Are these glitches expected? Is it possible there was some damage or connector which is seated poorly? Suggestions on what to take a look for?
More context:
Bringing up my DIY (Assembly Required) Reform 2 up for the first time. This includes the option for the NVMe and wireless card from CS. I noticed the occasional “color burst” glitch on the terminal during boot and some while in use. Seemed more often in graphical environment (sway). I am not sure which version was included on the included SD card, but I performed an apt upgrade which took a while and then the system failed to boot with a kernel panic.
I found references to “GPU glitching” with various scenarios, but not “general” glitches.
I positively correlated the glitches with wireless activity. After removing the wifi card and antenna, I now only get the (expected) GPU glitches on startup.
I was able to prevent further glitches from the wifi after re-installing the wifi with the antenna located between the trackball and the battery.
My hypothesis is the small non shielded breakout portion of the display connector is susceptible to noise from the wifi antenna. It is possible the arrangement of the wires is contributing (i.e. right angles on the differential pairs).
I am attempting to modify the title to indicate my hypothesis, but can’t figure out how to modify titles. I am guessing it is an “earned” reputation feature or reserved for moderators.
@bnys title update would be great to indicate wifi activity was causing the issue. It seems highly likely that this behavior is caused by the wifi activity on the antenna so close to the connector from my use of it today with nothing but small pixel noise on very dark screens (vs full blanking).
I updated my IPFS instance to v0.13.0 and the video seems to be available now.
Changed thread title. Have you tried shielding the display connector with kapton tape or a small piece of foil perhaps?
EDIT: I just watched the video. This is truly strange! I have two Reforms with the stock wi-fi card and neither behave like this. My personal model is even upgraded with a Laird antenna and does not do this. I wonder if there’s a fault in the display cable or if it isn’t properly seated.
I also sometimes get wifi-correlated GPU/screen glitches. I’ve been travelling a lot with the reform, and I’m beginning to think there might be location/wifi-specific circumstances at play. On one specific network I used for a couple weeks, the glitching was consistent and almost unbearable. While heavy web sites loaded- the screen would jitter/shake a lot. Now I’m home and back on my home wifi, and there’s no glitching at all. I didn’t make any software changes that I know of in between.
I did attempt re-seating the connector, but it did not have any effect. Note: the connector requires a good amount of wiggle because the connector is tight (which is good!). It is possible there is something about the display signal path which is “marginal”, but the case came assembled with the screen in it (and I haven’t disassembled it). There were some tiny blemishes on the screen which made me think it may be sourced from a “secondary” market due to supply chain issues (which I am totally fine with).
I do not suspect kapton tape will do much other than possibly holding the broken out wires in a “better” arrangement, but I do have some shielding tape left over from the Hackaday Remoticon “Basics Of RF Emissions Debugging Workshop” which I may try next week.
This may make sense as I believe there are scenarios where 802.11 infrastructure (e.g. Access Point) “informs” the client which power level to use (so they don’t use higher power than required). This is usually called (Dynamic) Transmit Power Control. Not sure if our cards/drivers support it though.
It could also be differences in the transmit power allowed based on the channel number and width. There can be drastic differences in maximum transmit power allowed depending on the (FCC) rules at the time of hardware/software. For some equipment I have, the range is very extreme (between 10dBm and 23dBm) depending on: channel, width, VHT, beamforming, and STBC settings
I had similar experience with apt upgrade. I saw lots of glitching during download stage which is in your video but minimal glitching during the installation stage. Presumably, there is significantly more wifi activity during the download stage.
I did try downloading a new package later to see if i could repro, I didn’t see any glitching but it was a pretty small package. I had not seen any more glitching after that, and no glitching on v2 at all. So maybe what I saw was different to what you have if it’s ongoing.
sysimage-v2 seems to have less reliable wifi than sysimage-v3, so maybe v3 is using a higher transmit power and that’s why I don’t have any glitching on v2, but that is me guessing.
Just did some tests at home, visiting speedtest.net and saturating wifi from different spots in the apartment.
At my desk, ideal signal strength: no screen glitches
One room away from the router, less signal: glitches slightly on downlink, glitches painfully on uplink
I also didn’t notice this kind of glitching until sysimagev3 (had another kind of glitch before but less often and not wifi correlated). I’ve also noticed more reliable signal strength using v3. Perhaps the regulatory zone is set differently?
So, I might do some experimentation with cables… In general, I’ve been making do but not happy with the antenna that came with the Reform. Will order that Laird part today and save playing with cables until installing it.
I wonder if there’s some easy/hacky way to shield the display cable.
The shielding solved the issue for me! I used some 10mm EMI shielding tape wrapped up to the hinge and connected it to the HDMI shield (which appears to be attached to “ground” but is oddly not low resistance to the USB port shield).
I’d like to attempt the same! I don’t have any EMI shielding tape… Anything you’d recommend I look for? Everything I quickly see is copper, but what you’ve used isn’t? And I guess it’s conductive so one has to be pretty careful about how it’s laid out? Any tips?
Correct, it is “braided” EMI tape. I purchased 10mm tape from here, but I can also send you some (inside US) if you like. DM me with an address.
You are correct that the tape is conductive. I was not concerned about the wrap on the cable touching anything, but the “tail” which grounds it to the HDMI shield could conceivably short something out. I routed one of the keyboard cables under it (pictured), to ensure it is held off the PCB.
I sloppily copied their technique and it seems to work! The screen glitching was painfully bad in some circumstances and easily reproduceable, and now it isn’t.
