I update the drivers and 3.05 BIOS.
Sadly, this problem is not be solved…
1 Like
Yes, just tried 3.05, the disk now cycles itself just sitting connected, without any transfer. (with my replacement AMD FW)
Well…
@Kieran_Levin ?
Confirming on my end as well that the issue remains on 3.05
1 Like
Thank goodness I bought a different laptop for daily driving since I can actually backup to my SSD that way. If only my AMD FW13 is still returnable.
1 Like
Power supply probably isn’t an issue per se. I’m now doing some sustained transfer test with external HDD, with this particular one being an HDD+Tiny SSD Hybrid drive. I assume this probably draws more power than my Gen3 NVMe in an enclosure. (although I do not have a device to measure)
Approaching 80GBs now, no issues.
Another update. Was just messing with the laptop before ending the night, plugged external drive to Port 2, experienced ‘cool down’ and can’t recognize drive as usual, I kept the drive connected and triggered ‘battery disconnect’ in BIOS, afterward I also held down power button for ~30seconds for good measure.
After waking the computer by attaching DC to Port 1, I noticed it seems like the external enclosure isn’t getting power at all. Disconnected external enclosure and connected android phone to Port 2 instead, no data and phone shows it’s charging the laptop. Therefore, it does seems like there’s some funky negotiation happening.
Maybe related to the post below?
(Recovered to status quo by disconnecting everything, and trigger ‘battery disconnect’ again)
You could also just use an ssd that is not some form of 970 evo XD.
Also dicounting power issues, anything but a 970 evo (on linux) works in my pretty power hungry asmedia usb4 enclosure, including a PM9A1 which is a lot more power hungry than the 970 evo.
Just did a full device read (512GB) at 2.8GB/s (the bandwidth cap is still there but somehow higher than last time I tested).
Was the 970 evo plus realtec enclosure issue ever reproduced on windows?
I use a GM7000 4 TB that’s made of an entirely different controller and entirely different NAND. It does not work, nonetheless.
TIL acer makes ssds.
What kind of enclosure are you running? Also is it not working on linux or windows or both?
The issue of the NVMe enclosure dropping out is very similar to what I’m running into with my iPhone 15 Pro Max that I posted in this thread: USB C is not compatible with iPhone 15 / Pro / Pro Max - #42 by Tyrael
TLDR: C-to-C cables constantly connect/disconnect; USB 3 A-to-C cable connects fine, but drops data link after a few minutes trying to backup phone; USB 2.0 A-to-C cable works fine.
I also have a TeamGroup USB 3.2 Gen 1 flash drive that is not seen at all by Windows when I plug it in, but works on other machines fine.
It seems more like the USB controllers between the AMD Framework board and what’s on the device aren’t playing nice.
1 Like
Hi ! I came across this thread and thought I could share my experience.
My setup
- Laptop: Framework 13, AMD Ryzen 7 7840, BIOS 3.05.
- NVMe/USB enclosure: ElecGear EL-2230C, JMicron JMS583 inside.
- NVMe SSD inside the enclosure: KIOXIA KBG40ZNS256G, known reliable and used as a boot drive for a few years without any issues.
- SATA/USB adapter: Sabrent ASM1351, paired with a Samsung 850EVO SATA SSD.
- Cables used: All known/confirmed good, tested multiple 2.0 and 3.0.
I use Windows To Go on a regular basis for gaming with the SATA/USB adapter and SSD. I find it convenient to separate my work space and my gaming space.
PCB view of the EL-2230C
Fault observed
I tried Installing Windows to Go on the EL-2230C (Windows 11 Pro) with a software called Rufus. I applied the image either from the Framework Laptop itself (from a Virtualbox VM) or from a Windows To Go (Windows 11 Pro) install booted from a SATA/USB Drive.
Depending on the case, I always end up with a corrupted image on the EL-2230C:
- Rufus fails to apply the full image to the disk and drops an error message.
- Windows To Go BSOD on start because of corrupt files.
- Windows To Go fails during OOBE because it corrups itself.
- Windows To Go fails after a few minutes of normal usage and does not to boot anymore: unrecoverable BSOD on restart like case 2.
This behavior does not occur at all with a SATA to USB adapter. I tried cloning my daily (known good) Windows To Go image from my SATA drive to the EL-2230C, but ended up in case 4 in only a few minutes.
What I tried
- Better cooling of the drive, adding thermal paste to the controller (the enclosure gets really hot).
- Several USB cables, including USB-A and USB-C as well as USB 2.0.
- Flashing several revisions of the JMicron JMS583 firmware on the EL-2230C (including firmware from other products).
1 Like
Yup, there are other folks having issues with one-cable Type-C portable monitors not working on AMD FW, despite the same monitor works fine on Intel FW, and other computers. Defintely seems like some controllers isn’t functioning as they should.
Yup, I use Rufus created WinToGo quite often. Even if an intact WinToGo is created on the external NVMe with another computer, and successfully booting the AMD FW with it, once the R/W gets intensive it will BSOD. Essentially the drive will become unstable or drops outright when activitiy becomes intensive, irrelevant to what the activity is.
Likely not the issue especially if the drive works fine on other computers. My enclosure can become unsafe to touch at the extreme, yet sustained file transfer, WinToGo creation, and WinToGo booting all consistently work fine with my Intel Framework.
1 Like
BTW. I think this is not USB-C’s problem. I use same enclosure and SSD with type-c to USB-A cable. I connect to framework official USB-A expansion card. The situation is the same.
7 Likes
hmm, not really. Same SSD, recently appears bad, speed 11MB/s, lots of error message…
insert into an RTL9210 (ugreen CM559) enclosure. it immediately goes
361681125376 bytes (362 GB, 337 GiB) copied, 374 s, 967 MB/s
362630086656 bytes (363 GB, 338 GiB) copied, 375 s, 967 MB/s
363569610752 bytes (364 GB, 339 GiB) copied, 376 s, 967 MB/s
364576243712 bytes (365 GB, 340 GiB) copied, 377 s, 967 MB/s
Are there any better firmware for ASM2364/2464?
The firmwares didn’t make any differences in my case.
I also didn’t have performance issues with the 970 evo it just didn’t hotplug on linux, when I pulged it in before booting it performaned just fine…
hmm with AS2464, plug before boot, the performance is good ~2GB/s. boltctl shows 20Gbps x 2 link.
But somehow if something happens to 970evo with AS2364, seemingly “triggers” some condition. using RTL9210 for some time, will somehow “heal” it.
No idea, maybe because of I added some capacitors to power of AS2364 enclosure?
Anyway, I can’t get x 2 link neither USB nor TB for devices I can reach.
With new 6.7.12 kernel and 3.05 BIOS, everything’s almost the same.
I have had this problem on a Framework 16 with a Sabrent NVMe SSD Enclosure on Alpine Linux with a Samsung SSD inside of it. It will work for 10 minutes then stop.
1 Like
Oh, so it does appear that this issue is applicable to both the AMD 13 and 16.
My previous correspondence with FW regarding this issue was a month ago, at the time they are still internally escalating it. I wrote to support again yesterday to check on the progress, I’ll update the post when (and if) I have more information.
Just curious. Did you get anything from framework team?
Nope, I have not. Neither me nor other community member(s) who’s already in engineering escalation that also reached out to Framework regarding this specific issue in the past few weeks received response.
1 Like