Question *STABLE* NVMe - USB Adapter?

[NewMaxx]

LOL, wish I'd seen this thread a few days ago. I ordered an Orico TCM2-C3 enclosure (JMicron chip), and can confirm it's flakiness. It works OK on USB 5Gbps, but connected to the 10Gbps port it hangs and is very buggy. Using a WD Blue SN550 500GB ssd, max speed I saw was 700MB/s read and 500MB/s write. The controller on the ssd gets burning hot as well. Oh well, should've remembered the USB to SATA days, JMicron was the worst back then as well.

I just ordered this (RTL9210 chip): https://www.amazon.com/Adapter-Realtek-RTL9210-Compatible-Samsung/dp/B07Z976BVK/

Should work much better. I don't need an enclosure, just using it for file recovery and cloning.

Yeah, the RTL9210 is the best of the 10 Gbps options for sure. I recently picked up and did a "quick look" at the JMS583-based M2X and I had trouble with it on one of my two test systems when it came to 10 Gbps port performance.

 

[samboy]

Update from my end........ even the RTL9210 controller is not completely stable for me and seems to drop out every couple of hours or so (both Intel and AMD platform). I have a 2TB Inland nVME drive inside - may turn out that some drives work better than others?

I'll double check the thermal paste; but one of the highlights was that RTL controller runs cool and it has dropped out when no obvious access is happening to the drive (i.e. it's not being triggered in the middle of some huge copy or ....)

My unfortunate conclusion is that none of these 3 controllers are reliable (at least in conjunction with a 2TB Inland nVME) at 3.1 Gen 2 speeds.

I'll stick with the RTL but only use the 3.1 Gen 1 ports; no real world impact in my case - just annoying that it can't operate reliably at Gen 2 speeds.

 

[NewMaxx]

What's the Inland reporting for temperature? Although it's difficult to judge if the controller/bridge chip itself is overheating. In my case the drive works on some machines and not others, so it's a bit unreliable.

 

[samboy]

What's the Inland reporting for temperature? Although it's difficult to judge if the controller/bridge chip itself is overheating. In my case the drive works on some machines and not others, so it's a bit unreliable.

This is what I see from Crystal Disk Info:-

----------------------------------------------------------------------------
(3) PCIe SSD
----------------------------------------------------------------------------
Model : PCIe SSD
Firmware : ECFM22.4
Serial Number : 19082120480101
Disk Size : 2048.4 GB
Buffer Size : Unknown
# of Sectors :
Rotation Rate : ---- (SSD)
Interface : UASP (NVM Express)
Major Version : NVM Express 1.3
Minor Version :
Transfer Mode : ---- | ----
Power On Hours : 286 hours
Power On Count : 503 count
Host Reads : 924 GB
Host Writes : 1743 GB
Temperature : 25 C (77 F)
Health Status : Good (100 %)
Features : S.M.A.R.T.
APM Level : ----
AAM Level : ----
Drive Letter : P:

-- S.M.A.R.T. --------------------------------------------------------------
ID RawValues(6) Attribute Name
01 000000000000 Critical Warning
02 00000000012A Composite Temperature
03 000000000064 Available Spare
04 000000000005 Available Spare Threshold
05 000000000000 Percentage Used
06 0000001D93CE Data Units Read
07 00000037CCF2 Data Units Written
08 0000012A2928 Host Read Commands
09 000002800E35 Host Write Commands
0A 00000000002E Controller Busy Time
0B 0000000001F7 Power Cycles
0C 00000000011E Power On Hours
0D 0000000001A5 Unsafe Shutdowns
0E 000000000000 Media and Data Integrity Errors
0F 000000000000 Number of Error Information Log Entries

You have a good point that there may be an issue with the Inland SSD itself and I may need to pop it onto a PCIe host card to see if it gives any issues there.

What would be the best way to test it? (Other than using it that way )

 

[NewMaxx]

Hmm, temperature looks okay (12A = 298 = 25C most likely?) but you can track it under use. Obviously the number of unsafe shutdowns is more concerning...

 

[samboy]

Hmm, temperature looks okay (12A = 298 = 25C most likely?) but you can track it under use. Obviously the number of unsafe shutdowns is more concerning...

Thanks

This is the same drive I have been using with all 3x controller cases/chipsets that I have. I guess each "disconnect" results in an "Unsafe Shutdown" .

