RT-AC87U died unexpectedly, getting hot. Is another failure imminent?

[Speedy-VI]

My RT-AC87U died unexpectedly today. Lost internet, checked the router and all LEDs were off, like it was powered off. The top of the unit was quite warm, almost hot, to the touch. I cycled the power switch several times and checked the power adapter output with a DC volt meter (19VDC). I let it cool down and tried again - nothing. I tried putting it into recovery mode by holding the Reset button in then plugging in power - nothing. An Asus router should not just spontaneously die right (or maybe they do)? So I removed 4 screws and pulled the top off hoping to find a blown internal fuse or some sign of heat damage or a burned component. Everything look perfect and if there is a fuse, its surface mount. So despite its internals looking brand new, it was still totally dead. Then I read a post on this forum that said try removing power and slowly cycling the power switch on and off about 10 times. I did that, plugged it back in, turned it on and it booted up! So maybe dust buildup in the power switch?

I have not yet connected to it to see if its functioning, but I can see both the 2..4GHz and 5GHz SSIDs, so its broadcasting. I just let it boot up and sit there with the top removed. There are 3 big heat sinks fastened to the main circuit board. If looking from the front, from left to right they are labelled SH3, SH2, and SH1. SH1 and SH3 are the same size, SH2 is larger. After powering it up and letting it sit for awhile, heat sink SH1 was getting too hot to comfortably touch. SH2 was warm but not alarming, and SH3 was a little warm. I also thought I caught a whiff of burnt electronics smell but it was very mild and I could not localize it. I can probably connect to it but I am concerned about why it suddenly shut down. Now I know why the top was so warm - there are 3 big heatsinks right below it. With the top off, heat sink SH1 is getting almost too hot to touch. Is this normal? I am worried maybe a cap or a voltage regulator is failing, and a circuit is getting cooked, which caused it to shutdown today due to overheating. But if that were true, it should have powered back up after I let it cool down to ambient. I could not get it to do anything until I removed power and slowly cycled the power button ~10 times. So maybe a thermal shutdown caused by a failing component, and a really dirty power switch?

Of note is that I updated the router firmware to v3.0.0.4.382.51939 about a week ago. The update went fine and I don't know if it's related in any way. Also, in my digging today I found some posts talking about how hot some Asus routers get, and how its best to orient them vertically. Having looked inside and seen where the heatsinks are, I think orienting it horizontally (i.e. sitting on a table) would be best, which is the way it has been sitting for years. I spent half the day dealing with swapping in another (Netgear R7800) that I was using as a access point, and dealing with a less than cooperative cable modem. I don't want to swap the RT-AC87U back in and reset the Netgear as an AP again if the Asus router is going to fail. While typing this I just put my hand on heat sink SH1 and its really hot. SH2 and SH3 jest get warm. Any ideas or suggestions?

 

[Kendo]

You could try using at least a silent cooler like I did if that's an option for you, even without mounting it directly to the router and using it in silent mode, it still shaved off around 14 degrees from the cpu and that's no joke.
(this is the AC86U below).
Others took the cooling even further with bigger temperature drops.
Just a suggestion.

p.s. using the same method gained 10-11 degrees on the AC68U (cpu).

 

[Speedy-VI]

Thank you for the suggestion, but I am not sure what you mean by a "silent cooler" and running it in "silent mode". Can you elaborate on this? Do you mean a bigger passive heatsink or something with a fan?

Generally, there are several things I could do to lower the temps and get more waste heat out of the system. I could add a bigger heat sink and/or mount a small fan in there and/or leave it running with the plastic top off. My concerns remain about why it shut down out of the blue yesterday, and why it would not power back up until I removed power and cycled the power switch 10 times. I have had this router since 2016. It has locked up a few times but it has never totally shut down like this before. If it is heat related, it must be because it got hotter yesterday than it has in the past. Something changed, and I am concerned that failure may be imminent. I don't want to go through the hassle of putting it back in the network if its going to die soon. I think I will just let it run all day today with the top off and see if it shuts down again.

