I’m interested in upgrading my Pocket Reform. Am I correct in that I would need:
MNT RCORE RK3588 Upgrade kit for MNT Pocket Reform, along with Intel Gig+ Wi-Fi 6E Module, and a Flex Antenna to make up a complete system? (from Crowdsupply, I’m in the US)
If so, I have a few questions about my current system so that I don’t spend good money after bad.
There are a few problems that I’ve been having that have me worried. I’m going to go through them with the hope that the new CPU will resolve them.
Video output issue: The built-in video was only showing in low-res (1024*768) after I upgraded kernels. Until this afternoon, but it has worked temporarily before. Puzzling, but I expect that it might return.
1b) In the process of troubleshooting the above, it was suggested that I boot off f an SD card. I flashed it properly, but it never seems to boot from it, the behavior is strange. (I have a thread on this, will get it if needed.)
Battery levels. Charging seems to have gotten inconsistent, sometime it is plugged in, and will just stop charging. Haven’t seen a pattern to it yet. Also, the battery levels don’t read within the OS any more, seems stuck at 6%. I see a lot of “power_supply BAT0: driver failed to report `voltage_now’ property: -16” in dmesg. The keyboard display seems to work (mostly, see below)
Keyboard controller issues. The keyboard seems to stop working periodically, again, haven’t been able to pin down a pattern, but sometimes as often as every 2-3 boots. Keyboard will not respond to system, and sometimes will not allow me to reset (hyper-enter, and then r) This is after I flashed firmware, both on the 20250623 version. Frequently, if I check, it has a T in the upper left corner of the display, blank rest of the top two lines, and the HID versions below. If I try to quickly enable the battery display right after booting seems trigger the issue. Sometimes the keyboard will light up mostly white, with some keys different colors. Very nice looking, but the keyboard is then unusable.
So, I’m thinking that the former two problems will likely go away with the upgrade, but I’m not sure on the former.
So, I’m inclined to buy the board, do the update, and start clean, if it’s thought that it would work correctly with the rest of my system.
No, I don’t think upgrading will help. I do not have the first issue you’ve mentioned but I am experiencing the second and third on my upgraded rk3588 pocket reform. The keyboard firmware is a bit unstable (I’m on the same version) and causes, I think, both problems. Sometimes I need to do a hard reset by toggling the power switch because the keyboard system is stuck; the keyboard works but the battery status is stuck on 0% and gives no indication of the power plug being connected (normally the -1 something amperes woobles a bit which i think is normal and seems to indicate it is charging) and the os reports 6% (coincidence?).
I do know that this is a known issue and there has been considerable effort put into finding a solution. However, as of late, I don’t believe a solution has been created. Something is just causing the controller to freeze. Sometimes you can reset. Sometimes you have manually reset by pushing the actually controller reset, and sometimes, the ONLY thing that works is powering down and flipping the power switch on the side.
The thing that makes this more difficult is that this issue seems a little random, in that some people have it, and some don’t seem to have any issues.
I have even received a brand new keyboard for mine (thanks MNT), and sadly it almost immediately did the same thing as the one it replaced.
My RCORE upgrade just shipped. That will be the first that I will mess with the Pocket again. I think your questions are well founded.
Edit: I just spoke with Josch and things look real good so far that this issue is solved in the most up to date firmware. I will look at my Pocket when I get home and try to report more on the specifics of this then.
I just installed the 23 June system controller firmware and the 17 Jun kb firmware. Both are from this year. So far it seems that the kb issue has been fixed. I have already psuspended several times. What I found was that if the led flashing while suspending was enabled it would hang the keyboard, but not crash it. It just wouldn’t reach the OS. You had to reset the kb. However, if the leds were not flashed while psuspending then it would work without issue. I have not had any manualk resets that I needed to do, so this has been a big improvement for me.
I used to have keyboard problems with my RK3588 Pocket Reform but since flashing the latest firmware I did not experience any issues anymore. This is also what other users observed. On the other hand, there are clearly some for which the latest firmware does not fix their issues. We talked about that in IRC yesterday because another user (“optimal”) with pocket reform keyboard issues showed up: 2025-09-08.log
As you can read in the log, others like me and @zeha do not see these issues on our devices. On classic Reform keyboard there were some issues (in the past) which correlated when having the OLED display active but user “optimal” said that they did not observe a correlation between the pocket keyboard issues and the oled. Minute chimed in:
in terms of strategy, what we could try is make several builds with certain features turned off and see if it behaves differently, but it’s a bit of a sledge hammer approach. like: pio+led stuff, oled i2c, trackball i2c. maybe it could also be possible to use the second core for debugging in case the first one crashes, or in general look into crash handlers with stack trace capability. and/or wire up swd and try to find a good reproducer
You can read the full conversation in the log above.
