More to the fundamentals, it’s kind if questionable whether higher or lower heat helps on an SSD. So long as the controller isn’t thermal throttling and slowing you down, a hotter SSD can actually counterintuitively help lifespan…
PCIe gen 4 SSDs can thermal throttle very easily even under open air, hence why a lot of them come with small coolers now. My hp envy x360 with a gen 3 SSD and an aluminum chassis, but no m.2 cooler, had temperatures above 75°C for the controller aswell as 50-60°C for the NAND quite often. That temperature occured without me even doing a lot. You see the temperature with programms like HWInfo. I solved this in a similar way Fraoch mentioned with a thermal pad under the ssd and a thin cooper plate i cut out on top. The ssd was a Samsung PM961
Don’t get me wrong, there’s no doubt about it that PCIe Gen 4 SSDs can run ridiculously hot. I’ve seen them get up to 105 before in other computers. I’ve actually been considering downgrading my SSD to a PCIe gen 3 one to drop power consumption. The intent of my last post was simply to point out that NAND temp is not necessarily a bad thing.
NAND flash is supposed to like temperatures around 40°C for the maximum lifespan. Having a good cooler can lower the temperatures below that mark. Though i don’ t think that you can run into this issue that easily in a laptop
Now that I actually have my Framework and those heatspreaders, I couldn’t dare to use them. There is just no room, not even for 1mm of copper.
I broke out the micrometers and measured 4.17mm from the mainboard base to the top of the cooler, which seems to be the component that stands tallest. I then measured 3.39mm from the top of the M.2 to the mainboard base. That’s a difference of only 0.78mm. No way was I going to put a 0.5mm thermal pad plus a 1mm copper plate here. I thought it would cause the copper to push against the underside of the lid.
That definitely sounds about right based on my experience. I bought this one that claims to have an installation height of 0.75mm, but I think that’s a bit optimistic. My calipers aren’t as precise but the heatsink is about 0.4mm and the thermal pad looks to be about 0.5mm, there’s a little compression but I’d say it’s adding around 1mm to the height
On installation I noticed it does stick up slightly compared to its surroundings, about 0.6mm above the adjacent battery. When I first secured the keyboard housing it looked like it was bowing ever so slightly, but after I poked it it’s looked normal since, and if I use a level it’s flush across the surface.
I do suspect it’s applying a slight amount of pressure between the keyboard housing and the mainboard, if it fits it just barely fits. Debating whether to leave it installed.
At that clearance the heat spreader will probably hurt rather than help. You won’t be able to get any air flow past the heat spreader, so you’ll just trap the heat instead. (0.78-0.75 is effectively zero, especially with keyboard deck flex factored in)
Your only bet at this type of clearance is either piping the heat somewhere else or dumping it into the keyboard. The first is not easy with 0.78mm clearance, and the second may come with unintended results.
Exactly my concern. I can say the heat is dissipating slightly worse than it did in the old laptop, though better than stock, but that means it’s probably dumping heat into its surroundings.
Guess I can try it without and weigh getting a more efficient drive. This one tends to thermal throttle unfortunately (not a huge loss, just reusing my previous laptop’s drive).
I went ahead and removed the heatspreader and that got rid of any perceptible flex in the keyboard deck, so pretty sure it was making contact… thankfully my workload doesn’t write large amounts of data very often so I’ll just deal with thermal throttling when it does occur.
If one were to attempt adding an NVME heatspreader I’d recommend using an extremely thin and malleable thermal pad like Gelid GP-Extreme 0.5mm as it deforms very easily to fill gaps when pressure is applied, and a heatspreader no more than 0.5mm thickness, but preferably thinner…
TL;DR, adding a heatspreader in this laptop will be difficult. Better off putting the money into a cool and efficient NVME if you’re buying new (SK Hynix P31 is often recommended).
After reading this, I’m thinking of a DIY solution as I couldn’t really find anything commercially available on the german market fitting the specs you mentioned. Of course I’ll wait for my Framework to arrive and the experience of others with the heatsinks mentioned in this thread first (some report initial success with the 0.5+1mm copper plate e.g.).
But if the 0.78mm clearance prove to be correct, I’d probably try a combination of these two:
It’s 20 instead of 22 mm width, but I guess that would suffice. Of course the actual thermal gains with such a hacky setup are questionable, but I’m willing to give it a shot.
For anyone interested in this topic I reused an few old 1mm thick thermal pads on the SSD controller and it makes contact with the top cover sinking a great deal of heat.
Also guys I think you actually want your flash chips to be hot when writing data to them, the controller does not want to be hot.
Hello,
I am thinking of using the Samsung 990 Pro in my Framework 13 with Ryzen 7. I think it’s known for running a little hot so I am definitely interested about a heatsink / hearspreader solution.
I came across these 0.15mm copper+graphene heat spreader and for only a few bucks, I think I will definitely try it.
Honestly?
All such ultra-thin heat spreaders just feel like a gimmick. Without fins, it can’t provide as much of an increase to the surface area in order to shed heat. But I’ve never looked into any test results for similar heat spreaders. It is cheap. Not much harm to try
Samsung claims to use a heatspreader label on the 990 PRO so maybe that already does all it can do. The tiny thermal mass of such a thing certainly won’t do a lot.