I am posting here today to ask a general question for which I’d appreciate opinions.
Yesterday I replaced the stock thermal pads with new thermal pads (Artic TP-3 120x20x1mm) in one of my 4TB NVMe drives (see photos of drive below). I have gained a solid 5C drop in the upper temperature range for the NVMe during testing, sometimes fluctuating to 6C.
The question I’m asking is: Is it worth a 5C (41F) (9F) drop to replace the thermal pads in my remaining 5 NVMe drives?
The test drive was my spare. It was not in an air flow environment, just the normal room air flow which isn’t much. I expect better results with active cooling of course, but using the test drive gives me an idea what can be obtained.
For transparency, the five NVMe drives within my NAS get up to 58C while running a scrub. I actually expect more than a 5C drop with these drives having active cooling to dissipate the heat, whereas the test had no forced airflow.
In the photos you can see the thermal pad material ooze out. I cut them to exact length to cover all the chips on both sides, sandwiched it all together.
I used a 1mm thick pad for both sides however I need to purchase a .5mm pad to lay between the heatsink and the NVMe if I decide to replace the remaining thermal pads. Having two 1mm pads was a bit difficult to put back together. I had to use a small vise. I do not recommend it and slowly compress the heatsink until I could install the screws. After sitting overnight, I’m retesting the entire NVMe drive right now to ensure the entire drive can write/read data (+10 hours remaining). I want to ensure I did not induce a problem. This drive already had some defects, hence I used it.
This would be quite high for a HDD, but should be safely within operating range for a SSD.
So I’m tempted to suggest that it may not be worth the hassle in this case. (No hard data to back this feeling.)
What! You mean I completely forgot to subtract 32? Well I knew it was way too large of a difference. I should have payed more attention in physics class.
The drive in question has a warning temp of 84C and a Critical temp of 89C. Yes, well within the limit. But as we all know, cooler electronics is generally better.
That is okay, I am not looking for specific factual data, but I’d take it if it were available. I’m looking for opinions.
In just over 6 hours the modified NVMe drive will have completed it’s testing, so far no errors and I’m over half way through. If this drive passes, I will order the .5mm thick pads and modify one more of the drives that is in the machine. Then run another scrub to see what the temps are. I know this will provide me an answer to if the active cooling will make a significant difference.
It would also be nice if I could reduce the fan speed to 800 RPM and still retain improved thermal cooling.
The testing of my spare NVMe drive just completed and is a PASS. I did not break it. I’m glad about that. So I did order some .5mm sized thermal pads. Unfortunately they won’t be here until Monday evening.
I did some research on thermal pads and it turns out there are some pretty efficient pads out there. I need to read more but for this adventure I am staying with Artic TP-3 material. The best material I found was conductive and very expensive. The TP-3 is budget friendly while still offering above average performance.
I will post an update once I have the next drive converted. I hope I see a 10C drop or better, but I may be day dreaming.