I think the next step is to rule out the SSD being an issue. I'll see if it is reliable at USB 3.1 Gen 1 speeds first (and I can keep using this as an external drive). If it is reliable then I guess I'll need to use it as a proper nVME internal for a while (and make sure reliability holds up with higher performance)

 

[BrallyO]

Update from my end........ even the RTL9210 controller is not completely stable for me and seems to drop out every couple of hours or so

Orico has a new RTL9210 adapter which advertises an interesting feature: It will auto-shutoff after 10 minutes of inactivity. I wonder if yours does the same.

Can you set a clock/timer to go off after 9 minutes and 30 seconds, and then see if the drive shuts off?

Here's a link to the Orico drive's product page: http://www.orico.cc/us/product/detail/6758.html

This may be a feature of this RTL9210 chip. I wonder if there is some way to reprogram the controller to not shut down after a period of inactivity, or to change the timer.

 

[samboy]

I've had the drive plugged in on USB 3.1 Gen 1 all day and there hasn't been any random disconnects. It's definitely been idle for more than 10 minutes and still present.

Just seems that USB 3.1 Gen 2 is not ready for prime time yet.......

 

[cyberjedi]

Update: my JMS583 adapter died after 2 uses (mentioned in post #8), it was disconnected while powered on, although there was no active data transfer going on. It has never worked afterwards. At least the NVMe SSD survived... I have sent the adapter back to Amazon, got a refund.
Now I have ordered 2 more adapters from Jeyi, as they seem like a reputable company and promote stable dual power delivery, etc., in their products:
https://www.aliexpress.com/item/33035709205.html RTL9210
https://www.aliexpress.com/item/32945237657.html (Black AC Dual, JMS583)
I will report back about their stability in April... I'm curious what we're gonna find, as this company provides firmware updates also.

 

[Monstieur]

The new 10 Gb/s external SSDs all use the ASM2362 with NVMe controllers - Samsung T7, SanDisk Extreme Pro.

 

[NewMaxx]

The new 10 Gb/s external SSDs all use the ASM2362 with NVMe controllers - Samsung T7, SanDisk Extreme Pro.

Yes, and the USB3.2 Gen2 (2x1) drives (20 Gbps) use a variation labeled the ASM2364.

 

[cyberjedi]

I'm happy to see the move away from JMS583, Jmicron was never known from their reliability...
Let's collect experiences on the 3.1 and 3.2 cables, as they seem to play a crucial part in determining data transfer speeds.
I also see an interesting phenomena: when using a USB C to USB A adapter cable with JMS583, the transfer speed swithced down to USB 2.0 - and I don't fully understand the underlying reason... I have tried several cables, same thing.

 

[Monstieur]

Yes, and the USB3.2 Gen2 (2x1) drives (20 Gbps) use a variation labeled the ASM2364.

Do you recommend getting an ASM2362 or RTL9210 enclosure now? I suspect companies will investigate ASM2362 stability on their platforms due to its prevalence in popular external SSDs.

 

[NewMaxx]

Do you recommend getting an ASM2362 or RTL9210 enclosure now? I suspect companies will investigate ASM2362 stability on their platforms due to its prevalence in popular external SSDs.

RTL9210 seems ideal but I haven't worked with it yet.

 

[samboy]

A small update from my end and inspired by BrallyO's suggestion

Turned out that regular USB 3.1 Gen1 was not completely stable either (only saw twice).

Looking at the windows logs there was a "Reset To Device..." error which usually means a buggy driver or .....

I found this Link which exposes power options for the "AHCI Link Power Management " in the Windows 10 power settings advanced area.

By default it seemed to be set to HIPM+DIPM and I changed it to HIPM.
I'm back to using USB 3.1 Gen 2 again and have not seen any disconnects for the last couple of days; which is definitely progress but the jury is still out. The drive doesn't seem any warmer; but presumably the drive power management is not as efficient without the DIPM setting.

Suspicion is that the RTL9210 has some incompatibility issue with the DIPM setting (could well be an intel chipset driver issue as well?).

I didn't intend to buy a "project" (I like things to just work) but this may be helpful to others and may help with the ASM2362 and JMS583 chipset as well?

I'll post back any updates (I need to test to see if this helps for AMD x570 as well)

 

[NewMaxx]

I didn't intend to buy a "project" (I like things to just work) but this may be helpful to others and may help with the ASM2362 and JMS583 chipset as well?

