Best case for cooling a Pi 3?

DirtyRob

@enderandrew its about sustained performance ... you need to increase processing and see where they cap out at. An open air one with no contact to anything solid to sink heat (just air on surface of chip packaging) just doesnt have the surface area and dissipative capacity. I dont mean to shoot the test full of holes but you didnt push any of them to a failure mode (throttling), so as far as im concerned those 2-3 deg differences in the fan-off ones and the flirc are still in the noise range to me.

What this does show me though is that you are probably not making good connection to the chip if on 3 other cases where there is no contact with anything at all you get roughly the same temperature. That just doesnt make sense.

Check this study ... with completely open air pi3 no case at all just laying out in the open. He pushed them to a failure mode so you could actually see the difference. I dont think its showing because of either 1) bad interface to the chip or 2) not a high enough demand on the processor. Remember too that heat increase on these things is not going to be linear. It will get to a point (open chip) where it simply cannot dissipate anymore. That the area you should be pushing them to.

BuZz

@enderandrew I don't agree and I think you need to test for longer I think - with load on my rpi3 (building on all cores etc), it gets significantly hotter and throttles with no case (open air), than using the flirc case.

enderandrew

@DirtyRob said in Best case for cooling a Pi 3?:

@enderandrew its about sustained performance ... you need to increase processing and see where they cap out at. An open air one with no contact to anything solid to sink heat (just air on surface of chip packaging) just doesnt have the surface area and dissipative capacity. I dont mean to shoot the test full of holes but you didnt push any of them to a failure mode (throttling), so as far as im concerned those 2-3 deg differences in the fan-off ones and the flirc are still in the noise range to me.

Open air cases have TONS of surface area. I think you're also assuming that open air cases don't use heat sinks at all. You can use heat sinks with the open air case.

The benchmarks at the end of the video show that a Flirc case beats no heatsinks or a tiny heatsink, which is common sense since the Flirc case is a giant heat sink.

I didn't watch all 10 minutes, but the intro and end of the video has no mention of them running an open air case in the test.

I'm curious to see an actual benchmark of the Flirc versus the JBtek open air case (with and without fan on) with the Pi 3. But with the Pi 3, the JBtek was cooler. You're saying the open air case has no surface area to cool, but you can't explain how it beat the Flirc in testing then.

DirtyRob

@enderandrew said in Best case for cooling a Pi 3?:

Open air cases have TONS of surface area. I think you're also assuming that open air cases don't use heat sinks at all. You can use heat sinks with the open air case.

You can use them, yes, but the surface area of those small heatsinks (without the fan) are nowhere close to matching the surface area of the flirc case. And as the other study points out ... those heatsinks alone wont cut it (hence the fan). Im not assuming that there is no heatsink. In fact i was comparing the "fan off" tests to the ones he claimed has a small ebay heatsink, because they do dont they? All other surface area is negligible since it consists largely of non conductive surfaces. You dont want your surrounding circuitry to be considered your "dissipative surface area" and you certainly cant count all the acrylic plates and hardware either.

The benchmarks at the end of the video show that a Flirc case beats no heatsinks or a tiny heatsink, which is common sense since the Flirc case is a giant heat sink.

Exactly. So im wondering how its possible that one sandwiched between two thermal insulators and has a huge plastic fan (in the off position) hanging over it obstructing air flow around the largest heat source on the board can get better temperature results than one out in the open with nothing on it at all

I didn't watch all 10 minutes, but the intro and end of the video has no mention of them running an open air case in the test.

Laying on a desk face up is the equivalent of being in that open air case by JBtek. There is little to no separation between the board and the backside acrylic piece ( i do see there are standoffs there ... but keep it perhaps 1/4" off the base?). I have to assume whatever desk he set it on was not thermal conductive because of the numbers he got. Theres no reason to think that an acrylic backing of the JBtek would perform any better than whatever he placed it on (barring maybe a shag rug or something). I will say though that i might be referencing his setup from another video he did. He has a couple. I think in this video he never actually shows the setup physically while hes running it ... just the PC monitor and code running. So I suppose we have to take some of it with a grain of salt.

I'm curious to see an actual benchmark of the Flirc versus the JBtek open air case (with and without fan on) with the Pi 3. But with the Pi 3, the JBtek was cooler. You're saying the open air case has no surface area to cool, but you can't explain how it beat the Flirc in testing then.

I tried to explain that, yes. I suspect bad interface between the case and the chip. Thermal pad or paste. Thats really the only explanation. For it to get hotter would have to mean that it was an air flow blockage issue, and that the actual contact with its heatsink was negated as a factor. Then those numbers would look reasonable to me.

I would like to see the JBtek one tested the same fashion as in the video. Fan on it looks to be effective. Im not convinced though that at more demanding processing temps that that would still hold true. Air cooling and this passive cooling case will cap out with different curves to them. A huge heatsink will handle heat differently than a fan combo will. They both will cap out somewhere with different slopes to them. Thats what im more interested to see.

space cadet

I'm using a zebra case with heat sinks and a fan. Works great and fun to put together.

DirtyRob

This post is deleted!

H-A-Z

Hi, you many be interested in reviewing the results of the experiments conducted by MR-Bodger on open case design with USB fan attachment and the ceramic heat sink thermal dissipation performance:

http://www.instructables.com/id/RPi-3-Cooling-Tests/?ALLSTEPS

H-A-Z

@H-A-Z *may

edm203

I have this wooden case, and it's terrible for keeping my pi cool.
DecaPi Slider

I bought a plastic case on ebay with a fan for about $6, and it's much better at keeping the temp down.

sergioad

An aluminium case enclosed in a flexible and breakable plastic shell to protect the internals

HurricaneFan

I have this aluminum case, no complaints:
https://flirc.tv/more/raspberry-pi-case

Rion

alt text

It's relatively easy to modify a case to do the same.
You just need something to hold the heatsink in place.

Rion

Here are the results with a massive heatsink and a fan.

Here are one more and final test with overclocking using a massive heatsink.

This is by far the best results in have seen and that's using an Aluminum heatsink.

tekn0

This case works as a heatsync and keeps temps in check. https://nostalgiatechs.com/heatsync-case-pi-amazon

edm203

@tekn0 That's a great case, but reviewers state it doesn't fit well for Pi 3.

tekn0

@edm203 it fits mine fine, the only issue is that you need a paperclip to pull out the sd card. as it's flush with the metal. this case was a life saver for the price because my pi3 would overheat after about 5 mins of neogeo play, now it has no issues at all.

nschloe

I've created stressberry to create standardized temperature signatures; hopefully we'll be able to properly compare all solutions with it. I'm hoping to be able to add a few more graphs in the coming weeks. Perhaps other people would like to run stressberry as well.