Abit IS7 reports wrong CPU Temp... I confirmed..

[mixerman]

I have the Abit IS7 and it reports to me through Motherboard Monitor that my CPU core temp is about 41C on average when the computer is idle. On my Thermaltake Xaser III case I have the thermal diode installed like thermaltake says to by putting the diode underneath the CPU chip between the pins. This diode reads about 32C at the same time that MBM 5 is reading from my CPU core a high 41C. Then when I am at load running Prime95 MBM 5 reports to me a CPU core temp of about 50C and my Hardcano 9's thermal diode is reading only about 40-41C.

So I think this is pretty good proof that the Abit boards DEFINATELY do report temps about 8-10C hotter than they actually are. Supposedly Abit claims they measure the temps from a different chip or something than other manufacturers do and that's why the temps are higher.. but in their minds MORE accurate then the rest. Well I would have to probably trust my Hardcano 9's probe at this point, being I PHYSICALLY installed in under my CPU chip and the CPU chip is sitting smush right ontop of it.

So I am definately convinced now that the IS7/IC7 series of motherboards report temps 8-10C higher then they actually are..

Question.. in order to fit the thermal diode under the CPU chip, I had to bend the diode in half because it was too long. Would this affect temp readings or not?

And when I put the diode under the CPU chip, the CPU chip didn't sit exactly flat in the socket, so I had to manually push down on the CPU chip and then close the socket latch. I think the chip is sitting a bit higher on the side the wires come out then it is on the other side of the chip. Is this is a problem? When I mounted my SLK-900u, I just put it on the chip and then tigheted all the spring loaded screws until they couldn't go anymore..
So does it really matter if you CPU chip doesn't sit exactly flat in the socket and one side is higher then the other?

NEW SYSTEM:
P4 2.8c (800fsb), Abit IS7, Kingston HyperX PC3500 (2 x 512mb), IBM Deskstar (60 gig @ 7200rpm)
OLDER SYSTEM:
P3 850mhz (100fsb), Asus CUV4X, 620mb Micron PC133, WD 40gig 7200rpm

 

[jihiggs]

since when has any external thermistor been accurate? what do you mean you had to bend it in half? there should be a visible area on it where two leads come together at a little bump. that is the temp probe, if thats not directly touching the chip then it will be wrong.

<font color=red> black </font color=red> <font color=green> white </font color=green> <font color=blue> yellow </font color=blue> <font color=orange> purple </font color=orange> <font color=black> red </font color=black> <font color=yellow> green </font color=yellow> <font color=purple> blue </font color=purple>

 

[Crashman]

Basically, the CPU die is on the top, the bottom is not as hot because...it's insulated! The only way to get an accurate reading would be to put the sensor between the heat spreader and the cooler, but that wouldn't work because it would hold the cooler off of the heat spreader! Of course you could drill a hole in the heatsink and force the heat sensor down to the heat spreader through that hole...

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>

 

[pIII_Man]

I don't beleive garolite or whatever cpu's are printed on conducts heat very well...you would be better off putting the temp sensor next to your cpu...touching the side than putting it underneath that would be a more direct contact IMO..don't p4's have an intigrated thermal couple anyway? If thats the case you could just use an aftermarket program that reads from this thermal couple?
Happy 4th of july

 

[Crashman]

That's what I was thinking myself about the integrated thermalcouple

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>

 

[mixerman]

Okay..

1.) I had to bend it because the actual flat prob part was say about an inch long. That center of the bottom of the P4 chip between the pins isn't that long, so I had to bend the probe in half so it would fit. I know that at the end the two probe wires come together and that is where it gets the reading from. That part is touching the bottom of the cpu chip.. I just didn't know if by bending it, that could break the probe or cause the readings to be off. I don't think it did because I get very good solid readings and the readings are always about 8-10c below what MBM 5 reports to me..

2.) Yes I know the p4 chip has a built in sensor, so that is what I see in MBM 5.. BUT.. if you go over to the Abit forums you will see that Abit Motherboards are using a different calculation to figure out the correct cpu temp, thus making all of the Abit boards (well Is6, Ic7) read higher temps than the rest of the boards out there. So by using this temp probe of the Xaser III case, I am pretty confident that the CPU temps are actually lower than what MBM 5 is reporting to me from the CPU core off the Abit board.. I understand that the actual core is on the top of the chip, and that by reading the probe from the bottom it might be a little lower, but I don't think it would be 8-10c lower!

3.) I asked if you guys think it is a problem that by putting the probe under the cpu chip, the wires have to get fed through the pins and out the side, well the side where the wires come out, the chip is sitting a little higher there in the socket. So in effect, the cpu chip doesn't sit perfectly flat in the socket and I had to push down with some force before locking the socket level to try and get it as level as possible, but it still isn't as level as it would be if all the sides of the chip sit perfectly flat on the socket... so I was wondering, is this a problem for either the chip or for the slk-900u that's sitting on top of it? I am not sure if the slk-900u is making a level even contact with the chip, it feels like it is, but by knowing the chip isn't perfectly flat in the socket makes me wonder.. should I be worried? My temps see fine.. MBM 5 is reporting 36C CPU, 29C Case, 27C PWM while I am writing this post.. so my computer isn't under stress right now, but even when under stress from Prime95 my CPU only goes up to about 50c in MBM 5 and up to about 42C on the Xaser III's thermal probe that sits under the CPU... so do I have anything to worry about?


NEW SYSTEM:
P4 2.8c (800fsb), Abit IS7, Kingston HyperX PC3500 (2 x 512mb), IBM Deskstar (60 gig @ 7200rpm)
OLDER SYSTEM:
P3 850mhz (100fsb), Asus CUV4X, 620mb Micron PC133, WD 40gig 7200rpm

 

[xeenrecoil]

heya mixerman:

I suggest you remove that sensor out from under the CPU, as it is causing a gap in which the result is your CPU is sitting crooked in the socket, there is a real possibility that something may happen and short out the CPU, as dust and other particles small enough to enter through the gap could do so and in result cause damage.
Also it is possible to cause physical damage to the pins on the CPU, which would void the warantee of the CPU.

XeeN

 

[mixerman]

then why the hell does Thermaltake tell you to put it there..?

Obviously the chip can't sit perfectly flat because the wires need to come out of the socket somehow..


NEW SYSTEM:
P4 2.8c (800fsb), Abit IS7, Kingston HyperX PC3500 (2 x 512mb), IBM Deskstar (60 gig @ 7200rpm)
OLDER SYSTEM:
P3 850mhz (100fsb), Asus CUV4X, 620mb Micron PC133, WD 40gig 7200rpm

 

[JimmyDean]

Lemme get this straight.. You put the wires over the socket, then the cpu in the socket ontop of the wires? That seems a very bad idea, of course the cpu cant sit flat, if it doesnt the hsf wont contact as well either. If I were you Id take it out immediately

<i>Royal Fusileres, Company C</i>

 

[mixerman]

yep well this is how Thermaltake wanted it..
it seems to be working fine..


NEW SYSTEM:
P4 2.8c (800fsb), Abit IS7, Kingston HyperX PC3500 (2 x 512mb), IBM Deskstar (60 gig @ 7200rpm)
OLDER SYSTEM:
P3 850mhz (100fsb), Asus CUV4X, 620mb Micron PC133, WD 40gig 7200rpm