KT133A chipset temperature monitor!

[Leo V]

I just found another cool thing to do with my A7V133 mobo:

I took an unused external temperature sensor left over from my good old Abit BE6. The A7V133 actually has a header for such external sensors! (It's called something like auxiliary power supply sensor.)

Well, since I'm running my KT133A chipset without the fan, I'd like to know how hot it gets. I've plugged in the heat sensor (it's like a tiny little tongue), and jammed it between the teeth of the KT133A heatsink. I'll note that I've already sanded that heatsink to perfection, and used thermal grease for better heat transfer. Voila, the BIOS as well as Sandra2001 give me dynamic, infernal temperature readings straight from the KT133A!

The chipset temperature is currently 46.0C/115F, I'll run a few Quake3 timedemos and report back the new temperatures.

 

[Leo V]

Okay, after about 12 consecutive Quake3 timedemos (CPU bound), the temperature reads:

CPU temp: 58.0C/136.4F (accurate)
KT133A temp: 48.0C/118.4F
Board temp: 29.0C/84.2F

 

[Mikewarrior2]

How do you know hte cpu temp is accurate?

BTW, you should try putting the probe in between heatsink and touching northbridge chip edge..


Man, that sucker must run hot, though...



Mike

 

[Leo V]

I've just read VIA's tech specs for the KT133A Northbridge. Its max operating temperature is rated at [update:] 85 Celsius! Although max power dissipation is "only" 3.5 watts!

I don't like to jam sensors under the heatsink, since that can drastically interfere with cooling. However, my current arrangement allows rather close monitoring.

What did I mean by "realistic"? I've read somewhere that A7V133's temp sensor is directly under the CPU. My MSI Pro2A reported much less realistic temperatures--for the same level of heatsink "hotness" the MSI would report 42C and the Asus would report 63C, to give you the general idea. Probably neither is perfect

 

[Viperoni]

<< A7V133's temp sensor is directly under the CPU >>



You don't know what you got yourself into Leo :Q

 

[Mikewarrior2]

you don't need to put the probe between northbridge and heatsink, just touchign the edge of the northbridge chip, under the heatsink and mb... this way it doesn't interfere with the heatsink.

As far as the temp probe, yes, it is much better at &quot;approximating&quot; temp than the MSI board, but still far from actually being accurate.


Mike

 

[Leo V]

I've tried attaching the probe edgewise like you've described. The thing is, it wouldn't hold there without adhesive. Maybe I'll give it another try.

 

[Mikewarrior2]

you could try taping it down to the mb, using electrical tape or something like that...


Mike

 

[Leo V]

Okay, I've gotten the sensor stuck edgewise between the chipset &amp; heatsink. The readings are actually lower! I'd guess for these reasons:

1) The most heat is generated at the center, not edges
2) The heatsink is very thin; in my original configuration (center of heatsink), the sensor was actually much closer to the chipset.

I'll make an improvement once I get Arctic Silver--I'll apply it to the heatsink, as well as between the sensor &amp; heatsink teeth.

 

[Mikewarrior2]

is it the stock heatsink?


Mike

 

[Leo V]

Yes, I unscrewed the fan and sanded the stock heatsink (lowest to highest grit, until it produced a reflection). I re-mounted this heatsink and used a conservative amount of thermal grease.

 

[Insane3D]

Leo, the in socket thermistor of all socket A boards just reads the temp of the bottom of the CPU PCB. It has about 6 or 7 layers of ceramic between it and the core... I used a flat style thermistor like you mentioned and slid it in between the HSF and the CPU PCB with the tip just touching the side of the core. This measurement is, on average about 5C warmer than the socket thermistor reading on my board.

 

[MCS]

The heatsink will most likely be hotter than the chip if it is doing its job properly (transfering the heat).

Thats why your readings are lower.

 

[Kenneth]

&quot;The heatsink will most likely be hotter than the chip if it is doing its job properly (transfering the heat).&quot;

This isn't thermodynamically possible, since heat exchange is always from an area of higher temperature to an area of lower temperature. Since the heatsink is absorbing heat from the chip, it must by definition be cooler than the chip.

 

[Neo_Geo]

Kenneth - You are absolutely correct.

Glad to see someone has some decent knowledge of thermo around here.

 

[MCS]

Thanks for correcting me.

 

[Charles]

Leo, how long does your A7V133 take to boot up?