Even with the latest qcacld2 driver, which does not generate the previous errors in the system log (thanks for that), the signal strength is still bad and recent disconnects occur in the “fringe regions”, where other devices work flawlessly.
I checked with 5 different devices lying next to the Pocket Reform and any one of them had more than 10 dBm better RSSI while using the same band on the AP. Another device in the next room even had the same RSSI. So there seems to be a problem with reception.
I saw that the Wi-Fi antenna sits on the standby-switch and trrs-jack board. I tried several other antenna types and locations (inside the case, outside the case, glued on the non copper part of the top cover, etc.), with no change in actual reception.
After I let the device stay powered off for while, I noticed that the measurements were much better. So I checked the thermals on the Wi-Fi-Chipset, which rose quickly and where then quite high, while not having a cooling device or heat conductive material attached. I then put a thermal pad on top, connecting it to the top cover and the signal strength shot up and was almost in same ballpark as the other devices, which seems to point at a thermal chipset issue. I did not yet do intensive transfers and/or long-term stability tests, but this seems to be promising.
I cut the pad to be exactly the size of the chip, so that it does not contact with the surrounding parts. I tested the top cover contact, by sticking the pad on the chip and see if the pad would adhere to the top if opened, which it did.
I am also reporting some wifi issues, after adding the thermal pad, it went from “few minutes” to “a dozen of minutes”, but as soon as the device gets a bit hot, the wifi gets disconnected and does not reconnect until a reboot.
Even when standing close to the AP.
edit: I moved the system to the SSD (from the eMMC flash) and it seems to be better so far, I will report back later today.
Yeah, my pad workaround is far from perfect as the CPU apparently generates just to much heat. When the thermal capacity of the top cover is reached, the Wi-Fi modules does tends to fall-back to the old pattern.
You could try to force the CPU freq to the lowest possible values.
This is kinda related to this topic but I wonder whether a custom backplate with many vias between the two copper layers would help dissipating the heat by increasing the copper surface.
Small update, I ordered some better thermal pads and with the thermal pad + having the OS on the SSD, I am able to get the wifi working for multiple hours now.
My guess is that having the OS on the flash memory makes the overheating issue worse, but I may be wrong.
I noticed that sometimes when the laptop is idle, the wifi disconnects/reconnects instantly when trying to load a page in Firefox, but that may be software/driver related?
Anyway, mostly sharing if it can help people having the same issue.
that would make sense, as I also run the root partition from encrypted NVMe. That would explain, why I had a much better experience with only a pad already.
I can confirm the random disconnect/reconnect problem while initiation a bigger transfer after a longer idle period.
Yep, have noticed the same thing on my Pocket, and indeed moving the OS to NVMe seems to have helped (though not entirely). I have a thermal pad on the way to try that out.
I’m seeing much lower wifi range with the 8M Plus compared to the A311D, locations where the A311D maintains connection, the 8M Plus will cycle between dropping out and connecting.
Does the A311D Realtek Wi-Fi chipset gets anywhere near the temperatures of the QCA chipset of the i.MX8MPlus?
Can you measure the temperature difference between the two?
The latest driver update for the QCA seems to limit the maximum transfer speed in 5GHz to 96MBit/s and therefore alleviating some heat and running much more stable. Can anyone confirm that?
I only have the internal thermal sensors to go by, but it generally seems the idle temps while connected to wifi of the a311d are less than the mplus - about 5-10C lower, total power draw is also lower on idle 5W vs 7W. I ran the mplus with the thermal pad on the QCA chip as well - though I’m not convinced my 2mm thermal pads were making good contact with the case lid
That would make sense. It already made a difference to throttle the MPlus to 1,2GHz to get it to drop below 70°C. If the A311D is that much cooler, then the wireless chipset will likely be also.
I’ve just added a thermal pad without transferring OS to nvme (yet, I’ve got one installed I’ve just not done the needful) and it’s not really helped, so I’ll try doing the transfer later and report back as to whether it helps with the wifi.
sadly that makes sense, as the thermal pad usability dwindles fast as the CPU heats up the topcover quite fast and then there is no heat capacity left to be useful for the Wi-Fi. My experience is, that it best worked right after startup, when the top cover is not yet hot to the touch.
