Question *STABLE* NVMe - USB Adapter?

Page 7 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

cyberjedi

Junior Member
Apr 4, 2010
23
13
81
I'm looking to buy a *RELIABLE* NVME-USB external adapter, that doesn't disconnect randomly, can keep up the close to 1GB/s speed.
So far all adapter's reviews contain frighteningly high % of reports of malfunction, random disconnects, sub-par speeds, falling to even under USB 2.0 speeds, etc.
There are 3 main chipsets in the market: JMicron JMS583, Asmedia ASM2362, and Realtek RTL9210 (no product has surfaced with the latter yet, so no REAL reviews, experiences yet UPDATE: it is on the market already).
Also ASUS has a new product, ROG Strix Arion, but there isn't enough feedback about it yet, and I mean real life feedback not just being able to run a few benchmarks, but real stress test to see if the device (adapter) can keep it up, without disconnecting or slowing down hard.
I would really appreciate any feedback from those who have GOOD experiences during stress-tests, like tons of random 4k writes with a stable speed, cloning 100's of GB-s at a stable high speed close to 1GB/s.
I know SSD's very well, so please stick to the point and don't start to educate me on how their speed falls after some writes, etc. I'm aware and this is NOT the point.
You can help if you can point me to a STABLE and RELIABLE NVMe-USB-C adapter that WORKS as expected.
Thank you in advance!
 
Last edited:

Andalu

Member
Jun 25, 2020
93
68
61
Will say few precautions again
1. always done;
2. even without the thermal pads applied I had however extract the the nvme card from the enclosure to have a better ventilation. Please consider that I also tried a benckmark for the JMS583 on a cold boot but I got the sudden disconnection (see the time);
3. This is the only thingh I haven't tried but having always gotten the following update message, I don't think there's much hope:
Updated Firmware 2.0.8.png
4. here the cable I used for both enclosures:
 
Last edited:
Jun 10, 2020
149
47
61
Please download latest magician app and firmware update your EVO 970 on a compatible desktop with spare pcie nvme slot and run tests with latest firmware.

also report your Evo temperature with crystal disk info immediately after running your crystal performance diskmatk tests
 
Last edited:

Andalu

Member
Jun 25, 2020
93
68
61
My both 970evo nvme have already the updated firmware (2B2QEXE7). The screenshots were taken immediately after the end of the tests with CrystalDisk Mark so the temperatures are correct as detected by CrystalDisk Info.
Please see the temperature (and the results) for the Sabrent Rocket on the ASM2362 enclosure:
 

Attachments

  • Image 010.png
    Image 010.png
    158.1 KB · Views: 56
Jun 10, 2020
149
47
61
JMMassProd.ini
Which one did you use??

I have removed the performance increase and did small (name and other) changes compared to jeyi i9 one. Also using the standby with 5 mins. Works best for me.

I have no other suggestions. Maybe there is a hardware issue causing a performance issue or a nvme compatibility issue??
 

Andalu

Member
Jun 25, 2020
93
68
61
@apoelistasaudi
My JMMassProd.ini is almost identical to yours of some posts ago, with StandbyTimer=5 and PerfShift=1.

I really don't know what else to think about. Most likely it can be an hardware issue related to the USB3.1 Gen.2 ports of my board but the weird thingh is that it occurs for the JMS583 only and for another USB3.1 to SATA3 adapter (VL715 chip) that sometimes is recognized by USB Device Tree Viewer and Aida64 as plugged into an USB2 port even if it is really connected to an USB3 port (and its benchmarking results are poor). Both of these software, however, detect the JMS583 correctly connected to the USB3.1 ports. But I don't have noticed same issue for the ASM362 enclosure and for other UASP devices.

Thanks for the support ;)
 
Last edited:
  • Like
Reactions: apoelistasaudi
Jun 10, 2020
149
47
61
I confirm that running tests above a set temperature on usb 3.1 gen 2,I think 66oC the results are low. Same for me too. Must be a safety feature.

Need better cooling. Usb 3.1 gen 1 is a bit cooler to keep nvme card cooler.

