I’ve just done the migration to SSD and I was able to get a better connection initially but once the system heated up again the signal dropped out.
I’m getting the impression that this isn’t going to work unless heat dissappation can be improved (I’m no expert and would be very happy to be wrong)
Then you come to the same conclusion as me.
From what i’ve read i’d assume that there’s something wrong on the iMX8MPlus with downclocking and power saving features. It should not run on this high clock speeds when the system is idle.
I’ve found a topic at the NXP community about iMX8Mplus running hot in idle.
I just got my pocket reform (hooray!) and I think I’m seeing this issue: I can connect to wifi for a while (an hour or so?) after boot, then the connection drops and can’t be re-established. So, a few questions:
What is a “healthy” range for pocket reform system temperature and how should I monitor it? I have just been using the “sensors” command and going by touch. So far, after light use, the lid gets quite warm, “sensors” reports 65 - 70C on soc_thermal-virtual-0, and I can’t re-connect to wifi. Is the output of “sensors” trustworthy? Is a different command better or more convenient in some way? Should I get a thermometer gun?
How should I monitor the wifi for signal strength? Currently I’m using the TUI application “wavemon”, and this shows signal levels about 10dBm worse than an adjacent device. Is that the appropriate measurement to compare across devices? Is there a better command for reporting RSSI?
How should I log wifi disconnects, or search existing logs for disconnect events? Just trying to correlate with system temperature or maybe something else…
Thanks in advance for any help. I’m so delighted with the pocket form factor and would like to help figure out these wifi/thermal issues any way I can, even though it has been quite a while since I used linux.
Today I also encountered WiFi issues while running on battery in a building with two WiFi networks each at medium signal strength. Connections dropped a lot and switched between the two networks and when I resorted to using my phone hot spot, that also was not really stable even with the phone right next to the Pocket Reform.
(Still running the Pocket from eMMC, not the SSD)
I am running my system from the NVMe and just installed the thermal pad on the wifi chip. Will see how it goes!
This morning I still had wifi reception issues, and this is when the pocket reform just booted, so I don’t think overheating is the only issue.
See my first post, this is “correct”. The reception is, even when inside a moderate temperature range still around 10dBm worse than other devices lying next to it. When the device heats up, this gets worse as now the disconnects begin to appear and often times the reception is getting worse.
And to top it up, my device is quite hot even just after being booted up.
So yes, I agree that it is not only heat but more than that.
Ahh, fair enough! I must have misread, as I thought it was only 10 dB after getting hot.
WiFi issues seem to be getting worse. This morning my Pocket Reform didn’t connect to my home network at all, only to my phone’s hotspot. It wouldn’t even list the SSID of my router even though that wasn’t a problem originally.
I’m sorry that you’re experiencing these WiFi reception issues. I have to switch back to the i.MX8MPlus processor (from A311D) for further dogfooding to see how much of this issue I can reproduce in the wild. It might be a combination of antenna specs and antenna placement. The antenna was chosen conservatively to match the type and relatively low gain of the one that the WiFi module/iMX8MPlus module was FCC tested with, so we could use the modular FCC certification. You as users can change the antenna design on your own responsibility, of course. So I propose that we can find a better DIY upgrade solution here together. There’s also different ways that antennas can be mounted:
Then, one can also look into more software details like the regulatory db which might limit reception if not setup correctly (iw reg …)
Additional option that came to mind:
but isn’t the reception problem triggered by the cpu overheating as a result of not clocking down instead of a weakness in antenna design?
FWIW: related to the M.2 card suggestion, I have tried a B-key M.2 USB adapter (with a Netgear wifi USB adapter inserted), and a B-Key M.2 SD card adapter (with an SD card inserted), in the modem / WWAN slot and neither seem to be detected by the Pocket Reform - although, I’m not sure whether I would need to manually load modules to drive things in the slot… in the case of the SD card adapter the device itself has LEDs that come on indicating that it is connected in hardware terms. Looking at the schematic I wasn’t sure how generic an interface the M.2 slot is, though, it seems to be specific to WWAN / modem (I’m waiting on that to arrive right now)
(NB I have not yet had issues with wifi myself)
I have no hard data on any temperature issues influencing WiFi. So far, it was anecdotal evidence from this thread (EDIT: not saying it is not the reason, but so far it is AFAIK an unproven hypothesis).
Is that an ADT-Link adapter? @bnys just pointed me to them. I haven’t looked into the pinouts yet but even without a driver, something should show up in lsusb.
fair point - I only mentioned it as it seems like perhaps an easier fix with potentially wider benefits.
The temperature cause was only a correlation, and may be only that and not a causation. What led me to conclude that heats has a role, was that the runtime of the device with the corresponding heatup increased the occurence of the disconnects.
So we have two separate issues in my opinion. The disconnect issue and the low signal strength issue. They interact with each other but may not necessarily be linked together.
What is certain is, that the device has a very low signal strength even when cool, “open topped” and with other antennas.
I can observe that the longer disconnects seem to not really be dependant on distance/RSSI. But as the signal strength is quite low, some access points tend to disconnect the device when it reaches a certain RSSI value, which it does quite fast due to the low reception.
This leads to, in my opinion, an intermixing of both issues. The reception problem triggers disconnects, but these can be resolved easier by reconnecting and the heat (probably) triggers the longer disconnects, which tend to last longer or need reboots and so on.
I have a couple of very cheap AliExpress adapters - the USB one is marked ADT-Link on the board, so yes I guess so. I’ll have another play with these tomorrow if I can, but I’m fairly sure I tried lsusb. Will report back in some form, although we’re heading off topic a bit so I might start a new thread.
Okay so I just did a very unscientific experiment with a number of antennas that I have. I’m sitting in roughly the same place in my friend’s house, I have no idea where the AP actually is, so take this all with a pinch of salt. Also know that I’ve migrated to encrypted nvme and I’ve also installed the thermal pad.
For context here, around the 35% signal strength mark with, hilariously, the highest gain wifi antenna, I was getting less than 1Mbps whereas with ~90% signal strength with this stubby boi I was getting around 200-250Mbps.
So: an external antenna helps, but size and placement matter. I thing it would be entirely possible to just drill a hole in the side plate and mount the SMA connector into it, I just don’t have the power tools available right now.
For what it’s worth, I don’t appear to get many spurious disconnects with the onboard antenna if I’m close enough to the signal, so it’s entirely plausible that the hacks (migration to nvme, thermal pad) have helped on that front. I think the remaining issue is just that the reception with the onboard antenna is not terribly good.
If you want to repeat any of this experiment, be aware that the U.FL connector on the imx8mp is MHF1, but MHF4 appears to be more prevalent these days so don’t do what I did and accidentally buy the wrong SMA → U.FL cable.
