By the way, the existing radiator for VIM2 (if the mounting points coincide with the Board) is well suited for replacing the top cover of the case. And it will act as both the cover of the body and to remove heat from the processor.
One top and Bottom, thermal pad Excellent, seems to easy now especially if they line up
Bottom one could just be a block, and fan on top or both if needed
I like this idea too. The only downside could be that the unit cost will go up - more time will be spent on machining a larger block (volume) of aluminium.
A larger radiator would also make more space for a larger, slower and thus quieter fan. We can put this in our to do list for investigation.
have the vim 2 and loving it waiting for my heatsink and new case to come in the mail before getting back into it as it was getting pretty hot My two concerns with the edge v are #1 how will thermals be compared to the Vim2 and #2 will the M-register be more easily accessible than the Vim2? I freak every time i have to use it as it’s small and far too close to other solder points for someone lacking fine motor skills.
Hi Anubis:
Yep, Edge is much higher performance than VIM2 which means that the Edge will have much more thermal also. Here are some efforts we did:
When we designed the PCB, we keep more copper to ensure better heat spreading from the CPU to other space of the board.
We are designing a specified cooling fan for Edge to balance both heat dissipation and lower noises
When we designed the Edge, our engineer @Terry came out the idea that use onboard MCU to manage the upgrade modes, and we already achieve that just press the Function button 3 times than Reset button, the Edge will boot into upgrade mode. We still keep the M-Register for some case the MCU firmware works fails.
They can be, agree
Although if there is nothing, it can help
Due to the underside and Many uneven components not a perfect solution.
They Need to done tight Spring loaded things like in the link Not much good
I did this small mod few year ago now on a mecool box running Hi 90c with benchmarks
As the Base was metal a small bit of Pad in place dropped them testings temps 15~20c
If you design ‘something new’ why not doing it appropriately then? It’s known that the thermal conductivity of thermal pads while being ‘better than nothing’ is otherwise crap compared to direct contact with a thin film of thermal compound (I think it’s better than a copper shim in between?)
But it seems Khadas guys are fascinated by annoying fans and stuff like that so a great and efficient heatsink allowing for a passive cooling setup is most probably nothing they want to provide.
On which numbers is this claim based? If I look at some performance metrics then RK3399 is slightly faster than S912 (see NanoPi M4 compared with last Vim2 number)
At a comparable price (cost), passive option and active option, passive cooling system is not effective and will not be able to ensure guaranteed operation of the equipment under all prescribed operating conditions. The passive system can be used as an auxiliary, under light operating conditions or under medium loads and favorable operating conditions (open space with guaranteed ambient temperature).
The current heatsink is a fail for passive operation since heatsink fins too small and the area where half of the SoC is underneath with no heatsink fins at all (since there must be something that wastes energy for nothing and starts to make annoying noise after some time called ‘a fan’). Designing the heatsink to be used with an inefficient thermal pad of course makes things even worse. So yes, you’re absolutely right: in the Khadas world passive cooling is impossible.
Anyway: Many many RK3399 boards with apppropriate heat dissipation exist at a competitive price. The only interesting Khadas RK3399 option for end users so far is the Edge-V and no price is disclosed yet. Let’s stop here.