Temperature check missing in AC66U firmware 354?

[njweb]

I know we are able to check the temperature of the N66U via the web admin GUI.

I noticed with my AC66U, which just started using yesterday, that there is nowhere to check the router's internal temperature (at least not in beta firmware 354)?

Is / are there any way / ways to check it (without switching to alternative firmware)?

 

[joegreat]

Is / are there any way / ways to check it (without switching to alternative firmware)?

Hi,

On the GUI only Merlin's FW would support the temp. display, and is still based on the ASUS firmware!

There is by sure a command line tool which will give you the information - only I do not know it...

With kind regards
Joe

 

[njweb]

Hi,

On the GUI only Merlin's FW would support the temp. display, and is still based on the ASUS firmware!

There is by sure a command line tool which will give you the information - only I do not know it...

With kind regards
Joe

Thanks!
Interesting that Asus chose not to include it on the AC66U (since they do / used to on the N66U).
I was also thinking there has to be a command line equivalent (maybe Merlin can post details on the specific command(s) and instructions on which cmd line interface to use).
If not, I may just have to try Merlin's magic when his version of fw 354 is ready.
To be frank I typically use stock firmware (unless it has issues that a custom firmware resolves). A situation like this is a good example of when
I might use an alternative.

I'll have to first ask Merlin if he is confident that his temperature is accurate given a post I read a week or so ago on the DD-WRT forums between Merlin (Eric) and BrainSlayer:
BrainSlayer wrote i took the code from asus. the formula you show is not the formula asus is using and it works also only if the wireless driver has been modified which isnt possible for the most users since they dont have the source codes unlike me

Eric replied:
then look again into your gpl archive. there are 2 places. within the webgui they simply just show the returning value. at the fan control daemon a average of 2.4 and 5 ghz interface is build.
in basic scheeme. 24iftemp + 5iftemp / 2

but its a little bit more complex since asus does a moving average by:
24value = (24value * 4 + new24value) / 5;
same for 5 ghz.

but there is no offset of 20 or something like that, your formula is wrong at 100%

take a look at rc/phy_tempsense.c for the moving average. the other location simply just uses the return value of the driver. but both code pieces indicated that the returned value is the temperature itself


This is on what I based my formula, from Advanced_PerformanceTuning_Content.asp:

Code:

function convertTemp(__coreTmp_2, __coreTmp_5, _method){
if(_method == 0)
return parseInt(__coreTmp_2)*0.5+20;
else
return (parseInt(__coreTmp_2)-20)*2;
}


This is where Asus used that 20 offset, hence I'm completely confused now. "



I did not see anything after thr above exchange, so I don't know what the conclusion was (smile) << I reached the per message image limit (includes emoticons).

 

[RMerlin]

The temperature was only on a (dead) page because the original RT-N66U had a cooling fan, which would throttle its speed based on temperature. That page was dropped after the fan was removed from the design, so out of the box the RT-N66U did not display any temperature either.

Brainslayer ended up changing DD-WRT to also use the same formula as me (that can be seen on the SVN timeline).

You have to apply the formula to the values returned by these:

admin@RT-N66U:/tmp/home/root# wl -i eth1 phy_tempsense
61 (0x3d)
admin@RT-N66U:/tmp/home/root# wl -i eth2 phy_tempsense
67 (0x43)

 

[njweb]

The temperature was only on a (dead) page because the original RT-N66U had a cooling fan, which would throttle its speed based on temperature. That page was dropped after the fan was removed from the design, so out of the box the RT-N66U did not display any temperature either.

Brainslayer ended up changing DD-WRT to also use the same formula as me (that can be seen on the SVN timeline).

You have to apply the formula to the values returned by these:

admin@RT-N66U:/tmp/home/root# wl -i eth1 phy_tempsense
61 (0x3d)
admin@RT-N66U:/tmp/home/root# wl -i eth2 phy_tempsense
67 (0x43)

I recall having a page with the temperature display (advanced tuning or performance tuning, something to that effect) while using the stock N66U firmwares (not sure which one though).
I just found the reference:
http://forums.smallnetbuilder.com/showthread.php?p=41269
So it was a hidden page that had to be accessed separately.
That explains it...

"http://<YOUR ROUTER IP>/Advanced_PerformanceTuning_Content.asp
or if default
http://192.168.1.1/Advanced_Performa...ng_Content.asp "

Hopefully this feature is still available (even if not officially supported) and works on the AC66U as well...
EDIT - Looks like it is no longer available, if I understood your post I noticed on page 6 of that same thread...


Thanks a lot! That's good news re. the code and formula for the temperature being confirmed / sorted out.

 

[RMerlin]

Thanks a lot! That's good news re. the code and formula for the temperature being confirmed / sorted out.

We're still not 100% sure about the formula. This is the formula Asus uses to calculate the temperature, however it does look a bit weird (Brainslayer called it "Voodoo math" on the SVN commit LOL), so we can't be 100% sure. Personally, I think there's a good chance this formula is indeed correct. Intel has a similar method of calculating the temperature of their CPU. The internal register returns the temperature in reference to tJunction, which is typically 100C for most CPUs. So if the register contains 40, then the CPU temperature is 100-40, so 60C.

The only way to be sure would be to have access to the Broadcom HW sheets, and knowing how Asus implemented it (sometimes, the value might be different based on an external resistor connected to some specific CPU pins, for example). Or to have someone with an infrared thermometer to see if the temperatures do make sense (keeping in mind this value is probably the internal, not external temperature).

Overall, I'd say: take it for what it is, which is, a reference value. It will at least allow you to see if the temperature is higher than other people report, or if there is a sudden increase in temperature. So long everyone calculates it the same way, it will be useful enough.

 

[njweb]

We're still not 100% sure about the formula. This is the formula Asus uses to calculate the temperature, however it does look a bit weird (Brainslayer called it "Voodoo math" on the SVN commit LOL), so we can't be 100% sure. Personally, I think there's a good chance this formula is indeed correct. Intel has a similar method of calculating the temperature of their CPU. The internal register returns the temperature in reference to tJunction, which is typically 100C for most CPUs. So if the register contains 40, then the CPU temperature is 100-40, so 60C.

The only way to be sure would be to have access to the Broadcom HW sheets, and knowing how Asus implemented it (sometimes, the value might be different based on an external resistor connected to some specific CPU pins, for example). Or to have someone with an infrared thermometer to see if the temperatures do make sense (keeping in mind this value is probably the internal, not external temperature).

Overall, I'd say: take it for what it is, which is, a reference value. It will at least allow you to see if the temperature is higher than other people report, or if there is a sudden increase in temperature. So long everyone calculates it the same way, it will be useful enough.

I see, thanks! Yes, it could be quite an ordeal to determine if it is accurate by factoring in all the variables, it it isn't as simple as measuring a temperature and subtracting that from a fixed value (such as the tJunction example).
Given the doubt you noted, to be safe I indeed plan to use it as a reference value / gauge for relative measurements only, as opposed to expeciting accurate absolute measurements (not that they aren't but nobody is 100% sure currently, as you noted).