Here is a picture of the RT-AC87U with the top removed. The heatsink with the red box around it (SH1) is the one that is getting almost too hot to touch.

 

[Kendo]

Silent cooler with lower voltage, less rpms that would be enough to help with the high temps.
You might wanna check what's under that heatsink if it's too hot or change it.
All I can think is that the temps were too high, maybe some other members had similar issues with the AC87U?

 

[Speedy-VI]

I have it sitting here next to me running with the top cover off. Just updated firmware to v3.0.0.4.382.52545. Heat sink SH1 is too hot for me to touch for more than ~1 second, which according to a quick Google search means its at least 50°C (122°F). I have some thermal sensors laying around from custom computer builds. I could attach one to the heat sink and connect it to one of my computer fan controllers and get a real temperature reading, but I don't know what "normal" is. I already know its getting too hot to touch and Google says for most people that is around 50°C. I am just going to let it sit here and see if it fails again.

I don't want to replace the router, but I also don't want to switch it back into my network if its going to fail soon. Adding more heatsink mass and/or pointing a fan at the heatsink means it will have to run with the top cover off, making it a bit of a jerry-rig. I am hoping someone here that has experience with the RT-AC87U or similar Asus routers, specifically what the normal operating temperature range of this heat sink is, will see this thread and chime in. I have not given up hope yet!

 

[bsdsource]

The SH1 heatsink is for the CPU. The RT-AC87U case it not ideal for heat dissipation since it doesn't have a vent on top. This particular router CPU at idle runs higher temperatures than most most ASUS routers. You might want to consider activate cooling your router. If you do a forum search or google search for the RT-AC87U you will find a lot of information about this router reference high temperatures as well as cooling solutions.

You should really check your router temperatures so you know what they are. If you can't see them in the webui you can access the console and run commands to see your router temperatures. You will also find the commands doing a forum search.

RT-AC87U cooling solutions

 

[Speedy-VI]

Thank you for your response and clarifying what heat sink SH1 covers (the CPU). I did some more digging which also confirms this. I did search this forum and the internet in general for info about the RT-AC87U, and found many reports of it running hot. I quite enjoyed many of the cooling solutions in the thread you provided. I had the router sitting in an upstairs bedroom where it gets quite warm, with a cable model sitting on top of it. I now realize that this setup was just asking for a thermal problem. The router has now been running for 24 hours (with the top cover off) and so far its fine. Heat sink SH1 is still really hot but I am now confident that this is normal for this model.

Regarding monitoring router temps, I don't see any reported in the Asus GUI (now running firmware v3.0.0.4.382.541634). I would try putting AsusWRT on it but the RT-AC87U is listed as no longer supported. I did find some info about a command line interface but I am not sure how to access it. I will search the forum for specifics on how to do this. I do see header J3 with pins for Tx, Rx, Vcc, and ground. I think this is a console serial port. Hopefully some weekend Google searches will steer me in the right direction. I still have no idea why removing power then cycling the power switch 10 times made it work. I suspect this is just evidence of a dirty power switch.

Thanks again.

 

[bsdsource]

Speedy-VI I've provided instructions below on how to telnet into your router so you can run Linux commands from the console. A serial connection isn't necessary is much more difficult that just using telnet. First enable telnet from the router webui.

Administration --> System --> Enable Telnet (then save settings)

Download the Windows putty telnet client from here: putty

1. start putty
2. enter your router IP address into the IP address box in putty
3. select the open button at the bottom of putty (you should now be connected to your router)
4. at router command prompt enter user name (same name you use to login to your router webui)
5. at router command prompt enter password (same password you use to login to your router webui)
6. you should now be logged into your router and can use Linux commands

Try the following Linux commands:

For the Quantenna SoC temperature:

qcsapi_sockrpc get_temperature

For the Broadcom CPU temperature:

cat /proc/dmu/temperature

These commands may no longer work. Do a forum or google search for temperature commands that may work. When you are done, disable telnet from the webui. You don't want this running for an extended period of time since it's not a secure protocol.

