Posted: Tue May 16, 2017 9:47 Post subject: NETGEAR R6250 - Very high CPU Temp (~80°C)
My Netgear R6250 with Kong dd-wrt shows a very high CPU Temperature. About 80°C ... that is a little bit high, or not?
The R6250's default firmware limited it to use only 1 CPU, and now with dd-wrt the second CPU is unlocked. At the Status-Page there is only one CPU-Temp instead of displaying two CPU-Temps.
Is it possible that dd-wrt combined both CPU-Temps?
So CPU_1 = 40°C and CPU_2 = 40°C -> 80°C
Where did you get those temps from? Around 75 is still ok. Ddwrt always showed one metric.
Edit: now I think I know what you meant. No, it's not like that, it's like on x86 CPUs where you have overall and each core temp. _________________ R6400v2 (boardID:30) - Kong 36480 running since 03/09/18 - (AP - DNSMasq - AdBlocking - QoS) R7800 - BS 31924 running since 05/26/17 - (AP - OpenVPN Client - DNSMasq - AdBlocking - QoS) R7000 - BS 30771 running since 12/16/16 - (AP - NAS - FTP - SMB - OpenVPN Server - Transmission - DDNS - DNSMasq - AdBlocking - QoS) R6250 - BS 29193 running since 03/20/16 - (AP - NAS - FTP - SMB - DNSMasq - AdBlocking)
I'm not sure if this is the normal temp for this router or not, but it's not right.
I just bought one on ebay, and found this thread searching to see if anyone else was seeing CPU temps 90 degC and higher... Looks like a lot of people are, which is probably why you can get them on ebay for a little more than $20...
So I just couldn't live with 90 degC CPU temps.. Just because I'm OCD or something (the router seemed to be working fine..) So tonight I took it apart. I had to de-solder the CPU heatsink, but under that was a dried out one of those blue foam heat transfer pads.. I ended up having to fold 2 pieces of copper foil I have around to make a shim, since the heatsink didn't actually touch the CPU without it, and added some normal grey thermal paste, then soldered it back together. CPU temp after that was right around 56 deg C, but while I was in there I decided to add a fan, which is keeping the CPU temp at 44 deg C stable..
I know it's been stated above that TJmax for the broadcom cpu is 125 deg, but I'm pretty sure that a normal intel CPU has a similar TJmax, but you don't want to run anywhere near that ever... most say stay under 65 or you shorten your cpu life...
I'm not sure if this is the normal temp for this router or not, but it's not right.
I just bought one on ebay, and found this thread searching to see if anyone else was seeing CPU temps 90 degC and higher... Looks like a lot of people are, which is probably why you can get them on ebay for a little more than $20...
So I just couldn't live with 90 degC CPU temps.. Just because I'm OCD or something (the router seemed to be working fine..) So tonight I took it apart. I had to de-solder the CPU heatsink, but under that was a dried out one of those blue foam heat transfer pads.. I ended up having to fold 2 pieces of copper foil I have around to make a shim, since the heatsink didn't actually touch the CPU without it, and added some normal grey thermal paste, then soldered it back together. CPU temp after that was right around 56 deg C, but while I was in there I decided to add a fan, which is keeping the CPU temp at 44 deg C stable..
I know it's been stated above that TJmax for the broadcom cpu is 125 deg, but I'm pretty sure that a normal intel CPU has a similar TJmax, but you don't want to run anywhere near that ever... most say stay under 65 or you shorten your cpu life...
Yeah "21st century OCD"... those temps are normal as said before. x86 Tjmax is 105 in most cases, so these SoC are designed to work fine at around 100. Fan was extremely overkill of course, and this "stay under 65 or you shorten your cpu life" is completely stupid of course too. _________________ R6400v2 (boardID:30) - Kong 36480 running since 03/09/18 - (AP - DNSMasq - AdBlocking - QoS) R7800 - BS 31924 running since 05/26/17 - (AP - OpenVPN Client - DNSMasq - AdBlocking - QoS) R7000 - BS 30771 running since 12/16/16 - (AP - NAS - FTP - SMB - OpenVPN Server - Transmission - DDNS - DNSMasq - AdBlocking - QoS) R6250 - BS 29193 running since 03/20/16 - (AP - NAS - FTP - SMB - DNSMasq - AdBlocking)
Yeah "21st century OCD"... those temps are normal as said before. x86 Tjmax is 105 in most cases, so these SoC are designed to work fine at around 100. Fan was extremely overkill of course, and this "stay under 65 or you shorten your cpu life" is completely stupid of course too.
Yeah, no idea where that supposed "rule" came from. 65 is a very low temp for virtually anything out there. Heck, that's basically an idle temp for most hardware. You'd pretty much not even be able to use anything if you had to "stay under 65". For instance, my Laptop runs at 48-53 on idle, then if i play an HD movie or some demanding game, it hits about 65-70. Doh, guess I need to shutdown, since i am nearing thermal core meltdown
However, i can somewhat sympathize with what he said about the heatsinks and overall cooling. It wouldn't hurt some manufacturers to add better heatsinks/pads/paste to at least make an effort to keep things as cool as they possibly can out of the box. A good solid quality heatsink and quality paste is more than adequate to keep things at a sufficient temp (especially for the OCD types )
I remember my last PC, absolute nightmare it was with the stock cooling. Right out of the gate the temps were at 95 due to this abhorrent and weak excuse for a heatsink they used on the Northbridge/Southbridge (which was one of them stupid combo sinks and all one piece). And worse, when I changed it out and put some quality paste in, the stock paste was this pink crusty ass bubble gum nonsense that took forever to scrub off and clean thoroughly, with the stock pads all chewed up and gnarly as well. Just a total mess and really no excuse for manufacturers to be cutting corners like that.
