Hi~Mattia
Firstly, thank you very much for your feedback and support!
We will take your suggestion positively, it is a valuable suggestion for us, I will feedback your feedback like our development department, please don’t worry, thank you again for your support, our metal case you can consider to use, or you can contact me at my email, I will give you a nice activity, thanks.
My private email address: runfa@youyeetoo.com
Well after the two days of uptime I can provide the updated thermal performance of the case at 25C room temp:
Idling for at least 18hrs: avg.53C, the case is really warm and at the edge of not being stored on a fire hazard materials like carpets IMO (but possibly I’m just too paranoid about that)
The NVMe figures for me are especially interesting: my NVMe has two temperature sensors, one shown avg.55C and another avg.75C. Judging by actual thermal status by hand, I see first one shows ambient temperature and second… I dunno. But they correlate to each other, and with fan I had few less figures of course
Building kernel 100% all four cpus for, I guess, eight hours: avg.63C, the case alarms my finger the temp is obviously higher than 50-55C when I touch it and it feels not okay. Sustainable though.
Overall stability of my system wasn’t affected much, but to warm up whole thing it takes about four hours to be sure. But it grabs heat into a massive CPU pillar very well (did not notice much CPU temp spikes really).
But please don’t take the figures too seriously, the main “offender” here is JH7110 itself, it generates enough heat to make the whole thing get such warm and my hopes for the future is better CPUs with better (ideally same as current typical smartphone ones) thermal performance while still being efficient.
For example, I had an old ASUS X101CH netbook with Intel Atom N270 from 2013, and it had passive cooling solution in a really thin case. It heated very well, sometimes to trigger it’s embedded CPU thermal shutdown protection at 105C (once I left it compiling kernel while laying on a sunny spot with surface temperature at 50C and it just went emergency turn off at some point)
Overall performance is still not quite bad, as I said, the main heat generator here is JH7110 itself (you really can’t get accumulated heat elsewhere without forced cooling anyway). But, as I said, I would like to see aluminum version of it (why not? Is aluminum that really expensive?)
3 Likes
I think just change the top and the iron bar to aluminum will be better. May be add an elevated bottom and thermal pad at the location of NVMe SSD.
2 Likes
I have always liked the idea of a metal case with two small fans on one side blowing cold air over the top and bottom of the board and then exiting on the other side of the case. Ideally with a thermal pad on the CPU dissipating it heat to the metal case through a pillar. Basically the airflow is interacting with the maximum amount of surface area of the printed circuit board which has tracks made of copper that are fantastic at conducting heat from hot parts of the board to cold.
2 Likes
Hello my friend, that’s exactly what we did with the metal casing and that’s exactly what we did !
Hi friend, your suggestion is very valuable to us, we are very happy to receive the real experience from our users, it helps us a lot, I’ve already feedback to our technical department about the demand of aluminium casing reflect, please look forward to see if we will release the aluminium version, thank you!
1 Like
Is there an image showing the fan placement inside the case ? Because looking at the current images of the metal case, I just can not see how your case matches what i said.
1 Like
Hello Friend.
First of all, thank you very much for your real user feedback, your suggestion is very valuable to us, we are very happy to receive real user experience, it helps us a lot.
Secondly, the performance of the JH7110 CPU you mentioned, the current performance of this chip can still meet the market for RISC-V most of the user’s needs, we are also committed to solving the problem of CPU heat dissipation, and we hope to be able to give users a better experience.
Finally, you can contact me at my email address, and I can recommend other RISC-V solutions that may be helpful to you, and I’m very happy to have in-depth research and discussion with you, thank you!
Personal contact email: runfa@youyeetoo.com
1 Like
Hello friends, at present the metal shell is not configured with a fan, only the acrylic shell is configured with a cooling fan, you can check this link, thank you!
Acrylic Case Buying：Aluminium Heat Sink + Fan
I don’t think these number are that good. For me idling without any thermal solution also produce around 50C. And running stress-ng on all cores produce around 75C. I installed Waveshare fan and heatsink (which is basically horizontal variant of ice tower) and now idling at 30C and under max load at 40C. All at ambient of 24C. I don’t have any cooling for SSD and it for sure heats the board. But in my experience SSD coiling is not really necessary for use with SBC. I realize that it’s not fair to compare passive cooling with efficient active cooling especially in open air, no case. But here is is.
1 Like
@strlcat What software do you use to read NVME temperature? I use lm-sensors/sensors but it does not return it.
1 Like
Your kernel much be compiled with CONFIG_NVME_HWMON=y to use sensors command to monitor NVMe.
3 Likes
Hi, I use smartctl -A /dev/nvme0, but a sysfs interface is also available at /sys/devices/platform/soc/2c000000.pcie/pci0001:00/0001:00:00.0/0001:01:00.0/nvme/nvme0/hwmon1/temp*_input
As @cwt noted, you need to have CONFIG_NVME_HWMON enabled prior to that.
3 Likes
Hey there,
Even if I’m critical about thermal performance (anyway, passive heatsinks suck when being small enough), I shall note that case design is good enough for using it almost everywhere and embedding it as an appliance. It’s not all-weather proof of course, but sufficient for most indoor applications.
So take my initial criticism with a fine grain of salt 
1 Like