Router temperatures

[follower]

Hey guys, I hope everyone is okay during the challenging times.

I run electronics recycling business on the side and I repair electronics as a hobby. Component level repairs from PC motherboards to iPhones. I have my small lab with what's needed for my fun projects. A few resurrected RT-AC86U routers are sitting on my shelf. Those fail in big numbers. Asus is reusing parts from previous models and they don't fit properly. Some of you may have noticed a number of huge thermal transfer pads. The gaps between the ICs and the heatsink are as large as 3mm. It's built as cheap as possible with hope for the best. There are many thermal stress points on this PCB. It lights up like NYC night lights on the thermal camera image. Lead free solder doesn't like thermal stress points. VRMs fail as well as RF ICs. I had to replace few VRMs and reflow a RF IC to bring a few back to life. The repair is not worth it. Fix one thing and something else fails. The bad design always wins. Using copper shims and thermal paste drops the temperatures you report by 20C. A fan on top drops it further another 20C. My general recommendation is to avoid this model. If you already have it avoid on/off cycles to reduce the thermal stress. A fan is a workaround temporary solution. When the VRMs fail you get one LAN port light lit or some other LED half lit. When the RF IC fails you get Channel 0 in your UI. I don't know why Asus uses 19V PS and then drops the voltages down with multiple VRMs. It only creates more failure points and generates more heat. Other routers do fine with 6-12V PS. I can't find schematics online for Asus routers. Looking at the components used nothing needs 19V. Good luck to RT-AC86U owners.

A funny thing is some people still saying it's ok. They even say 100C-120C is ok. If we follow their logic we don't have to use heatsinks.

Release - RT-AC86U 11th January 2021 Firmware version 3.0.0.4.386.41634

Yes. But you will need a USB drive of some sort inserted to trigger the process. (There may be another way, too) Enable SSH access LAN only. In a terminal (Putty is great for this) log into your router You will use the VI editor to create a file and enter contents vi /jffs/post-mount (Enter)...

www.snbforums.com

Interestingly they are so quite these days.

 

[Amwjujo]

I have the same router for 2 years now. Running stand alone with Skynet, Unbound DNS and VPN. Never had issue with temps at all. It may as well be related to AiMesh.

I think mine has the same age. Someone mentioned earlier in the thread that it might not be a software problem but it might be a hardware one - depending on the batch lot or the place where the router was manufactured. It could be true as i see so many different opinions from people having the same router and then reporting different temperatures( low or high) running the router in different environments at different temperatures, but then I am wondering: How come the router use to run perfectly fine( i mean at an acceptable temp) with all the firmware versions before 386.1 and after this we stared to have this issue with it getting hot.

I am not running AiMesh on it - though I am planning on doing it in a week time - and I also use: Diversion, Skynet, YazFi, Cake, Unbound, and OpenVPN.
I keep my router on a shelf close to the ceilling next to a media server PC and at first I thought it might be to hot up there and this could be the cause. I removed the router from his place and have it placed in the room on a table and let it run for a couple of days to see what happens - the temperatures were still at 92C.
The only thing that worked and dropped the temperature a lot was the pair of fans that i bought, which made the temp to drop to 56C - 60C and before the router upgrade the temp was around 74C.

I will try to move the fans from router USB to a separate 5V power source as @Tech9 suggested to see if the noise from the fans will drop and I'll keep you updated.

 

[Matthew Patrick]

It's definitely a hardware thing. And it's also not because of old routers. As I said. Mine is a new late 2020 unit..

 

[gattaca]

I will try to move the fans from router USB to a separate 5V power source as @Tech9 suggested to see if the noise from the fans will drop and I'll keep you updated.

Yes, the noise will drop substantially by spinning the fan slower. Caution: Some 12V fans may not spin up at way lower voltages - all depends on how they are designed. Push / shove use a PSU with an adjustable voltage output and then dial down the output voltage or use something like the small LNA/ULNA inline resistors such as is supplied with Noctua FLX models to slow the fans speeds. These are really just short cables with in-line resistors of varying ohm values to lower the voltages which makes the fan spin slower. They work great - just keep them to 1 per fan per Noctua's instructions. I've used them for years to slow my Noctua fans to near silent. However, they will not work on every fan, nor every Nocta fan - again b/c of the fan's design and what's needed to start it spinning at lower voltages. For instance, I tried to use one on a 5V Noctua and it would not spin up. That fan wanted/needed 5V to start spinning. Always, YMMV. Stay safe, stay alive! Peace.

 

[gattaca]

I don't know why Asus uses 19V PS and then drops the voltages down with multiple VRMs. It only creates more failure points and generates more heat. Other routers do fine with 6-12V PS. I can't find schematics online for Asus routers. Looking at the components used nothing needs 19V. Good luck to RT-AC86U owners.

I think.. 19V is used b/c by supplying a higher voltage, designers can spec the PSU to produce less current / lower amps. There may be some regulation rules in play too, IDK about that. It's the old adage, "you transmit power with voltage" and is why our transmission lines run at a few 100K volts. Nothing in a PC runs ever at 19V but it is a very common PSU voltage for many laptops.

