cancel
Showing results for 
Search instead for 
Did you mean: 

NVMe m.2 laying flat on mobo uncovered. Is the cover a heat-spreader? (TUF z270 Mk1)

tgm1024
Level 7

UPDATE:
30-JUL-2020 (Thur)
[/HR]On my motherboard (TUF z270 Mark 1), the cover is obviously plastic with no heat pads/spreaders underneath it. Further, the venting is small. I've been running it without the cover and will continue to. I will be experimenting with heatsinks.


ORIG: 15-JUL-2020 (Wed)
[/HR]ASUS TUF z270 Mark 1 (roughly 3 years old)
Samsung 960 EVO NVMe m.2 (same age)
(case: Corsair 570x, yadda....)


3 years ago I hesitantly installed the NVMe horizontally, but kept the cover off because the lack of airflow worried me. Things have at least seemed ok thus far, good grief I hope I didn't just jinx everything.....

However, moving forward both with this and in future builds:

1. Should I install an SSD heatsink on the thing? The online community isn't in unison on this: Some point to how heat actually helps NAND memory lifespan (to a point), and others are freaked by the controller on the thing.

2. Is that cover (that I never installed) actually a heat spreader/sink itself?
2,163 Views
3 REPLIES 3

Megalamaniac
Level 9
I've been seeing this question a lot lately and the only references to heat helping (that i found) are from about 8 years ago. I didn't do a deep dive on this, but I was able to find an article that's less than a year old pretty easily. I did quick search of "heat actually helps NAND memory" and found this article - https://www.maketecheasier.com/does-nvme-ssd-need-heatsink/. Very informative on the subject, but in a nutshell heat kills.

I've got a couple of the new Evo plus 970's and a Evo 960 currently in my system. Without a heatsink the 970 being used as a system drive would quickly hit temps of 95c+ for the controller, and start thermal throttling. This was on a RVIE and it's not really possible to sink the M.2 socket that on the MB due to space limitations, nor was the cover a heatsink. The MB recenty failed, so I changed it out to a RVIEE, which has built in heatsinks on the socket cover, along with a DIMM.2 heatsink. I also had an Aquacomputer x4 SSD waterblock I picked from Amazon (they screwed up and upsold it to me for $4 awhile back, couldn't pass it up), but haven't used it until now. Now my system SSD controller temp hasn't gone past 41c, and my sinked SSD's are maxing in the low 50c's.

Also to note, my 960 started out as a system SSD and been in a system since I got it almost 4 years ago, and I only noticed today that it's S.M.A.R.T. info finally dropped from 100% to 99% remaining life, according to HWiNFO.

I'm not familiar with TUF Series boards, but if your cover has thermal pads under it for the SSD, it should be a heatsink. Either way, the cooler the better, IMO.

Megalamaniac wrote:
I've been seeing this question a lot lately and the only references to heat helping (that i found) are from about 8 years ago. I didn't do a deep dive on this, but I was able to find an article that's less than a year old pretty easily. I did quick search of "heat actually helps NAND memory" and found this article - https://www.maketecheasier.com/does-nvme-ssd-need-heatsink/. Very informative on the subject, but in a nutshell heat kills.

I've got a couple of the new Evo plus 970's and a Evo 960 currently in my system. Without a heatsink the 970 being used as a system drive would quickly hit temps of 95c+ for the controller, and start thermal throttling. This was on a RVIE and it's not really possible to sink the M.2 socket that on the MB due to space limitations, nor was the cover a heatsink. The MB recenty failed, so I changed it out to a RVIEE, which has built in heatsinks on the socket cover, along with a DIMM.2 heatsink. I also had an Aquacomputer x4 SSD waterblock I picked from Amazon (they screwed up and upsold it to me for $4 awhile back, couldn't pass it up), but haven't used it until now. Now my system SSD controller temp hasn't gone past 41c, and my sinked SSD's are maxing in the low 50c's.

Also to note, my 960 started out as a system SSD and been in a system since I got it almost 4 years ago, and I only noticed today that it's S.M.A.R.T. info finally dropped from 100% to 99% remaining life, according to HWiNFO.

I'm not familiar with TUF Series boards, but if your cover has thermal pads under it for the SSD, it should be a heatsink. Either way, the cooler the better, IMO.


Hmmmm......well 8 years might not be too old to invalidate any substrate level tech info. It is for higher end devices, the devices as a whole, but for base-level NAND thermal reactions much of it I suspect would be similar to how we dug up 25 y.o. white papers on OLED to try to figure out what the heck Samsung was doing wrong with their larger displays.

Thanks for the reply.....I'm not "deep diving" either, but did you notice in that MakeTechEasier.com article you posted that they themselves refer to a "warmer is better" article (from 2012, like you said) with a caveat that it has to stay between 50°C and 70°C. But when I looked through that article, I think MTE completely misread what they (EEWEB) were saying.

From MTE, near the bottom:
The heatsink is the slowest to heat up due to its sheer density. The fact that the NAND flash temperature readings are relatively higher compared to the motherboard heatsink shows that it does a good job of taking heat away from the controller. This is actually great because, contrary to popular belief, NAND flash performs optimally and is safest at temperatures of 50°C and upwards, as long as it stays below 70°C, beyond which the data integrity can get compromised.



....and that 2012 article they point to (https://www.eeweb.com/profile/eli-tiomkin/articles/industrial-temperature-and-nand-flash-in-ssd-prod...) does an incredibly deep level analysis and simply says that for the most part, the acceleration of errors (I forget the term) lowers with temperature. So I didn't see a clear cut "warmer is better but below 70" from EEWEB.

Bottom line, your assessment of "lower is better", even if it gets appended to by some with "probably always" is what I'm going with. Seems the most sound advice outside of a laboratory. I'll be checking out the mentioned heatsinks. Thanks for that.

Watch my 960 croak within a week of me putting it on....

Initial post updated: Cover is plastic. No heat spreader.