Will try to add a heatsink on back side if I manage to cut the transparent case. I run test with no a/c on and temperature rose to 67 on second consecutive test, later became 68.. from second test on I had poor results.

First test results were really high will post tonight a photo.

Your hardware behave same as mine. Maybe the 5 mins standby will be added in hardware in the future since that does not always kick in in software, and maybe a better temperature/voltage control in hardware to keep nvme card cooler.

To cut the story short, there is a high temp issue causing performance degradation.

Before the firmware update disconnects

After the firmware update, no disconnects and standby kicks in earlier (but sometimes does not get activated I believe)

Update. Seems that running 4 tests after replugging cable even with 68oC did not cause any performance degradation so what I have said above is not confirmed. Just happened. Don't know why I have no clue because now I have no performance problems. Maybe a new firmware will fix this in the future. However heatsink temp was pretty near DHW temp:).
 
Last edited:
  • Like
Reactions: Andalu

Andalu

Member
Jun 25, 2020
93
68
61
Glad to hear you have no performance degradation. Which nvme disk have you used?
Based on what I could notice with my nvme disks, samsung ones are more sensitive to heat (and more hot) than sabrent. In fact, using the ASM2362 and a sabrent nvme, after 10 benchmarks one after the other (without ever disconnecting the device or rebooting the system) the results were always in line with what you would expect from nvme disks, even if the temperatures reached a peak of 65 degrees. Proof of this is the screenshot I attached yesterday (https://forums.anandtech.com/attachments/image-010-png.24361/) and that refers to the 10th consecutive test performed. I think that high temperatures can be not only a problem for the enclosures but also for the nvme controllers.
 
Jun 10, 2020
149
47
61
Toshiba but still I am not happy with the nvme enclosure. Should have been cooler and still replugging usb is not a proper fix.
 

Attachments

  • diskmark gen 2 type usb 3.1 gen 2.JPG
    diskmark gen 2 type usb 3.1 gen 2.JPG
    99.1 KB · Views: 36

Andalu

Member
Jun 25, 2020
93
68
61
I should have the Orico M2PJM-C3 by next month so that I can also try the RTL9210B chip and maybe also adapt into it the JMS583 card in the hope to have clearer findings. But I ordered it only few days ago and it has to come from China.....
 
Jun 10, 2020
149
47
61
Engineering time. Will try to find a proper glue in the next few days to hold heatsink in to place.
Test 1 temp after first run on usb 3.1 gen 1. seems pretty cool (a/c on) 45 oC max temp.

Few minutes latter while crystaldiskinfo was monitoring temperature, this has risen to 57oC
 

Attachments

  • A4F9B262-7347-4138-BF3A-3408D6B965F6.jpeg
    A4F9B262-7347-4138-BF3A-3408D6B965F6.jpeg
    373.9 KB · Views: 50
  • 0302DC6B-3147-4D91-B9C7-15D9201575FD.jpeg
    0302DC6B-3147-4D91-B9C7-15D9201575FD.jpeg
    326.9 KB · Views: 59
  • new heatsink test usb 3.1 gen 1.JPG
    new heatsink test usb 3.1 gen 1.JPG
    98.3 KB · Views: 50
Last edited:
  • Like
Reactions: Andalu

Andalu

Member
Jun 25, 2020
93
68
61
@apoelistasaudi
Well done! :cool: It can be a good solution to better dissipate heat (it wasn't a great idea from Orico to use a plastic enclosure). I'm curious about the temperature of the test on a USB3.1 Gen.2 port as I can see from the image that the toshiba controller seems to be a bit hot too :)
 
  • Like
Reactions: apoelistasaudi
Jun 10, 2020
149
47
61
@apoelistasaudi
Well done! :cool: It can be a good solution to better dissipate heat (it wasn't a great idea from Orico to use a plastic enclosure). I'm curious about the temperature of the test on a USB3.1 Gen.2 port as I can see from the image that the toshiba controller seems to be a bit hot too :)

Thanks. I don't think I had a choice. :)

Have uploaded the usb 3.1 gen 2 diskmark and diskinfo/temp image few posts above (post 158), test was carried out before adding the extra heatsink. Max temp was 69οC so far. I believe now temp should be much lower than that. (Similar to full aluminium body)

5 minutes Standby feature usually fails to activate after carrying out a Crystal diskmark. Needs a better firmware I believe.
 
