And some infos / notes ..
Please use this firmware only for this type .. eg. the same router.
there are too much "chinese" versions of it.
It must have same SOC MTK7628, 64Mb RAM, 2xLAN ports and 8Mb spi Flash.
My server IP was changed to 89.173.96.22, so all links above are broken.
I can't edit all previous posts. (for such a long time)
rolly
108
This is a similar issue and device to Generic China "on-wall" router YTXC-958K 7628N with @JRD_McLAREN @MOV et al.
[also artas26 gustawa21 ... can't @ them until I've been here longer]
Here I am interested in the unrequested resets. It's newly installed and a promising recipe for great annoyance to the user on a business Zoom call... 
There are two here but I'm only visiting. It's not mine and I haven't taken it apart. Mostly I'm trying to fend off a more severe case of buyers' regret!
It reports itself as a YTXC_xxxx with xxxx being the lastfour of one of its MAC addresses. The status page shows "Firmware version 1.2", "Software version ROUTER". There is very little for a geek to look at and figure out what it is, and the post above is the most useful info I found so far 
.
This one looks most similar to the picture at https://www.ebay.co.uk/itm/355879526999 = https://archive.ph/89Zuq , found with search "300M Wall AP", but terms like "in-wall router", "wall-mounted AP", "China OpenWRT Router", "Single-Gang WiFi AP" and YTXC are also relevant. On the front: five blue LEDs, three buttons, one mains power switch, RJ45, USB.
My observations,
- Device is serving wifi, then... the wifi signal disappears, the lights on the front have mostly gone off, and it restarts after about a minute. Repeat at unpredictable intervals.
- It appears not to be a thermal problem in this case. After a protracted exercising with
iperf3 , warmth can be felt on the front panel and it will restart a couple of times in a two hour pounding. However the restarts are not obviously clustered - unless I've been insufficiently thorough, which is possible. If it was thermal then I think after the first crash there would be more, sooner.
- It appears to not be directly throughput-induced. When the laptop is close enough I can get a 300Mbit/sec link (as reported by
iwconfig) and convincing throughput to confirm that. However "close enough" is only a few feet away and the signal drops off rapidly.
- It does this in router or AP mode alike, so I'm not distinguishing between those here.
- Link speed seems to be over-optimistically chosen... is that by the AP? I say this because feeding it heavy traffic will cause the link speed to drop even as low as 6Mb/s at what I considered a reasonable distance. Yes there are stone walls and aluminium foil in the floors, but another 4-ducky TP-Link is less troubled by range problems.
- Assuming it was voltage brown-out, I stuck a USB lamp (3x white LED) in the front to load the power supply harder, and gave it some more pounding. I saw no flicker in the light, nor obvious increase in crashing.
- It's new enough that I wouldn't think capacitor aging is relevant unless the PCB happened to be assembled with really old stock..?
- One was tested new out of the box, but is now part installed and inaccessible. I found nothing obviously different that way.
- @JRD_McLAREN 's meter pic (of a minimum voltage) strongly suggests brown-out but I don't know where that was measured, and I haven't seen the circuit / traces for the PSUs of CPU vs USB socket.
- I haven't tried disabling wifi and/or pounding it via the wired front panel - might not have time this trip.
- I believe that a crash, even if wifi-induced, would cause front panel connectivity to drop also. This is because it can NAT the packets from front to back, so I assume they're hitting CPU.
- There are reports of wifi driver bugs affecting similar devices, and I haven't looked into that.
- There are suggestions (here somewhere?) of RFI (radio frequency interference) being the cause but I don't have links. It does make sense that occasional reliability issues might not be picked up if the design is mostly "dump this reference design on a PCB and shuffle it to fit a single-gang box". It also seems to be supported by the unpredictable nature of the resets.
- I see it has a LuCI interface on port 80. I don't know much about LuCI and other ports are closed - I have no logs.
- With a
watch -n0.2 iwconfig or equivalent, one can see the AP throttle linkspeed or disappear. This makes it easier to script the monitoring of the problem. I've done that but run out of time to leave it just sitting waiting; my suspicion is there are no crashes if the device is idle. It can help to tell the laptop to not auto-connect other nearby APs.
- I haven't tried back-powering it at 5V into the front USB port (which is not exactly the AI's suggestion). This might be an interesting experiment and only needs a power bank + USB-A widowmaker(!) cable.
- I haven't tried pushing through traffic with specific repeating data, in hopes of hitting some magic "crash data pattern". It doesn't seem like a promising approach.
- Other details... it doesn't take PoE (tested with Netgear GS308P). It wants mains at 90~265V. Photos in post above suggest some care was taken to isolate the low/high voltage sides, but I'm only going on the PCB cut slot. If you want scary stories about too-cheaply-made USB power devices, go watch some BigCliveDotcom . I believe this one is mounted in a plastic wallbox, but haven't seen it. Mounting in a metal box seems likely to be unhelpful for signal propagation.
Possible hardware mods,
- aluminium foil shielding in strategic places, grounded if possible... to a 0v pin? There is no mains Earth connection. Being careful not to introduce mains voltage hazards!
- adding smaller decoupling capacitors in strategic places. The SMD ceramic ones can often be piggy-backed. Be aware of the benefit of multiple caps of different values = https://electronics.stackexchange.com/questions/541956/0-1-uf-in-parallel-with-larger-1-uf-to-reduce-noise-from-ic , https://forum.arduino.cc/t/multiple-decoupling-capacitor-in-parallel-solved/523181
Possibly relevant?
My requests,
- Software solutions would be great
- Hardware solutions might be great for some people, but I suspect these will get washed off with the money hose
- Recommendations for a better in-wall AP would be welcome, especially if form-factor-compatible with the single-gang socket box
I had some help from Grok3 in putting this together. It's a great search engine when you don't know the keywords.
rolly
110
With the factory firmware I got, holding reset for 10 sec while powering on (front mains power button) does the trick. YTXC in-wall router, crashing / shutdown / restarting
It sounds like with custom firmware there can be additional issues...
The main problem with these devices is poor PSU ...
My one restarts 3-4x per day ...
Try to power it via USB power bank, as you say ...
and you will see....
rolly
112
OK thanks, that'll be the next thing to try when I'm there. It may be a while.
(FTR my posts were originally in a new thread "YTXC in-wall router, crashing / shutdown / restarting" which is now merged)
