scan wont admit theres a problem with my q6600

[moogle]

he's not listening. we've said to him to run it with the intel hsf normally and let it damage itself. if the temps are that bad at idle I'm sure it'll go soon. Also with the inferior intel hsf if the temps are so high the hsf would be hot hot hot. If not then he would know it's a faulty sensor. But he just doesn't seem to listen...

[Grey M@a]

he's not listening. we've said to him to run it with the intel hsf normally and let it damage itself. if the temps are that bad at idle I'm sure it'll go soon. Also with the inferior intel hsf if the temps are so high the hsf would be hot hot hot. If not then he would know it's a faulty sensor. But he just doesn't seem to listen...

I wouldn't say that the high 40's is a concern at idle. Although what case is he using and what fans are in the case. The amount of times I see cases with poor fan setups which creates heat spots all over the place. Mine idles high as I crammed a 6600 quad, 4gb memory and a full size dual slot 8800GTS into a shuttle sx38p2 pro so I expect high temps tbh

[stevie lee]

polly 69.. added apology in previous post.. further to that pm you sent..

[rabbid]

he claims to be using a lot of fans, but we arent told which directions they are facing, maybe they need to be looked at as well as they can be cancelling each other out if poorly setup.

[moogle]

he claims to be using a lot of fans, but we arent told which directions they are facing, maybe they need to be looked at as well as they can be cancelling each other out if poorly setup.

easiest way to rule out it's fans would be to make a temp test bed out of the case.

[Saracen]

Ok, for the last time, THE SENSOR IS *ON THE CPU DIE*. You could try a thousand different motherboards and still get the exact same reading, because they're all reading from the same damn sensor i/o circuit.

Erm .... I'm no Intel engineer, but that isn't quite how i understood it.

Yes, the case sensor is on the chip itself. But it gives an analog reading. The AD conversion, which gives the results that will be displayed in BIOS or by monitoring software, takes place on the motherboard.

So yes, the CPU provides the sensor, and thus will perform in the same way (though not necessarily give identical readings if external factors have changed, such as a different mobo giving a slightly different mounting location which results is chassis airflow differences affecting the hsf to ambient temp differential), but the results reported will vary if the motherboard calibration or AD conversion is different.

We agree entirely on the implications, aidan, but unless I'm mistaken in the above, the second half of your quoted statement (about the 1000 mobo's) isn't necessarily the case. In fact, it's that variation in ADS conversion and calibration that is part of the reason why user temp results aren't an accurate way of assessing if a chip is in-spec or not. As I understand it. I do, of course, stand to be corrected if anyone knows better.



Note : In the above, I'm referring to Tcase, not Tjunction, for which the process is different, but the oft-quoted 71c refers to Tcase not Tjunction.

[Saracen]

easiest way to rule out it's fans would be to make a temp test bed out of the case.

That's one way. Another is to point a domestic 15" fan at the open side of the case. It'll shift ALL the air inside the case out of the way in a hurry, and as long as you take care not to cause back-pressure on the hsf fan, would resolve case fans as being an issue. Anyone who does this does it at their own risk. I've used it an I'm happy to taker the risk. But anyone else doing it takes their own risk.

But two things. One, polly says he's experienced, so presumably he knows enough to know which way round fans are ... and should be (*). Second, it won't move the situation on much in that it won't help determine if the CPU is within Intel spec or not, since Intel spec requires a compliant overall system thermal profile. It might be an interesting diagnostic point, but won't move the situation on much either way.




* Note : which way round they should be is a bit of an open topic. System airflow is not an easy thing to predict, and ultimately, it comes down to suck it and see. Also, of course, adding more and more fans doesn't necessarily improve either airflow or system cooling. Again, as an experienced user, polly will presumably know that, too.

[aidanjt]

Erm .... I'm no Intel engineer, but that isn't quite how i understood it.

Yes, the case sensor is on the chip itself. But it gives an analog reading. The AD conversion, which gives the results that will be displayed in BIOS or by monitoring software, takes place on the motherboard.

So yes, the CPU provides the sensor, and thus will perform in the same way (though not necessarily give identical readings if external factors have changed, such as a different mobo giving a slightly different mounting location which results is chassis airflow differences affecting the hsf to ambient temp differential), but the results reported will vary if the motherboard calibration or AD conversion is different.

We agree entirely on the implications, aidan, but unless I'm mistaken in the above, the second half of your quoted statement (about the 1000 mobo's) isn't necessarily the case. In fact, it's that variation in ADS conversion and calibration that is part of the reason why user temp results aren't an accurate way of assessing if a chip is in-spec or not. As I understand it. I do, of course, stand to be corrected if anyone knows better.



Note : In the above, I'm referring to Tcase, not Tjunction, for which the process is different, but the oft-quoted 71c refers to Tcase not Tjunction.

All correct, but if the Tcase sensor wire is too thick, or in this case, too thin. The readings will show an out of proportion temperature increase, and all motherboard sensor chips will be mis-calibrated in equal proportions and show similar readings.

[phill8800GT]

Did you have any screen shots of this? Idle and load temps? Pics of the heatsink/fan combination that was used at all??

[Grey M@a]

This is the hottest mine gets running SMP F@H on a 6600

[Wiczy]

All correct, but if the Tcase sensor wire is too thick, or in this case, too thin. The readings will show an out of proportion temperature increase, and all motherboard sensor chips will be mis-calibrated in equal proportions and show similar readings.

Surely 'too thick' or 'too thin' indicates a manufacturing fault? It certainly does to me.

[Chris P]

This thread is really being beaten to death now - So is locked

Unless the original poster would like to post in Return to Mark's post? I will re-open the thread

Regards