Last edited:

Andalu

Member
Jun 25, 2020
93
68
61
5 minutes Standby feature usually fails to activate after carrying out a Crystal diskmark. Needs a better firmware I believe.
I confirm that the standby does not activate from the end of the CrystalDisk benchmarking (also for me the standby is set to 5 minutes).
In that regard, the ASM2362 enclosure does even worse, staying warm and with the led on (so still powered) even if safely removed and even with the system off.
We're still a little far from getting the perfect enclosure :(
 
  • Like
Reactions: apoelistasaudi

cyberjedi

Junior Member
Apr 4, 2010
23
13
81
Today finally all 3 enclosures of mine have arrived.. With 3 different chipsets, for testing purposes.
 

Attachments

  • 106386400_628603151088644_2090827921757839185_n.jpg
    106386400_628603151088644_2090827921757839185_n.jpg
    259.9 KB · Views: 66
  • 83864267_738705206886851_4495422538160052744_n.jpg
    83864267_738705206886851_4495422538160052744_n.jpg
    84.2 KB · Views: 69
  • IMG_20200630_202708.jpg
    IMG_20200630_202708.jpg
    691.6 KB · Views: 69
  • IMG_20200630_203315.jpg
    IMG_20200630_203315.jpg
    457.2 KB · Views: 66
  • IMG_20200630_203617.jpg
    IMG_20200630_203617.jpg
    540 KB · Views: 59
Jun 10, 2020
149
47
61
Cyberjedi looking forward for your tests.

Andalu usually I dont have a standby problem unless I run either Crystal two programs (diskmark and diskinfo). (also the intel driver update and improvement program may interfere with the standby feature so uninstall). I wonder how the two programs even when they are closed, disable the standby capability of nvme enclosure. Maybe if we find how then we can fix or maybe a better firmware can tackle this issue???

Still I believe the problem with the enclosures is voltage regulation when idle. When on idle voltage should drop significantly to keep temp down. 5 min standby feature does a good job on this (cools enclosure and nvme card) until the enclosure drive is accessed again.

Maybe a better firmware can control the idle temperature?? or maybe a hardware revision or a board redesign?

Solutions until today:
1. For orico to add a second heatsink on back side to keep enclosure cooler. Otherwise dont buy the trasnparent poly carbonate enclosure, only buy a metallic one. Cost of heatsink about 15 Euros, You will need a dremel rotary tool.
2. Firmware 0208 corrects the disconnect issues
3. Firmware 0208 with 5 mins standby option cools down the enclosure sooner.
4. Uninstall Crystal diskmark and diskinfo when you are done with your tests, also unistall the intel driver update and improvement program to stop the app from waking from standby the nvme enclosure . Also for the USB suspend feature (power plan options) I think disable as well when running laptop on mains power. With above options I have no unnecessary wake ups from standby i.e. without accessing the nvme disc.

I am 90% happy with the above since my nvme enclosure is now running cool. Hope future patches exist to fix remaining issues.

33 o C when on standby. (digital thermometer set to object).
 
Last edited:
  • Like
Reactions: kaefers

kaefers

Junior Member
Jun 12, 2020
9
7
36
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.

Hi
I recently bought a RTL9210-based Orico M2PJM-C3 (https://www.aliexpress.com/item/4001137870584.html).

I did come with firmware 1.21.10 apparently and worked fine, but kept disconnecting from my pc after about 10min idle time... I did checkout your uploaded files as well as https://www.station-drivers.com/ind...ository&Itemid=353&func=select&id=634&lang=en , but I just cannot manage to get anything flashed, always getting a fail.

