Hello,
I have a MNT Pocket Reform with RK3588, the MediaTek Wi-Fi, running the default Debian with Sway. I recived the device in mid August 2025, with the “redesigned RK3588 adapter board”.
The only change that I made to the hardware is that I put a WD SN5000S NVMe in the m.2 slot (Key M).
Since last week I did experience issues with the system controller. Below roughly 70% charge, I had a “T” on the OLED but only when using the OLED-Menu. In this state the device could not be started. These problems disappeared while the charger was plugged in. Sometimes I had to cycle the power switch.
Firmware Update
This Tuesday, I flashed the newest firmware to both system controller and keyboard using the provided “oneliner scripts”:
curl -sfL mnt.re/update-pocket-sysctl | sudo bash -
curl -sfL mnt.re/update-pocket-keyboard | sudo bash -
I had a higher stability, but only for a brief time.
Yesterday I experienced the device getting highly unstable without a charger. At about 98% charge the system controller got a timeout (“T” on OLED) and debian crashed a few seconds afterwards (screen freeze, burn in, restart). After the reboot the system was unstable (seamingly random crashes while booting or using cli).
I decided to cycle the batteries from 100% to 0% once and fully load them.
After draining the batteries and fully recharging, I only see 0% charge and 0.0V in both batteries all the time.
Furthermore the USB-C PD negotiation does not work sometimes, I have to cycle the power switch to get it working.
Moreover, since this morning the Pocket seems to have destroyed my WD SN5000S NVMe which is not recognized anymore by the Pocket or any other computer. Using a different SSD (WD Black SN770) in the Pocket works, so the slot or the PCIe bus seem to be fine.
More about the SSD and PCIe
With the (now defective) SN5000S plugged in, I see
pcieport 0004:40:00.0: bridge configuration invalid ([bus 00-00]), reconfiguring
pci_bus 0004:41: busn_res: [bus 41] end is updated to 41
in sudo dmesg about every 15 secounds. Still lspci reads the SSD fine:
0003:30:00.0 PCI Bridge: Rockchip Electronics Co., Ltd RK3588 (rev 01)
0004:40:00.0 PCI Bridge: Rockchip Electronics Co., Ltd RK3588 (rev 01)
0004:41:00.0 Non-Volatile memory controller: Sandisk Corp WD PC SN5000S M.2 2280 NVMe SSD (DRAM-less)
Other commands like lsblk or ls /dev do not show the SN5000S or partitions of the device.
The Slot (Key B) with the MediaTek Wi-Fi appears to be fine. Wi-Fi is still working.
Right now, the Firmware is 20251030 for both system controller and keyboard. The kernel is 6.17.7-mnt-reform-arm64 from earlier this week, although updating to 6.17.8 a few hours ago did not solve any of my problems and the SSD remains dead.
For the timeouts from the system controller:
USB-PD does not work consistently. I checked with a USB-Multimeter that most of the time only 5.0 Volt are reaching the device. The red LED is not glowing.
As suggested on https://support.mnt.re/, I did try to bridge the charger port and run the device solely via USB-C, which appears to fix the timeouts of the system controller. Also then I get the 19.7 V USB-PD. But the red LED does only show up very briefly after plugging in the cable and is also not shining while Debian is running (only the blue LED is glowing). In fact the red LED only glows once for less than a second after plugging in the charger.
Short version of events in chronological order:
- Unstable / unbootable below ~70 % charge
- Firmware update → did only temporarily fix
- unstable at ~98 % charge
- I decided to fully cycle the batteries
- after cycling once, always 0 % charge, 0.0 V
- SSD not recognizable
- Pulled 6.17.8 kernel via apt
- checked the charger board as suggested by https://support.mnt.re/ and bridged the charger port with a jumper.
My questions are right now:
- How to fix the battery gauges (voltage is always 0.0 V, for the individual batteries and for the total)? Current is fine.
- What could be the root cause here?
- Is it still safe to use the NVMe Slot (Key M)?