Question 4TB NVMe overheating in external enclosures

[twistedtree]

I have a PNY 4TB NVMe M.2 SSD stick that I have now tried in 3 different external enclosures, and in all cases the SSD overheats and disconnects. Has anyone found a combination of 4TB NVMe + external enclosure that doesn't overheat ?

In more detail, the enclosures are USB-c and all get great reviews, 2 of 3 include thermal tape to conduct heat from the SSD to the enclosure case. But when I run a backup from a MacBook to the SSD, they all overheat and disconnect after a few minutes. I suspect that part of the problem is that the NVMe stick has flash chips on both sides, and in all the enclosures the thermal tape only creates a heat sink for one side of the stick.

I suspect this is a combination that simply can never work under any sort of R/W duty cycle. Engineered to fail?

 

[Lafong]

I'm using a 500 GB WD NVMe in an external enclosure without incident.

Its chips are just on 1 side.

The enclosure included thermal tape, but I have not used the tape.

I don't leave it connected constantly.

When connected but not being used, temps are in the mid 40s in a warm room (80 F/27C).

I wrote 366 GB of data to it. Took 25 minutes. Max temp reached was 63.

I have no experience with double sided NVMe sticks. Your guess about them might be right. I've never experienced any throttling. I suppose the enclosure might make some temp difference. Mine is made by TDBT and was about $22 at Amazon.

 

[twistedtree]

Thanks for the data point. A lot of people are using smaller SSDs succesfully, including me. I have on 750GB one that works fine. But this 4TB, double sided device just doesn't seem to be usable with any sort of sustained traffic. All PNY can tell me is that 70C is the max temp, and they point their finger at the enclosure. But they can't tell me what teh 70C temp limit is. Ambient free air? flash chip surface temp? Controller chip surface temp? And they can't tell me whether it's designed to operate in free air, or if it requires some means of heat extraction. And if heat extraction is needed, they can't tell me how much. It strikes me as not just a poorly engineered product, but a completely un-engineered product. And I get silence from the enclosure companies. Just use it and hope for the best. I'm so sick of tech companies dumping crap on the market and wasting hours of my time.

So heads up, by all indications it's impossible to operate a current generation 4TB NVMe stick in any of the many USB C enclosures. There just isn't any thermal management, let alone any thermal engineering. It's too bad because it would be a really nice package, but it just doesn't work.

 

[Lafong]

It strikes me as not just a poorly engineered product, but a completely un-engineered product. ..............................So heads up, by all indications it's impossible to operate a current generation 4TB NVMe stick in any of the many USB C enclosures.

May be a correct conclusion. Wondering if you found any online instances of 4 TB doing OK in an enclosure if NOT a PNY product? Wondering if it's PNY engineering issue or a 4TB issue per se?

Is there some understandable and valid correlation between temps in an enclosure and the capacity of the NVMe? I'm guessing there is...at least apparently.

 

[twistedtree]

I haven't found anyone using a 4TB stick in an USB enclosure, and posted here in hopes of finding one. So far there is no existence proof.

Generally the highest capacity chips will run the hotest, and the highest capacity NVMe will have the most chips. I expect that's what has driven this to be double sided with components. With the next generation of flash chips I would expect 4TB to require fewer chips, run cooler, and the problem will move to the new largest NVMe sticks that a jamb packed with chips.

 

[dwd999]

I haven't found anyone using a 4TB stick in an USB enclosure, and posted here in hopes of finding one. So far there is no existence proof.

Generally the highest capacity chips will run the hotest, and the highest capacity NVMe will have the most chips. I expect that's what has driven this to be double sided with components. With the next generation of flash chips I would expect 4TB to require fewer chips, run cooler, and the problem will move to the new largest NVMe sticks that a jamb packed with chips.

I've only used a 1TB stick in a Sabrent enclosure. As a general rule, all of the heat is generated by the controller chip and the cooling solutions are designed to spread the heat out to the memory chips away from the controller. This would explain why most enclosures only have a thermal pad on the one side where the controller would be located. I noticed in my enclosure that the thermal pad had a shiny protective film over it which I had to remove; this is not covered in the instructions. Since I also sometimes obsess about heat, I bought a small handheld rechargeable fan that I can use to cool my enclosure if I think things are slowing down.