Dump from the MPtool, default for the enclosure:

1593611242491.png

From the log.txt
1593611538730.png

Any pointers much appreciated :)
 

Andalu

Member
Jun 25, 2020
93
68
61
About the standby mode, I tried to connect two USB3.1 Gen1 to sata3 adapters (VL715 and ASM1153E chipset) and even after CrystalDisk Info closed, the standby mode worked properly for both devices after a few minutes of inactivity. The failure to activate standby mode for the JMS583 enclosure depends only on the chipset that has a problem going into a low power state under certain conditions. But I don't know which is the way to solve the problem.

Also for this reason, it will be interesting to know the results of cyberjedi tests, especially for the JMS583 chipset that, from the attached image, has the most recent revision "A2".
 
Last edited:
  • Like
Reactions: apoelistasaudi

kaefers

Junior Member
Jun 12, 2020
9
7
36
I ordered a total of four btw - just to see if there are any differences and because I could not help myself..
JEYI i9 GTR (RTL9210) - https://www.aliexpress.com/item/33035709205.html ,
JEYI i9 NVME (JMS583) - https://www.aliexpress.com/item/32874418125.html
ORICO M2PJM-C3 (RLT9210) - https://www.aliexpress.com/item/4001137870584.html
Asus Arion (ASM2362) - https://www.asus.com/Optical-Drives-Storage/ROG-STRIX-ARION/

So far I only got the Asus and Orico, still waiting for the JEYI. I also tested out a Sabrent Rocket Nano (JMS583), has some connectivity issues with X570, but main reason I sent it back is because the DRAM-less 2242 m.2 basically sucks.

Once I got all of them together, I will be running some synthetic benchmarks and more importantly imho real-world file transfers to see how they fare.
 
Last edited:

Andalu

Member
Jun 25, 2020
93
68
61
@kaefers
about the failure to update the firmware, could it be that version 1.20.12 (which is also older than the 1.21.10 already present) is simply only for the RTL9210 chipset and therefore not fully compatible with the RTL9210B chipset?
The attached log shows the error related to the different "Pkg Type" ("PCIE" for the 9210 chipset and "PCIE_SATA" for the 9210B). Also the "PCIE_PWRCUT_THRES" which should disable the pcie power after X minutes in idle mode seems to have a different value than the others listed in the various .cfg and .txt files of the "RTL9210_fw1.20.12.7z" package.

I've ordered the same Orico M2PJM-C3 enclosure but I'm afraid I'll have to wait many days before I get it.
 
Last edited:
  • Like
Reactions: kaefers

Charlie22911

Senior member
Mar 19, 2005
614
228
116
@kaefers
about the failure to update the firmware, could it be that version 1.20.12 (which is also older than the 1.21.10 already present) is simply only for the RTL9210 chipset and therefore not fully compatible with the RTL9210B chipset?
The attached log shows the error related to the different "Pkg Type" ("PCIE" for the 9210 chipset and "PCIE_SATA" for the 9210B). Also the "PCIE_PWRCUT_THRES" which should disable the pcie power after X minutes in idle mode seems to have a different value than the others listed in the various .cfg and .txt files of the "RTL9210_fw1.20.12.7z" package.

I've ordered the same Orico M2PJM-C3 enclosure but I'm afraid I'll have to wait many days before I get it.

From what I see, it seems the 9201 is NVME only, while the 9201B is a separate product that does dual NVME/SATA.
 
  • Like
Reactions: kaefers

kaefers

Junior Member
Jun 12, 2020
9
7
36
@kaefers
about the failure to update the firmware, could it be that version 1.20.12 (which is also older than the 1.21.10 already present) is simply only for the RTL9210 chipset and therefore not fully compatible with the RTL9210B chipset?
The attached log shows the error related to the different "Pkg Type" ("PCIE" for the 9210 chipset and "PCIE_SATA" for the 9210B). Also the "PCIE_PWRCUT_THRES" which should disable the pcie power after X minutes in idle mode seems to have a different value than the others listed in the various .cfg and .txt files of the "RTL9210_fw1.20.12.7z" package.

I've ordered the same Orico M2PJM-C3 enclosure but I'm afraid I'll have to wait many days before I get it.
From what I see, it seems the 9201 is NVME only, while the 9201B is a separate product that does dual NVME/SATA.