 

[Speedy-VI]

Thank you. I am no expert with this but I have made good progress. Here is what I did (so far), which has worked out pretty well.

1. Enabled Telnet and also SSH on the router.
2. Instead of Putty, I used an SSH client and terminal emulator called ZOC8 which is actually pretty nice. I was able to connect via SSH with ZOC8.
3. I tried the commands you provided and got back the following:
   • temperature_rfic_external = 0.0
   • temperature_rfic_internal = 47.0
   • temperature_bbic_internal = 65.0
4. I assume these are Celsius. I don't know why temperature_rfic_external = 0.0.
5. I also found an obscure post in an Australian Whirlpool Forum called Stupid Asus 68U Tricks which provided a Shell Script written for an Asus 68U that makes a router log entry with the temps for CPU and both radios.
6. I uploaded the script to the /jffs partition on the router and ran it. I am not a Linux person so I can't comment on the script but it didn't look like it did anything nefarious, and it mostly worked.
7. When I run the script, I get syntax errors but it does cause an entry in the router log that shows temps for the CPU and one of the radios (called eth1). The other radio temp called eth2 does not seem to report.
8. To run the script every 30-min, he uses the command "cru a TEMPS "*/30 * * * * /jffs/log_temps.sh".

The author of the script goes on to explain that Asus clears the chron table on every boot. Older Asus firmware allowed a special script to be run on boot but they disabled this. He has a workaround involving plugging in a USB flash drive and using NVRAM variable script_usbmount to specify a command line to be run whenever a USB device is mounted. Since USB drives are mounted during boot, he uses this to get the scripts to run even after a reboot. This is all a bit over my head but an interesting read. That script looks like this if you are interested.

#!/bin/sh
#the following line ensures the script only runs once per reboot cycle
mkdir /home/root/.asusrouterlock || exit
chmod 700 /jffs/*sh
cru a TEMPS "*/30 * * * * /jffs/log_temps.sh"
/jffs/log_temps.sh # optional – logs temps immediately after boot
nvram set script_usbmount="sh /jffs/myasusrouter.sh"
nvram commit

Below is a screen shot of the ZOC8 terminal window showing the login, running the commands you provided and the result, then running the script. The second image is a screen snip of the log file in the Asus GUI. I drew a red box around the log entry that is triggered by the script. Ignore the wrong date. For now, I got what I was after which is the CPU temp. The command you provided for the Quantenna SoC temperature provides 3 values, one of which is 0.0. The script reports temps for CPU and eth1 and eth2, but with no value for eth2. I don't know if the 2 SOC temps are the same as the 2 radio temps that the script gets.

I kind of went down the rabbit hole a bit with this script that seems to mostly work. I may go ahead and use his log temps script but change it to log temps every 1 minute. That way if the router fails again I can look at the log and see what the temps were before the failure. His work around to get the script to automatically run after a reboot is interesting but I don't think I will bother to try to implement it. Anyway, THANK YOU again for providing these commands. I think the router is OK and I just need to make sure it is well ventilated. I will put it back in service and keep an eye on those temps.

 

[vaboro]

heat sink SH1 was getting too hot to comfortably touch

SH1 is the CPU. It should be hotter then the other two.
After reading your initial post only I could infer that your CPU might have gotten extensive load after firmware update.

 

[vaboro]

I have had this router since 2016. It has locked up a few times but it has never totally shut down like this before.

Have you switched your router on and off pretty often or it stayed mostly on all the time? 6 years imho is quite a long run for the router. A combination of high CPU temperatures and switching on/off often may cause solder joints to deteriorate. You may google for the topic "GPU reballing" to learn more about solder joints deterioration due to overheating. This might be the reason of your router's powering off unexpectedly.
You may also find reading these posts interesting:

RT-AC87U/R: Cooling solutions

Anyone tried water cooling it yet? Sent from my iPhone using Tapatalk

www.snbforums.com

RT-AC87U/R: Cooling solutions

Anyone tried water cooling it yet? Sent from my iPhone using Tapatalk

www.snbforums.com

They basically say that high operating temperature might be the reason of your router's unstable behaviour.

