POMdev wrote: > Thanks all for the very useful responses. > > I believe this "Radio Loses WiFi Connection" issue to be so serious as > to deserve major escalation. My 6 radios are very nearly unusable, most > failing after just a few minutes. Of course, not everyone is > experiencing this yet, but enough have so far and the numbers likely > will just keep growing. The entire ecosystem is threatened if the radios > fail in large number. > > I tried turning everything off except for one new simple wireless AP, > the main router, and the pc running LMS, including phones, cameras, > televisions, IOT devices, and Bluetooth devices. Still, the radios wifi > deteriorated and eventually lost connectivity. This suggests RFI of some > sort disrupting the system, or that 6 radios on one access point is too > many. However, these radios have been around for years, but this > disconnection issue started only earlier this year, and has gotten worse > during the summer. > > The neighbors are not that close, and their wifi signals are at least > 20-30 db lower. I haven't gotten out the (quite heavy) spectrum analyzer > yet to look for interference, but it looks like that might be necessary. > I shouldn't need an LNA to detect what is presumably a high level > signal. > > I tried setting the debugging flags, to increase the driver verbosity, > but these had no effect (because the ar6000.ko driver was build without > debug support?). However, a debug build may be needed. > > Despite time spent, I have not made much or any progress with this > issue, being impeded by lack of the AR6102 chip specs and the source > code (no access to this site's private repository.) The closest public > source so far is some AR6kSDK.build_sw.18 code and various versions of > AR6kSDK.3.1, some of which is close to but not the same as the code > referenced in the various patch files. The various open source > decompilers produce dubious looking gibberish. > > To reiterate, I see three issues: the device is too easily disrupted, > fails to report a connection loss, and the supplicant or other layer > does not otherwise notice a lack of connectivity and initiate corrective > action. An optimum response would be to address all three. > > I examined the serial console log file started after over a day of > disconnection. In this session, the radio is connected to a simple bgn > router/access point located very close to the 4 of the 6 radios used by > the 'important' (no, not me) users. The rssi of this access point at the > test radio is 25 db below another modern AP 10 feet away. The bgn AP is > on channel 1, which somehow greatly improved its chances of being > detected by the ar6000 scan, and presumably connectivity, although not > enough. Connected to the office AP, the SB radio is more stable. > However, its UE office mate, 30 feet away from the same AP, drops out > anyway after a few minutes, so rssi is not the only factor. > > The 'wpa_cli scan_results' frequently show only 1 to 3 access points, > that is, the three nearby access points are frequently not all shown > even though the signal strengths are all good to excellent. > Occasionally, a neighbor's AP shows up, even though its signal strength > is 20 db less than the lowest house AP. Also, the connected AP sometimes > doesn't show up in the scan, a clue. > > I then switched each radio to the strongest AP signal, on the three APs, > but they still kept dropping out after a few minutes, except the UE in > the basement, and the one with the serial cable connection, which last > much longer. > > Out of frustration, I concocted a script to ping the gateway, and reset > the wireless when connectivity is lost. I cannot yet predict from the > statistics (e.g., when the errors skyrocket and/or the data rates > decline) when connectivity will be lost. So it's the ping method, which > is reactive and result in an outage of several seconds. This should be > handled by the supplicant, or other service, but since that is not > working, this is a start. > > Some good news: "wpa_cli reassociate" restores connectivity in many, but > not all, cases. It works fairly quickly, when it works. > > And, my script has been running on my 6 radios, and they are still > connected, a big success!The AR6102 datasheet is available here.
https://datasheetspdf.com/mobile-datasheet/AR6102.html Sent from my Pixel 3a using Tapatalk ------------------------------------------------------------------------ slartibartfast's Profile: http://forums.slimdevices.com/member.php?userid=35609 View this thread: http://forums.slimdevices.com/showthread.php?t=111663 _______________________________________________ Radio mailing list [email protected] http://lists.slimdevices.com/mailman/listinfo/radio