 

[Amwjujo]

It's definitely a hardware thing. And it's also not because of old routers. As I said. Mine is a new late 2020 unit..

It makes me wonder... if it is a pure hardware issue, how come the routers were running perfectly fine before 386.1 .....

 

[Wade Coxon]

It makes me wonder... if it is a pure hardware issue, how come the routers were running perfectly fine before 386.1 .....

Mine was running a bit hot, but was definitely 15 degrees or so hotter on 386.1.

In that version of firmware, the CPU wait state was disabled, which added a huge heat load. Merlin switched that off himself.

The EEE (Energy Efficient Ethernet) was also disabled, which creates heat for every LAN port in use.
I have turned that feature back on, and now only use the one LAN port on my router.

The suspicion is that these two settings were likely made to accommodate whatever nifty new things the AIMesh 2.0 code required.

So the hardware was already poorly assembled and hotter than it should have been, but the additional heat load pushed it over the edge.

 

[Matthew Patrick]

It makes me wonder... if it is a pure hardware issue, how come the routers were running perfectly fine before 386.1 .....

I think it's always been like that. But even tho the thermal management in the hw department isn't that good. It's good enough , that's why it's never been a problem. Until now when asus needs to disable some things that creates more heat. Then kaboom throttling everywhere...

 

[MvW]

You realise you're endorsing the poster insulting Merlin, right?

Sorry, I didn't mean to say anything wrong but from your reaction it seems I have. If I posted anything offending please bear in mind that English is not my native language and I would certainly never endorse anyone insulting Eric, I have way too much respect for him and insulting isn't my style anyway. I don't know how to correct this, sorry, my brain is letting my down and I'll be home schooling my kid today, all day. So Eric, if you read this, and are offended, it was never my intention. Sorry. Sometimes I feel like I'd better return to read-only mode like in the past few years... sigh.

 

[Jack Yaz]

Sorry, I didn't mean to say anything wrong but from your reaction it seems I have. If I posted anything offending please bear in mind that English is not my native language and I would certainly never endorse anyone insulting Eric, I have way to much respect for him and insulting my style anyway. I don't know how to correct this, sorry, my brain is letting my down and I'll be home schooling my kid today, all day. So Eric, if you read this, and are offended, it was never my attention. Sometimes I feel like I'd better return to read-only mode like in the past few years... sigh.

nothing against you, just a friendly reminder to read through what someone is referencing when they link to another post
please keep contributing, it's good to see new and returning members on these forums

 

[BreakingDad]

This seems to be working on mine, temps down to minus 60

 

[Centrifuge]

if it is a pure hardware issue, how come the routers were running perfectly fine before 386.1 .....

Yeah, that's kind of the elephant in the room, and CPU's don't have thermostats and just decide to raise the temperature, they produce more heat by processing more instructions. Nobody has explained what all the extra stuff being processed is pre and post an update to 386, all other things being equal. Is it just Aimesh or buggy firmware, but its all closed source, so you don't know.

 

[geralds34]

Mine was running a bit hot, but was definitely 15 degrees or so hotter on 386.1.

In that version of firmware, the CPU wait state was disabled, which added a huge heat load. Merlin switched that off himself.

The EEE (Energy Efficient Ethernet) was also disabled, which creates heat for every LAN port in use.
I have turned that feature back on, and now only use the one LAN port on my router.

The suspicion is that these two settings were likely made to accommodate whatever nifty new things the AIMesh 2.0 code required.

So the hardware was already poorly assembled and hotter than it should have been, but the additional heat load pushed it over the edge.

How do you enable EEE (Energy Efficient Ethernet), or check its status?

 

[gattaca]

My 35+ years of performance engineering says ASUS needed the pCPU cycles to accomodate the "AI Mesh" and/or other features being added as I think @Jack Yaz mentioned earlier. I'd buy that being a very valid reason. Think about it..

So ASUS turned OFF pCPUWAIT and maybe other "power saving" measures to get more performance from these routers to accommodate new features vs maybe having their support lines ringing about "poor performance" on AI Mesh or whatever is being added. Consider all the goodies ASUS has added vs 5 years ago. All these need pCPU cycles to work - well.

TL;DR details..
As I've stated in several other threads, I've been running my AC86 routers with the "green stuff" disabled for many years now - every chance I get. I've also disabled it all in my new AX86. That's also another reason I added the fans to CMA!

It's a not-well-known or understood point that today's pCPUs (Intel/AMD/ARMs), pNICS, ethernet ports, HDD, SSD, etc.. are literally many times "sleeping" more than they are working. All that sleeping, just like us, takes time to wake up. It's trading lower energy use for performance. This "waking up" adds 100s of uS for EACH state transition and huge latency to processing! The industry, in general, has glossed over this too. What features in BIOS do you think are the first things high-frequency-traders disable where uS and ms can mean millions of $$$?

