[ho1Aetoo]

[kernel-panic69]

And the Fritzbox and Ubiquiti are not the R9000. It is an overall design issue. I can send you mine and you can take it apart and see for yourself. It may not necessarily be related to the wifi chips themselves, rather the flash chip. I am fairly certain that what caused mine to tank was lack of proper heat dissipation. I probably could make toast on the heatsink under load. You are comparing different devices and circumstances while touting that the hardware can handle the temperatures.

Furthermore, I will tell you this: The logic circuitry and electronic hardware for certain equipment I used to work was rated above ambient temperatures we actually operated under and guess what? It would glitch and fail at random inside the temperature tolerances all the time. So take that theory and toss it out the window, because lowering ambient temperatures on the electronics is what fixed the issue. IDGAF what the design fault tolerances are on electronics. I could write a short novel on what I have seen fail due to ambient temperatures in an industrial environment where ambient temperatures were within the design tolerance parameters.

[ho1Aetoo]

Your R9000 has never worked, you bought it defective...
Anyone can read the temperatures of the radio chips in the WebIF.
I'm just saying that the temperatures for the chips are no problem.

I have run a 24h stress test on my R7800 in midsummer!, it has sent on the 5ghz radio 24h with ~900Mbit while the temperatures of radio wlan0 went up to 67°C....
These are basically ridiculous values.
My active air cooled QCA9984 card reached 75°C in the same period.
Every not actively cooled MiniPCI express card becomes easily 25°C warmer.
My wle1216v5-20 reaches ~100°C in a very well ventilated case

No need to panic just because the radios in a router reach 50 or 60°C
Massive radio dying in the R9000 rather points to other problems ...

You could try to re-solder the QCA chips...
I would laugh if that doesn't work and something else is defective like the CPU.

[kernel-panic69]

[msoengineer]

Ok, I was wrong and BS is right, 5ghz died and not 2.4....


The puzzle deepens and starts making more sense...I think I know where this is all going to shit...

The R7800 has passive antenna's which means the PCB has dedicated hardware&amps directly on the main mobo for BOTH radios...

The R9000, meanwhile, has active antenna's and no amps on the main mobo. There's no physical space left on this mobo to put the array of filters & amps on the main mobo, as they did on the 2.4ghz radio on the bottom section. So, Netgear off-loaded it to the antenna's in order to save space. That means that two connectors per each antenna must have DC power going on it in addition to the actual 2.4 & 5ghz signal.

All 4 antennas appear to get all their power through the tiny u.fl connector. there is no other wiring, otherwise. I am wondering if this is the shit design point of the whole thing?!!!!

Somehow it must be what causes the 5ghz to fail...I had the LED's all off from day one on the router- so, the extra power the antenna leds would pull, should not be the issue...there must be something else that's going on with the amps "pulling" too much power, but I am now much more convinced it's the active antenna design that is contributing to the failure.


CONDUCTORS tend to INCREASE their resistance with an increase in temperature. We all know this router runs warm and GIANT heatsinks were put in place...

So, the heat inside raises the wire temperature which in turn will increase resistance in the wires which will cause the amps to go low- if the voltage stays constant. Low power conditions on chips is what kills chips (I learned that from MSOE & ROCKWELL/ABB).

The real kicker is that two sets of the 2.4ghz wires (4 wires) run, squeezed, between the heatsinks. those wires must be heating the surrounding 5ghz wires that go to the two left antenna's (when looking from the front side of the assembled router), AND those are the longest run wires for 5ghz to boot which further contributes to IR losses....

I guess the real learning lesson here is to avoid ANY router that uses active antenna's that rely on the tiny u.fl connector to get their power for the active components in the amps.

If anyone else has other insights, let me know. I am open.

That said, I really don't think there is anything that can be done to repair the board in a cost effective way. Maybe a new QCA9984 chip would fix it...I have no idea. But, I've watched enough Louis Rossmann repair videos to know that there could be other cheap components responsible for supplying power to the chipset which could be a contributing factor. Without any schematics and more specialized equipment, I can't ever be sure where to start.

But, for now, I will blame the active antenna design as a contributing factor and poor cable management. All the wires should have run under the board on the much cooler side of the router.

This link shows off the 2.4ghz area
https://i2.kknews.cc/SIG=16o53hg/141r0001roq5s5s63rqp.jpg


Here is a full disassembly done in china of the R9000 that gets into some serious detail:
https://kknews.cc/zh-tw/digital/zm96b9g.html

and this is a translation into english but google won't load the photos for some reason...
https://kknews-cc.translate.goog/zh-tw/digital/zm96b9g.html?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=en-US

[kernel-panic69]

This does explain a lot. I'm still curious as to what is between the chips and the RF shields, though. That link I sent you likely denied you (I just checked my bookmark of it), but it originated here: https://fcc.io/PY3/16200339

I am curious what is on the underside of the RF shields. I would hope it is a conductive material that transfers heat. I am game for any way to remove the antennas and bits and pieces for the wireless and have a strictly ethernet router firmware image.

FFR: If I buy another R9000 or an XR700, it will be gone through and have the wiring routed better and the gratuitous fingers crossed, among other things

[msoengineer]

They use gray thermal pads to transfer the heat to the heat shields which then transfer the heat to the heat sinks.

There are also gray thermal pads underneath that attach to a giant, nearly whole mobo size, black plate. There were 4 pads on my plate, vs the three pads in the FCC photos...
Two for the QCA radios and one for the main cpu. The 4th one on mine seems to be under the 60ghz radio chip.

The wires are too short to be routed any other way... you would need to order separate wires and run things yourself.

[kernel-panic69]

Which I think the antenna ends on these are soldered on, fun. What a pain in the ass. I may get industrious and tear into mine. It would be awesome if it were just a matter of replacing the antenna wires.

[msoengineer]

For sure it's not as easy as replacing the wires...

my hunch is that either the 9984 5g chip is fried or the power rail and supporting power rail components have failed due to be driven too hard.

That's just a hunch, though....

We'll see where this leads. I am still trying some things behind the scenes with folks in the industry...

I've already wasted more time on this than the stupid thing is worth when it's working...

I may try to load the kong build that allegedly works with bunked radios and see if I can turn off 5ghz to get the router running again, at least on 2.4ghz...

[foz111]

i can confirm kong's last build will flash on mine as well as stock fw but not found a bs build that flashes, radios down on mine also. not sure if this will help from some time ago when i was attempting to get bs build on to it.

i have uploaded the text file/putty serial output from an attempt to tftp flash r45677 factory-to-ddwrt.img to my dropbox: https://www.dropbox.com/s/d4b1el07t0il2nu/not_accepting-f-2-ddwrt-r9000.txt?dl=0

[Alozaros]

[foz111]

This was the thread you helped me with Alozaros along with kong factory-to-ddwrt.img fw you linked in post #2 [url] https://forum.dd-wrt.com/phpBB2/viewtopic.php?t=328048 [/url] sorry no webupgrade.bin file stored on pc, from memory i couldn't upgrade to bs build from kong, flash completes fine from memory but fails to boot up, i cant even ping it but then flash back to stock via TFTP and back working again minus radios obviously.

It would be great if BS could provide an up to date build that would boot with broken radios, be very handy to run a vpn on especially if the cpu could be clocked @ 2.0Ghz.