V fan on 5V it is not even audible.
When i get the items i ordered i will test it using the same method and see how much Difference it makes in temperature.
If i can already get a 10+ degrees difference, with cheapo heatsinks, a larger one should get me to near ambient {ambient temperature is currently around 20-22 degrees}
@@AndreasSpiess Just gotten my passive china heatsinks for the Raspberry Pi 4.
With very very very Basic heatsinks it runs at about 60C.
The china passive+ 50mm Fans runs around 35-40C. {ambient temperature is around 28C} {12V 50mm Fan running at 5V}
So about 20-25C cooler thanwithout Fans/ultra cheapo basic fans
I have not stress-tested it yet, so i do not know how much temperature difference i can get with a TORTURE test.
I did notice that the Position you place the Fan has effect on its Cooling preformance.
I just bought last night: www.amazon.ca/gp/product/B07C9H9LJN/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1
I implemented on my other raspberries the PWM control reusing your code to minimize noise, dust, wear, etc.. I only added to the code MQTT so I can have a dashboard on Homeassistant with all the values. Also added the shutdown button you show in one of your videos to turn on/off my RPs.
Works like a charm. Thanks Andreas!
So you really "pimped" your Raspberry!
...of course maybe some Arctic silver between the sink, shim, and Pi cpu....can't forget that.
You are right!
Any chance you can replace chinese fanless pad with copper or metal. Run the test again, the pad is make it worst.
For the moment I do not plan other tests.
The PoE hat has an integral fan which only comes on when the temperature rises.
a good solution because otherwise it would be hard to cool the CPU with a hat
I just use a couple of copper coins with thermal paste and a small fan I had laying around and it works well lul
I do not like fans... But good to know.
I was happy to get the FLIRC case yesterday. It cools well and looks good, but boy it stinks! The plastic smell is headache inducing!
I did not recognize it. It sits in my lab since weeks and nobody is complaining.
@@AndreasSpiess Interresting. I was so bad i first thought I didn't remove the plastic from the heatpad. But I had. The smell comes from the plastic side panels. I noticed that the side panels also have some discolored spots that won't go off with isopropyl alcohol: www.dropbox.com/s/pbzmn5njf4a7mqk/pi-case.png?dl=0 I am not sure if this is just normal from the molding or maybe I got a batch with a botched plastic. I now have the Pi4 in the FLIRC case with stresstest-ng on 100% CPU load on the balcony for a few days so it hopefully loses its smell.
I'm building an arcade cabinet and the pi doesn't have to look nice. Definitely going with a 60x60 fan 👍🏼
Good Choice!
I’m subscribing just for how he said baaaii
Welcome aboard the channel!
Wahaha, If only this had been out before I bought the Chinese fanless thing ^^;
Talking about fans, the thicker they are the longer they last and the usual killers are: sleeve bearing wear and dust ingress into the bearing. (experience talking here.)
Same for me.
But I'm pretty sure there is some heat transfer problem between CPU to case.
I can't believe the Chinese Passive with all its massive Aluminum fins is so far from the FLIRC !
Being myself a big fan of no fan use, I just bought the Chinese Passive one also !
@@Llop66 Fins are only useful if you have air movement.
Putting the thing flat on the desk isn't really going to help at all.
Ideally it should be standing up with the fins vertically (linear flow, convection) but standing up with the fins horizontally would probably help.
Note, putting it inside a wooden drawer does not help one iota. I had customers do that with something at a previous job. Let's just say it (AMD SoC) got hot enough to throttle.
@@bunnymaid You're right, standing vertically should help air flowing (natural convection).
But still. Andreas said he tried that case with a fan and didn't get much better result : That makes me thinking there is CPU to case heat transfer problem.
@@Llop66 Shame that black body radiation isn't going to help at those temperatures.
I bought 2 of those fanless things, one grey, one gold. the latter ignores the USB and has a lump for the RP3 on the back. Not had a chance to do anything with them yet.
The thermal compound looks like blu-tak.
Thanks for the fan lesson. Good to know!