 

[twistedtree]

I've only used a 1TB stick in a Sabrent enclosure. As a general rule, all of the heat is generated by the controller chip and the cooling solutions are designed to spread the heat out to the memory chips away from the controller. This would explain why most enclosures only have a thermal pad on the one side where the controller would be located. I noticed in my enclosure that the thermal pad had a shiny protective film over it which I had to remove; this is not covered in the instructions. Since I also sometimes obsess about heat, I bought a small handheld rechargeable fan that I can use to cool my enclosure if I think things are slowing down.

Thanks. I just want a 4TB drive that works, not a science experiment, and to not do the manufacturer's QA job for them. I'm just going to return all this junk and let their problem be their problem instead of my problem.

 

[fzabkar]

Thanks. I just want a 4TB drive that works, not a science experiment, and to not do the manufacturer's QA job for them. I'm just going to return all this junk and let their problem be their problem instead of my problem.

So far you are only theorising that temperature is the culprit. Why not blast the SSD with a fan and then torture test it? If the SSD still falls over, and SMART is reporting a normal temperature, then the problem is elsewhere. In fact plenty of people are reporting stability issues with certain chipsets.

Here is one very long thread on the subject:

https://forums.anandtech.com/threads/stable-nvme-usb-adapter.2572973/

 

[Rokinamerica]

Perhaps something like this might help?
https://www.amazon.com/ineo-Enclosu...coding=UTF8&psc=1&refRID=DMJ16J9J5GPVBTPVJKM2

 

[twistedtree]

So far you are only theorising that temperature is the culprit. Why not blast the SSD with a fan and then torture test it? If the SSD still falls over, and SMART is reporting a normal temperature, then the problem is elsewhere. In fact plenty of people are reporting stability issues with certain chipsets.

Here is one very long thread on the subject:

https://forums.anandtech.com/threads/stable-nvme-usb-adapter.2572973/

I am attributing it to heat, mostly because it works fine until the data rate goes up and the device gets too hot to touch. Then it disconnects. The longer I let it cool, the longer it will run the next time. And I agree what you have described would be the next step. But I'm not going to do it. I buy products to solve problems that I have. I don't buy them to create work for myself. If they don't work, I have no use for them.

A few years ago I decided I was sick of the <Mod Edit> products people put out and that going forward I would go through no more than one round with tech support on a product, and if it didn't work, I'd just return it. I'm tired of doing QA, compatibility testing, and debugging of other people's products. That's their job, not mine. As consumers we have been so conditioned to put up with this crap, that it's what we get most of the time.

Thanks for link. I hadn't seen that particular discussion, but have seen others talking about the different chip sets. The first enclosure I bought use the JMS chipset, and it would crash my Mac consistently. Searching online I found this was a common complaint, aloing with recommendations fro the RealTek chipset. So the next two that I bought had the RealTek chip. They don't crash the computer, but they do disconnect when they heat up.

I talked to one of the enclosure's support people and they said it should work and that if it's overheating it's the NVMe's problem. But they couldn't tell me what 4TB NVMe they have tested, nor what heat dissipation the enclosure is capable of. And PNY (NVMe vendor) said the stick can be used in an enclosure, but when I told them it was overheating they said it was designed for motherboards not enclosures. So far I can't find anyone who has tested and is using a 4TB stink in a USB C enclosure. No manufacturer, nor any consumer. So far there is no existance proof that this can work, and I'm not going to do their thermal engineering for them on it.

I have wasted somewhere between one and two days on this, done a round with tech support with two companies, and there are no solutions other than "point a fan at it". That's exactly the sort of crap engineering and crap products that I'm not going to waste my time on. I have now returned all the enclosures and the NMVe stick for refunds. I'll go back to 2.5" form factor SSDs. At least they seem to work.