Methodology
Here are some questions that everyone who wants to tune their computer to the last detail deals with: Does it make any sense to remove a sticker from a powerful M.2 SSD format? Is it worth the effort and, if something goes wrong, the breach of warranty? How much will the heating be reduced? As a part of detailed tests, we also replaced the original stickers for ones of commonly available materials (paper, plastic foil, aluminum…) just out of curiosity.
Each company approaches manipulating their stickers a little differently. One more benevolent, the other not so much. For example, Samsung obviously does not want you to manipulate the sticker as you’d like. This is evidenced by the fact that the stickers on the 970 Pro/Evo (Plus) are particularly fragile and have notches in the corners. One perpendicular notch to the corner and the second one, in the immediate proximity, perpendicular to this previous one. We have no doubt that the toughest guys with the help of a hair dryer, thin razors and patient hands (and perhaps some other tools) will not damage such a sticker, but this operation is definitely not for everyone.
However, Samsung is an extreme case and most manufacturers won’t complicate your life that much. They either do not cut the stickers in any way, do use a shiny/firmer surface or they incorporate PET/PVC and even some metal elements. In such cases, the most difficult task of the whole process after peeling off the sticker is to store it appropriately and not forget where this archive is located. 🙂
One thing is how easy it is to get rid of SSD stickers, another one is whether it makes any sense to even try. That’s why we chose the WD Black SN750 from the range of modern SSDs, on which we tested various scenarios. First, of course, namely, what is the difference in cooling with and without a sticker. Although the WD’s sticker is thin, it is also firm thanks to its mesh structure, so there are not the slightest difficulties with its removal (and possible reapplication).
Testing methodology
In addition to how much better you can improve cooling of an SSD by removing the sticker, we were interested in comparing paper (and how its thickness affects heat transfer efficiency: tested with traditional 80, 190 and 300 gram white) with a thin and a thicker PET film. The thinner film used in the test is a standard packaging film that you can buy in every stationery store, and the thicker one is a printing 100-micron film, and we applied it on the SSD with the smooth side.
Then we have aluminum, i.e. thin aluminum foil. For these metallic materials, it is important to take care of the SMD isolation so that you do not kill the SSD immediately. For example, Silicon Power uses a metal structure for stickers. Aluminum is probably used in their production, too. We also used one of them (from SP A80) and adjusted it to a uniform size. This sticker is already thicker than the aluminum foil and consists of multiple components. The surface is probably made of PET and all these stickers, of course, use glue, which might not play a big role, but has an effect on the result, too. We put our alternative materials between the SSD and the heatsink only as a filler, without gluing. And they all always covered the same area and spots.
The largest heat surge from the test SSD came from the Sandisk control unit (which is WD’s own product). For workloads, we used continuous sequential write of large files for 600 seconds. In the end, we took the highest recorded temperature (from HWinfo) and just before the end (i.e. still at full power) we took thermal images of the heatsink. We then also read the maximum and average temperatures from it, too. In this case, the higher the heat, the better. The more efficient the heat transfer from the SSD.
The used heatsink is from the Gigabyte Z390 Aorus Pro. This motherboard (with the Core i9-9900K processor) is used for all testing. This is in a “bench-table” environment, but with airflow. A pair of Noctua NF-A12S fans, regulated to approximately 550 rpm, is responsible for the airflow around the SSD cooler. The entry air temperature was properly controlled during the tests and stayed between 21 – 21.1 °C.
Having spent a lot of time building and repairing high power solid state RF amplifiers I have a few recommendations: to get the maximum heat transfer it is essential to get good contract between chip and heat-sink so no stickers, no heat conducting pads just a thin spread of good thermal paste between the controller chip and the heat-sink you can apply a thermal pad to the non component side for stability. The controller chip is the only component on the board generating heat, the memory chips don’t so you are effectively heating the memory chips buy placing thermal pads over them and attaching them to the heat-sink!
Finally a good article, that makes sense and shows the tests about this issue.
Now all is clear and thank you for publishing this.
Can you also test combination no sticker + thermal paste + heatsink?
I have removed the sticker, put ton of thermal paste on SSD and attached lapped cooper heatsink (6mm thick).
But now I’m little aftraid that cooper surface may cause conductivity between SMD components and cause short circuit.
So the original sticker may also serve as electric isolation layer.
Thanks for your tip! I wrote it on a to-do list. 🙂
You’re better off putting a thermal pad between M.2 and copper (you can use a runner band to keep attached).
“…use a runner band…” I take it that was a type-O and you meant “Rubber band”. You should NEVERR use a petroleum product in this kind of application! Rubber bands arre made up of a petroleum ingredient. It doesn’t take long for a low-quality petroleum item such as a rubber band to break down.
Go find some old letters or greeting cards that you or a friend may have collected and saved. They probably bunched them up with one or more rubber bands. In a climate controlled, AC-cooled, environment, the lifespan of the rubber band may be as long as 4 or 5 years.
That would be a very optimistic estimate and based on a high quality batch of rubber bands. However, in a high-heat producing environment, such as the temperatures in a computer case or in an NVMe enclosure, (with or without a fan), the same rubber band wouldn’t last 6 months.
Hence, never use a rubber band for anything that has the potential of heating and then cooling at regular intervals, (or for that matter, at irregular intervals). In fact, just don’t use a rubber band for anything, particularly something that can create any amount of heat at all.
FWIW
Thank you!
Great help!