So yeah, issues seem to persist but they are hard to reproduce and do not affect everybody.
Sorry, I messed up. I do not have keyboard but power management problems. So it’s the readouts in on the OLED display about battery status which are stuck after (and reporting 6% to the OS) a while. So it’s the “sysctl” firmware which is stuck?
Thanks for the update @rwv and glad to hear @2disbetter that the current keyboard firmware appears to be stable for you.
For me, the keyboard is stable on version 20250623. I also run the 20250623 release of the System Controller firmware. If a keyboard is unstable on 20250623, it can be a damaged RP2024 chip, a bad internal cable, or an old version of u-boot on RK3588. If your keyboard sometimes doesn’t work at boot (when you have to type in a password, for example), it means that the command for toggling the USB hub chip reset is missing in u-boot. In that case, please do: sudo reform-flash-uboot emmc
Since adding that fix to u-boot, I’ve never encountered the keyboard not working on boot again over hundreds of boots.
Regarding System Controller, there are still situations where it can hang, which we are investigating. In the meantime, the FW has a watchdog feature enabled that recovers from this thang after 10 seconds. You can usually see that the system controller is hanging if there’s a “T” letter in the OLED display in the top left position when you’re issuing an OLED command from the keyboard. (T = timeout).
Situations/causes for Sysctl hangs that I know of:
I2C bus is busy forever, which can be caused by misbehaving charger chip. Rare, but can happen if you do a Sysctl flash update at exactly the wrong moment. Recovery: unplug and reconnect batteries. To work around this, we should make sure all i2c functions are using timeouts and test @zeha’s MPS watchdog code. I might also get a HW fix for this into Charger Board V2.
Certain updates to RGB LED backlight from Linux code (probably what @2disbetter is experiencing with psuspend). Cannot reproduce by changing backlight with Hyper+trackball. Cause yet unknown. Reproducer wanted.
Directly after shutdown (poweroff) of SoC, happens for me regularly, cause unknown. Recovers by itself after 10 seconds due to watchdog. Could be related to SPI (maybe an endless loop in pico sdk).
I was not aware that the u-boot version plays a role here.
Currently, RK3588 Pocket Reform machine.conf has EMMC_BOOT=warn because you can potentially soft-brick your unit if flashing the wrong thing to emmc. With the current version of reform-tools in the repos, reform-check will not tell you that your u-boot version is outdated if EMMC_BOOT=warn because why notify about an outdated version on a platform where you “should not” flash a new version because it’s dangerous? I think we talked about this in IRC back then but it has been a while. I now changed this logic in this commit:
With that, reform-check will tell you about an outdated u-boot version on emmc even if EMMC_BOOT=warn but also print a small note of caution. I guess given what you said above this change has your blessing? If yes, it will be part of the next reform-tools release
There are many times that if I change the backlight with hyper+trackball, it will hang and get stuck. I’ll see if I can reproduce it with any regularity.
OK, I wonder if this is a 5V power related instability (LED change related power surge), but it doesn’t 100% check out because the rp2040 is powered by a separate 3v3_standby rail over a separate cable (UART connector), and instability on this rail would also cause sysctl to crash. So maybe it’s related to the PIO updating function.
Maybe just a bit more nuisance here but when I notice the keyboard hanging, it is when automated led changes are done from the terminal. If I just do one, for example, to turn on the keyboard to green at a value of 50 that works with out issue. But if I am trying to pulse the leds on the keyboard where they start dim and grow rapidly bright and back to dim over the course of 6-8 commands, this is when the keyboard will freeze. Now this pulsing happens when starting or stopping psuspend. I’m not sure that works on the RK anyway, so take that with a grain of salt.
A combination of updates, and not using a connected usb3 hub to charge, seems to have taken care of most of the problems. So, back to the first part if this:
Am I correct in that I would need:
MNT RCORE RK3588 Upgrade kit for MNT Pocket Reform,
along with Intel Gig+ Wi-Fi 6E Module, and a Flex Antenna to make up a complete system? (from Crowdsupply, I’m in the US)
With the RCORE upgrade kit, you receive the board to replace the stock imx8 board. The RCORE has not wifi so you probably want to include that in the order. You receive the replacement board, and optionally the wireless card and the antenna. That and some minutes of your time and you have your computer upgraded. Note that the internal storage is replaced with the board so you will have a fresh installation if you are not currently using a NVME storage.
I ordered this all months ago so I received the first version, which does not have bluetooth. I think the new one that is currently being shipped does include bluetooth.
I thought everything was made clear on the Crowd Supply page, I apologize if that’s not the case. In any case, you ordered all the right items that you need.