 

[Speedy-VI]

Yes it has been clarified that SH1 is for the CPU. Thanks to bsdsource, and a post I found on forums.whirlpool.net.au, I have been able to query the CPU and SOC temps. CPU is running around 66°C (with the top cover off) which I think should be fine. I downloaded an SSH script from the whirlpool forum and after some tweaking, have it reporting temps to the router log once per minute. I don't know if the failure was related to the firmware update - I did it about a week before the failure. I have since updated it to v3.0.0.4.382.52545. Its been running for about 30 hours and so far its fine.

Regarding your 2nd post, the router has been pretty much online continuously, and it does sit in a room that gets hot in the summertime, and I had a cable modem sitting on top of it. I now know that was pretty dumb. I had to replace it with a Netgear R7800 that I was using as an access point for the downstairs. I think I will leave the Netgear upstairs, configured as the main router, and use the RT-AC87U as the downstairs AP. Its cooler downstairs and if it fails I will not lose the path to the internet, so I won't have everyone in my family yelling at me that their internet is down. Your point it well taken. Maybe 6 years is all I should expect out of a consumer router. The insides look fine though I did not inspect the underside of the main board. Maybe I should.

I have been browsing new models and am debating whether to try a Ubiquiti product this time around. I read some reviews of various Asus WIFI 6 routers and thermal problems seems to be common. I don't think we have any WIFI 6 capable devices but might as well plan for it. I am looking at the Asus RT-AX88U and the ZenWIFI AX. Reviews of both are mixed. Do you have a recommendation?

Thanks!

 

[vaboro]

Do you have a recommendation?

I don't have any AX devices yet either and run two RT-AC86U's now, though I changed thermal pads on both, attached fans for additional colling, increased RAM to 1024MB and tweaked them both to GT-AC2900 allowing to use 160MHz channels by AX devices if any.
If I were to upgrade my routers I would probably consider AX86U but it's just a matter of personal taste I would say.

 

[vaboro]

CPU is running around 66°C

By the way I think this CPU temp is normal.

Regarding your 2nd post, the router has been pretty much online continuously

If it's always online the probability of solder joints' deterioration is much lower. The temperature's going from low to high very often make solder joints micro crack and lose contant with time.

 

[Speedy-VI]

Thanks for the advice. I will read up on the AX86U. Reviews of the RT-AX89X are pretty bad - constant reboots, and most are blaming high temps.

Yes I think the CPU temp is OK, assuming the same metrics that apply to computer CPU's apply to router CPU's. Generally if you are under 90°C you are OK, though under 70°C is preferred.

Your point about solder joint degradation makes sense but I think this old RT-AC87U still has some life in it. This whole failure episode has made me aware of how hot these things run, and not to do stupid things like put it in the hottest room in the house and stack a cable modem on top of it. By moving it downstairs and ensuring it has adequate ventilation I think it will be fine. And the script is logging temps into the router log so I can easily keep an eye on them. So I went from maybe needing to buy a new router to just wanting to buy a new router.

 

[bsdsource]

Cool log script. Glad you used SSH instead of telnet. These commands should also get you the desired temperature results:

2.4 GHz Temp

printf "2.4 GHz: %s°C\\n" "$(wl -i "$(nvram get wl0_ifname)" phy_tempsense | awk '{ print $1/2+20 }')"

5 GHz Temp

printf "5 GHz: %s°C\\n" "$(qcsapi_sockrpc get_temperature | awk 'FNR == 2 {print $3}')"

CPU Temp

printf "CPU: %s\\n" "$(cut -f2 -d':' /proc/dmu/temperature | awk '{$1=$1;print}' | sed 's/..$/°C/')"