However, i can somewhat sympathize with what he said about the heatsinks and overall cooling. It wouldn't hurt some manufacturers to add better heatsinks/pads/paste to at least make an effort to keep things as cool as they possibly can out of the box. A good solid quality heatsink and quality paste is more than adequate to keep things at a sufficient temp (especially for the OCD types )
As most have said, it generally doesn't matter, but two things here...
1. Staying 'far' under the rated max core temp is normally only useful for increasing clock speed over the rated* value. Thermal noise is proportional to [the square root of...]** both temperature and bandwidth, and is created by the random motion of electrons from thermal excitation. Sounds sexy, eh? So if you increase frequency w/o reducing temp, you run the risk of this induced 'thermal voltage' to create errors in the data flow, thus causing kernel panic, cache collision, memory corruption, and/or the destruction of the planet. Over time, this noise (slightly) increases, which is where the axiom to 'stay as cool as possible' originates. This is similar to a magnet's induced magnetism on another iron-containing material. The higher the material's temperature, the greater the inducement rate.
2. Consumer routers are notorious for little or no air flow; so where do you think most of that heat goes? It can cause other components to prematurely fail from the convection. So this is another source of the aforesaid axiom.
* Note that the 'rated value' and 'max temp' are entirely arbitrary, based on a bell curve of the silicone wafer's quality out of the fabrication facility; this is simple 'speed binning'.
** The thermal voltage at the terminals of any resistive element (including the millions of transistors in an SoC) is:
V = sqrt(4 * [Boltzmann constant] * T[Kelvin] * [freq bandwidth] )
This is exactly why smaller process transistors can run at lower voltages, at least once the fab gets the quality high enough. _________________ #NAT/SFE/CTF: limited speed w/ DD#Repeater issues#DD-WRT info: FAQ, Builds, Types, Modes, Changes, Demo#
OPNsense x64 5050e ITX|DD: DIR-810L, 2*EA6900@1GHz, R6300v1, RT-N66U@663, WNDR4000@533, E1500@353,
WRT54G{Lv1.1,Sv6}@250|FreshTomato: F7D8302@532|OpenWRT: F9K1119v1, RT-ACRH13, R6220, WNDR3700v4
Last edited by jwh7 on Fri Sep 15, 2017 17:47; edited 1 time in total
...and is created by the random motion of elections from thermal excitation. Sounds sexy, eh?
So every 4-8 years when elections are held, thermal noise is thus increased in our hardware? Imma go out on a limb here just a touch and say it was a typo for "Electrons".
Seriously though, that was very well explained. I can picture you saying all of that with a white lab coat on behind a desk with a bunch of torn apart hardware and wires all over the place.
jwh7 wrote:
Consumer routers are notorious for little or no air flow
I've now noticed some actually putting a small fan in there, though I cannot remember which manufacturer it was. My D-link 880 doesn't have one but has really good ventilation ports placed well throughout. It usually stays at about 68-70. Don't they make those cooling molds too you can get, like they have for Laptops? I'm sure someone can build their own too if they wanted to
...and is created by the random motion of elections from thermal excitation. Sounds sexy, eh?
So every 4-8 years when elections are held, thermal noise is thus increased in our hardware? Imma go out on a limb here just a touch and say it was a typo for "Electrons".
Trump has clearly produced a tremendous amount of thermal excitation.
LOL; nicely done. Thanks...fixed.
When I took apart my N66U to put in the microSD card, I added bolt mount spacers and a custom-made foam spacer (for the LEDs) so it could dissipate heat (and allow me to swap the microSD w/o disassembly). Looks factory. That thing has a huge heatsink in it; I decided not to reset it like I normally would. _________________ #NAT/SFE/CTF: limited speed w/ DD#Repeater issues#DD-WRT info: FAQ, Builds, Types, Modes, Changes, Demo#
OPNsense x64 5050e ITX|DD: DIR-810L, 2*EA6900@1GHz, R6300v1, RT-N66U@663, WNDR4000@533, E1500@353,
WRT54G{Lv1.1,Sv6}@250|FreshTomato: F7D8302@532|OpenWRT: F9K1119v1, RT-ACRH13, R6220, WNDR3700v4
I remember my last PC, absolute nightmare it was with the stock cooling.
The correction way is: cpu >> High-Density Polysynthetic Silver Thermal Compound >> thermal pad >> heat sinking of copper=around 5 to 10 years of used. but for economic way and OEM life of cycle then 3 years are in support to be. ( how about Apple???). heat sinking of aluminum after years will creating a lay aluminum oxidation state that increase double fact of thermal resistance. more addition, some OEM used glue Thermal Compound paste which decreasing the fact of heat transferring. you can look and searching any pc board that uses like this manner all are having problem. if you have a time and know what you doing then open router and clean heat sinking problem up. expect that you have not change or replace, upgrade router soon. other wise the cpu can handling up 120 degree C. with out heat sinking, temp of cpu float around 110 C without problem. but cause safety, most cfe have setting around 95 C. if temperature is overed, then cpu stop and router is required restart. so you neeed to deleted all 5 line codes relating in cfe to made cpu continue running without care about temperature. you can made some script setting if temperature reach 120 C then force router shut itself down. due limited of design of router, you should not used usb hd , router not intend for that.
)
So if you increase frequency w/o reducing temp, you run the risk of this induced 'thermal voltage' to create errors in the data flow, thus causing kernel panic, cache collision, memory corruption, and/or the destruction of the planet. 
Over time, this noise (slightly) increases, which is where the axiom to 'stay as cool as possible' originates. This is similar to a magnet's induced magnetism on another iron-containing material. The higher the material's temperature, the greater the inducement rate.