Several years ago, I wrote a paper showing how we could easily gain 30%+ in performance/capacity by turning on "High Performance" settings in both the firmware/BIOS and OS on sever class machines. In most cases, the manufactuers default all these to "Balanced Power" modes which are middle-of-the-road and defaulted to ON to meet "energy certifications...."

Who cares if you might have to buy 30% MORE pHW to run the same workload? I bet your CFO does - right? If I put a switch on the motherboard and told the buyers that I'd intentionally thottled their PCs by 30%! How do you think that would play? Well, that's exactly what the manufactures have done, all-be-it at the request of the clouders and it's tucked away in firmware, BIOS and OS settings.

So I'll explain one step further. What do you think is the highest expense todays clouders incur running their huge datacenters? From their PoV, every watt counts b/c that's burning their money... and if customers do not know better or fully comprend the implications, they believe/trust clouders are delivering them best perfomance possible for their expensive instances - right? Quite, quite wrong! If you need the absolute top/best performance, well that instance costs more! Stay safe, stay alive! Peace.

 

[gattaca]

How do you enable EEE (Energy Efficient Ethernet), or check its status?

depends on the model.
SSH to the router then from the command line use --> /bin/pwr status to see the states.. You can also use that to change the setttings which are not preserved on boot... you have to setup script to flip'm for that.

The AX86 has more options than the AC86... IDK about others.

 

[dscline]

I will try to move the fans from router USB to a separate 5V power source as @Tech9 suggested to see if the noise from the fans will drop and I'll keep you updated.

If you're currently powering them via USB, you're already on 5V power. Assuming these are 12v fans, you may not be able to dial them down much more and have them still reliably start on power-up.

 

[goodwin_c]

If you're currently powering them via USB, you're already on 5V power. Assuming these are 12v fans, you may not be able to dial them down much more and have them still reliably start on power-up.

For some 12V fans, even 5V may be too low to start, especially if there is some air resistance (but enough to keep rotating) Personally, I'm trying to keep voltage around 5.5-6V to ensure that fan will always start.

 

[Amwjujo]

They are dual fans which I run them on the USB port, so you might be right ... already on 5V, but they are so noisy ... i even removed that metal net they came with but still noisy.
I don't know if I am allowed to post links here, but I can post a link to the product if anyone is interested in having a look...

 

[Deleted member 62525]

I think mine has the same age. Someone mentioned earlier in the thread that it might not be a software problem but it might be a hardware one - depending on the batch lot or the place where the router was manufactured. It could be true as i see so many different opinions from people having the same router and then reporting different temperatures( low or high) running the router in different environments at different temperatures, but then I am wondering: How come the router use to run perfectly fine( i mean at an acceptable temp) with all the firmware versions before 386.1 and after this we stared to have this issue with it getting hot.

I am not running AiMesh on it - though I am planning on doing it in a week time - and I also use: Diversion, Skynet, YazFi, Cake, Unbound, and OpenVPN.
I keep my router on a shelf close to the ceilling next to a media server PC and at first I thought it might be to hot up there and this could be the cause. I removed the router from his place and have it placed in the room on a table and let it run for a couple of days to see what happens - the temperatures were still at 92C.
The only thing that worked and dropped the temperature a lot was the pair of fans that i bought, which made the temp to drop to 56C - 60C and before the router upgrade the temp was around 74C.

I will try to move the fans from router USB to a separate 5V power source as @Tech9 suggested to see if the noise from the fans will drop and I'll keep you updated.

I dont think its firmware related at all. Otherwise everyone would have this issue. If you have upgraded to 386.1 and after performed proper hard reset it should be fine. As some people mentioned it could be hardware issue related to some of the units. In that case I would dissemble the unit and check if there is any heat sinks and they are correctly mounted/installed. Running temps 80-90 should be fine and I would not be concerned.

 

[Wade Coxon]

I dont think its firmware related at all. Otherwise everyone would have this issue. If you have upgraded to 386.1 and after performed proper hard reset it should be fine. As some people mentioned it could be hardware issue related to some of the units. In that case I would dissemble the unit and check if there is any heat sinks and they are correctly mounted/installed. Running temps 80-90 should be fine and I would not be concerned.

I guarantee you, it is related to the firmware.
We know of a very specific set of changes that were made that contribute to the worst of the temperature increases, namely the CPU Wait and EEE settings.
The remainder is something deeper, and probably AIMesh related.

You are right though that it is also absolutely hardware related. Some units have poorly fitted heatsinks, and there doesn't seem to be any pattern in manufacturing dates that anyone can discern yet.

Having disassembled the router, users are seeing those extra thick thermal pads. These are likely the cause, as thermal transfer through such a thick pad is going to be sketchy and unpredictable.

In short, the firmware change, has made an underlying hardware issue more prominent for some ac86u owners. Everyone will see a small temperature increase, but those with dodgy heatsinks will see a larger increase.

In my own case, the temps were pushed beyond 100 degrees, and the CPU started shutting down cores.