Thanks for your input!
I guess the 9210B is the NVMe/SATA chip and different from 9210 NVMe-only, and not fully compatible. I wrote the seller on AliExpress and Orico, asking nicely for a FW update... let's see.

Also, somehow, I did manage to get a partial update done, got to 15% then failed, but the dump from MPtool now reads as below...

So now I seem to have FW Ver 1.17.4 (I don't know where this is from, there is either 1.20.x or 1.18.x or 4.17.x but I have no 1.17.4 anywhere lol), and it also seems PCIE_PWRCUT_THRES is 0x00, so I should be good - need to test it out still if the enclosure works properly :)

EDIT:
Some images of the Asus and Orico enclosures. Clearly (as usual) the Asus is overdone and it's thermal pad is also not great (this is an old thermal pad, ordered new ones, the orignal Asus one did not allow me to close the lid properly, thermal pad too thick.. I really like the Orico, although the Asus does a better job in terms of thermals, but weighs like 3x as much.

EDIT2: Seems like the Orico enclosure still works, CDM7 about the same as before..
 

Attachments

  • Orico M2PJM-C3 partial_failed update.png
    Orico M2PJM-C3 partial_failed update.png
    52.9 KB · Views: 61
  • 01.jpg
    01.jpg
    713.5 KB · Views: 59
  • 02.jpg
    02.jpg
    668 KB · Views: 49
  • 03.jpg
    03.jpg
    1.1 MB · Views: 54
  • 04.jpg
    04.jpg
    589.1 KB · Views: 52
  • 05.jpg
    05.jpg
    1 MB · Views: 66
Last edited:
  • Like
Reactions: Andalu

kaefers

Junior Member
Jun 12, 2020
9
7
36
I have put the ORICO M2PJM-C3 (RLT9210) and Asus Arion (ASM2362) to the test with a 2TB Samsung 970 Evo Plus, OS Win10Pro, OS drive a Samsung 1TB 970 Pro, Asus X570 C8I.

Aside from the thermal advantage for the Asus, I would go with the smaller and overall very well performing Orico...
 

Attachments

  • 01.jpg
    01.jpg
    221.8 KB · Views: 73
  • 02.jpg
    02.jpg
    245.6 KB · Views: 62
  • 03.jpg
    03.jpg
    186.5 KB · Views: 60
  • 04.jpg
    04.jpg
    180.3 KB · Views: 51
  • 05.jpg
    05.jpg
    160.9 KB · Views: 52
  • 06.jpg
    06.jpg
    116.1 KB · Views: 47

Andalu

Member
Jun 25, 2020
93
68
61
@kaefers
thanks for your report ;)
Have you also had the opportunity to verify the temperatures reached with the two enclosures during the tests?
About the ASM2362 enclosure, could you please tell me if it still remains with the led on (so still powered) even if safely removed? I have this issue with my ugreen ASM2362 in all windows OS.
No other automated disconnection for the RTL9210B when in idle mode?
 

kaefers

Junior Member
Jun 12, 2020
9
7
36
@Andalu
Not really checked for the temps, I had the enclosures lying on top of my M1 in the exhaust air to make sure both are keeping cool and not throttling in any way.
During earlier testing, just the drive on a wooden desk, Asus (custom thermal pad) 55C or less, Orico prob got into the low 60s C, but didn't pay too much attention there either.
Drive has not disconnected since at all, and the blue led is still on. I RDP / VNC a lot locally and remotely, so some drive disconnecting would be rather annoying :)

[ Temps also very much seems to depend on the SSD itself. I borrowed an Adata SX8200 Pro the other day which ran quite a bit cooler than the Evo Plus, although I am not a fan of the SM2262EN on the Adata. It is a fast SSD but it was some random quirkly behaviour that I just don't see with the Samsung. Adata / SM2262EN SSDs seem to slow down quite a bit once you fill them up / once you run out of (the generous) SLC cache. The Evo Plus is the OS of my 2nd system, I ordered a 2TB WD SN750 for the external enclosure, as the speed is consistent, and that is all that matters to me, being limited by USB 3.1 Gen2 anyway. ]