I recently started to use my Vision Five 2 for some tests and realized it becomes rather hot without active cooling, and a few small heat sinks applied to the chips did not change this. Basically, the whole board and all connectors heat up and dissipate heat it you load the CPU for prolonged time.
While the Vision Five 2 lite presumably consumes less power due to reduced clock rate and reduced functionality, it has also less board size to dissipate heat. Are there any plans to produce solid aluminum cases with build in heat sink, like the Flirc Case for the raspberry pi?
If you take the 8GB-version, add an SSD and also consider the cost for a decent PSU, a proper casing doesn’t double the cost any longer. I may want to put the board into light server duty, and for this, I prefer a proper case and not a naked board, sitting on spacers, maybe with a fan mounted with cable ties.
If you are using NVME, it is recommended to use active cooling. NVME is installed at the bottom of VF2 Lite, which increases heat and accumulates on the CPU.
The Raspberry Pi 4B does not have a built-in SSD, so it can use passive cooling. Raspberry Pi 5, The dealer suggested using active cooling.
The power consumption and heat of RISCV are higher than those of ARM.
I do not own that case, but looking at their website it appears to only have a single aluminum pad for what you would expect to be the hottest chip on a RPi5 board (BCM2712 SoC). But depending on what you are doing, the chip with the highest temperature is typically going to be the Power Management IC. I have a RPi5 that I run LibreELEC on to playback 1080p AV1 files (It will eventually be replaced by a RISC-V SBC with either a hardware AV1 codec or enough processing power to decode AV1 video files in software). That PMIC chip can easily hit double the temperature of the BCM2712! I do have a very large USB 3.0 solidstate drive hanging off the SBC which is helping to push the PMIC so hard (and the RP1 I/O controller would be helping aswell).
Anyhow the design of that case would not be one that would interest me. But maybe your use case is one without any power hungry USB devices attached. In my mind if Flirc got the design so wrong on the RPi5, for what I would consider a major use case, I would not be holding them up on a high pedestal as a good example of what I would want for other SBC’s.
With an M.2 SSD on the board itself, I don’t have a need for an external SSD. Therefore, I expect to have lower load on the PMIC (But the M.2 SSD needs power as well). From what I can see, the official heatsink/cooler for the Raspberry Pi5 has thermal pads for the CPU, PMIC and Wi-Fi, so cooling for the PMIC may be needed. I still don’t like active cooling solutions for such relatively low power devices, It should be possible to cool it with a large, silent heat sink (those small fans sometimes fail earlier than expected and may be hard to replace, if they are custom designed).
The two cases I like for the RPi 5 are the “raspberry pi pure copper sheet heatsink passive case” (8 sheets being better than 3) and the “raspberry pi armour case”. But both are not exactly what most people would call conventional cases. And neither support the M2 addon hat.
They are good at what they do, the copper one being a lot better. In my mind either style would be nice for the VF2 or VF2L SBC.
I completely agree with your idea, No need to clean accumulated dust。My Raspi 4B uses passive heat dissipation, External fans are only used for cooling in summer, and no fans are needed in winter.
I’m just guessing that passive cooling may not be suitable for VF2 lite with built-in nvme, We need to wait for VF2 lite to be released for physical testing.
If anyone has an interest in cooling there is a really down to earth (and funny) book about it called “Hot Air Rises and Heat Sinks: Everything You Know About Cooling Electronics Is Wrong” by Tony Kordyban. It will bring you from knowing next to nothing to a total expert in a very short time - because it is so engrossing and entertaining.
It focuses on fundamental misunderstandings of heat transfer and the common, yet often incorrect, assumptions engineers make about cooling electronics. It is mostly real world examples of how heat transfer principles were misapplied and then explaining why and the lessons that were learned. It is basically teaching by showing how people in companies got it wrong.
I have stress tested both boards for more than an hour with zstd and xz (I can’t tell how much extra heat the GPU would use, I don’t use it due to performance/driver issues).
The VisionFive 2 reaches about 76°C (at an ambient temperature of 24°C, with a small heatsink/fan mounted, but the fan not turning, in an enclosure which is open at the top). The mounted SSD reaches about 70°C.
With the fan spinning, the temperature is well below 50°C at full load.
The VisionFive lite reaches about 63°C (on standoffs, sea picture below). I can’t tell about the SSD temperature, the current SSD doesn’t report it (or the drivers and software in Ubuntu can’t read it out).
With the fan spinning, the temperature is 45°C at full load.
To cool the VisionFive lite, I mounted a small 40mm 12V fan from Noctua using M 2.5 standoffs and powered it with a USB adapter from Noctua (Noctua doesn’t allow this from a technical point of view, because the fan is 12V).
Very cool trick the way that you are mounting that fan.
A simple trick for higher natural convection cooling is to mount the board vertically off the ground to improve laminar airflow. That way there is less resistance to natural air flow (hot air rises) on both sides of the board.
Another improvement to that is to mount a fan directly below the board which is still vertical, blowing cold air up, that way you are cooling both sides of the PCB which is in effect a large copper heat sink. And you are gaining a minute amount of extra cooling aided by natural convection.
I am concerned that the price of solid aluminum cases may be greater than or equal to $19.9. 
I can’t understand English books. 