I'll test this on JMS583 today if I get the chance! Thanks.

edit has no impact on JMS583

 

[eddieobscurant]

If you guys need the latest firmware for jmicron or realtek here they are from jeyi link
They should probably work with other brands with the same controllers but you try it at your own risk.
I have the jeyi i9 (jms583) and it upgrades to fw 0208
I downloaded the one with 5minutes (there was no delay though)

 

[razel]

Thanks. Just updated my Pluggable USB-NVME (JM583) based. It was on firmware 204, now 208. No performance difference noticed. Unlike others, I had no issues with my JMicron. I probably will now. The Pluggable comes with thermal pads and I only put that on the controller. It only heats up when you copy files to it. It's fine when applying game patches. It's probably seen 300hrs+ of gaming usage as an external drive for the xbox 1x which shockingly does support USB 3 at 5Gbps and does transfer at those rates. I tested when using two USB3 drives and copy a game from one USB-SATA-SSD to the Pluggable.

I also just bought the Orico RTL9210 based version for my other Intel 660p. I'll report if anything is odd with that one once I get it.

 

[fantom3l]

For those interested i have uploaded RTL9210 firmware v1.20.12 from 12.02.2020 + v1.18.1 from 28.10.2019 and upgraders for x86 and x64 ( read the note inside) along with examples of configuration for upgrade from different devices.
You can find thepackage HERE .
Used already on 2 devices i have and i did not brick mine. As you will see in configuration you can change name of your device+ stanbytime .... + a lot more.Just edit the config as you wish, if not use the generic one.
Also for JMS583 i have uploaded all firmwares i could find during time HERE including the latest 0208 with different configurations for standby time including the no standby one.

 

[samboy]

For those interested i have uploaded RTL9210 firmware v1.20.12 from 12.02.2020 + v1.18.1 from 28.10.2019 and upgraders for x86 and x64 ( read the note inside) along with examples of configuration for upgrade from different ........

Thanks for uploading this; I've upgraded my drive successfully.

The "AHCI Link Power Management " post I made earlier seemed to have helped; but the "device disconnecting" issue still occurs; just less often.

I'll try running with the new firmware and report back later........

 

[ppptech]

Been lurking in this thread for awhile. I bought a Sabrent EC-NVME based on the JSM583 bridge chip and I can't use it on my Razer Blade 15 Advanced. Works fine on an Alienware 13 R3 I tested it on and my Macbook Pro though. Maybe some sort of problem with newer chipset/cpu? Anyone else have similar hardware?

 

[eddieobscurant]

I found out that if i connect my jms583 on the usb ports that come from the cpu, i don't have any disconnections, but when connected to ports from the chipset i have huge issues.

Could someone verify that? You could check which usb ports come straight from the cpu and which from the chipset through your manual. I have a ryzen am4 motherboard

 

[ppptech]

You could look up what it looks like in usb device tree viewer. I only have issues with it on USB 3.1 gen 2/Thunderbolt port. My USB A ports work fine at lower speeds.

 

[razel]

Just received my Orico PFM2-C3 NVME-USB adapter with the RTL9210 chipset today. Used it for 3 hours as a x1x external drive and it's much cooler than the Pluggable Jmicron based. Sadly out of the box it's firmware didn't support TRIM. If you bring up 'Defragment and Optimize Drives' in Windows, it detects it as an HDD instead of SSD.

Thanks again to famtom3l, I updated the firmware from 1.16 to 1.20 and it now supports TRIM. So far so good after the unoffical firmware update. I'm pretty sure Orico could release one themselves, if they don't I probably won't buy from them. I'll wait for a competing product from StarTech or Sabrent since they have released firmware updates for their SSD to USB adapter products. Sabrent surprized me most with 2 updates to their cheap $10 SSD to USB adapter.

If you plan on buying be very careful with it. It's very slim and you can easily nick the controller or NVME chips when assembling.

I also wonder about shielding. Both Pluggable and Orico have metallic cases. Seems like the Pluggable one isn't shielded to USB ground while Orico appears to be which is good.

 

[cyberjedi]

Some news: the ASUS ROG Strix Arion, which is the most expensive solution on the market, uses the ASMedia ASM2362 chipset (and thermal pads). Good to know...
https://www.tech-critter.com/review-asus-rog-strix-arion-m-2-nvme-ssd/ says:
"For our case here, the temperature is hovering at 34~38°C when it’s left idle and topped at 48°C when we performed a file copy loop for 1 hour under the ambient temperature of 28°C. " - so it appears to be stable!
Another source: https://pricespy.co.uk/computers-accessories/storage-media/external-enclosures/asus-rog-strix-arion--p5214561/reviews
"Tested on ASUS Zenith Extreme 1 with Threadripper 1950X. Tested on both AMD Native USB 10Gbit Controls and ASMedia 3142. Did not work stably on the native controller, first unstable device." - not so stable after all on every platform...