Thermal probe under the IHS ???

I've got an Enermax thermal probe (see pic) and a Celeron 1.2Ghz chip. My goal is to get the thermal probe to make direct contact with the CPU die just like I had on my P3 733. What I did there was insert the probe under my waterblock and rested it up against the side of the die. This gave me perfect on die temp readings. My problem with the Celeron chip is that it's got the IHS in my way. I can't get my probe to slide under the IHS to get to the core. My assumption is that theirs some glue holding the IHS to the chip. I think if I can get that out of the way I will be able to slide my probe all the way in so it touches the side of the CPU die.

Any suggestions on how I can get some of the glue out of the way without removing the IHS and without endangering my chip ?

zoopa_man said:

I've got an Enermax thermal probe (see pic) and a Celeron 1.2Ghz chip. My goal is to get the thermal probe to make direct contact with the CPU die just like I had on my P3 733. What I did there was insert the probe under my waterblock and rested it up against the side of the die. This gave me perfect on die temp readings. My problem with the Celeron chip is that it's got the IHS in my way. I can't get my probe to slide under the IHS to get to the core. My assumption is that theirs some glue holding the IHS to the chip. I think if I can get that out of the way I will be able to slide my probe all the way in so it touches the side of the CPU die.

Any suggestions on how I can get some of the glue out of the way without removing the IHS and without endangering my chip ?

Click to expand...

Hmm this is a tough situation.

There is silicone glue on the chip, keeping the IHS on and its basically impossible to get it through the glue.

I don't know if this sounds right but take a razor blade slowly start from the corner and push the blade in abit and you should cut one little spot of the glue. Then you cut the thermal probe up, meaning cutting off the plastic cover and exposing the probe a little bit.

Hoot has more directions on how to cut the the probe to make it work more efficently.

Other than that you maybe stuck

Yah, Hoot helped me a while back with my PIII 733 and this projct. My probe allready has the extra plastic cut back. So it's thiner then what you see in the pic. I just put that up for reference. I think your right, I may be stuck. I'm hoping someone has a creative way of removing a small amnount of the glue without disrupting the IHS or killing my chip.

zoopa_man.....I'm in the same boat w/PIII 1.26 S,
be great to get the probe under the spreader. That
is, without risk??

Really annoying that the mobo manufacturers ignore the built in diode. It's been there since the venerable PII, Afaik.

Might actually be easier (and more accurate anyway) to build something hooked up to the SMBus and the internal diode (like voidyourownwarranty.net did for the Palomino).

Well......for now I'm using the Intel HS/F, I just might
drill the base of the HS, and insert the probe into the
hole w/epoxy

The stuff on voidyourownwarranty.net is interesting but a bit over my head. I'd have to do a ton of reading to figure it out for Intel chips and my Abit ST6 board. I'm good at soldering but the electronics theory behind it may get me lost.
I'll do some research and see if that is indeed an option for me...If i'm sucesfull im sure many others would love to have the input on how to do it themselves.
If I pull this off i'll have gone form quazi geek to full blown nerd. LOL

zoopa_man said:

The stuff on voidyourownwarranty.net is interesting but a bit over my head.

Click to expand...

Naah, it's just not very detailed since it's not meant as a 'how-to'. Take your time reading the docs and I'm sure you'll do well.

I'm really angry that something as current as an ST6 still doesn't read from the diode. They have no excuse, since the Intel chips have a diode for a long time now.

Anyways, I'll try mod my old GA-6BXD to read from the internal diodes once I can find the bigger Maxxim chip here. My soldering is bad enough with standard stuff, nevermind SMD parts .

I did a ton of reading and your right. It's not all that bad...I just need to find a maxxum chip that you can see LOL and where I will be able to get to my SMBus on my Abit ST6 motherboard.

I'm sure i'll end up starting a new thread for this. I want to wait for my board to come in the mail first and see if I can 't find a header to get to the SMBus. there's no way i'm soldering to my ram or to my cpu. And I need to find a place where there's a chip that i can see. those suckers are small ! they make a jumber look huge. LOL Thanks for the idea though...I can't believe I understand this stuff. I'm smarter then I think. Whooo Maybe college is worth the 8000.

Huh...what? The ST6 DOESN'T read from the diode? I'm fairly sure it does. I'm very certain that it does. If you smack it up with CPUburn, the CPU temperature in MBM goes up leaps and bounds within 10 seconds (what my MBM updates are set to). Once you take off CPUburn, it goes back down leaps and bounds within 10 seconds as well.

Look at the voidyourwarranty article, and you can see that besides being inaccurate, the socket thermistors lag the diode temperature a GREAT deal. This is especially apparent when the load is taken off the processor and the chip cools down from LOAD temps to IDLE temps; the lage

I would say the vast majority of Intel boards out there read the onchip diode.

Since I'm running my chip so cool, I can't show really dramatic contrasts between load and idle temps, but here's some snips from the interval log I made...(Sensor 1 is my unreliable ambient, unreliable because the exhaust from the Glaciator blows smack into the BX chip, and Sensor 2 is the CPU). Case temp is about 33 C

| 1/18/02 | 4:02:14 PM | 1335 MHz | 37° C | 38° C
| 1/18/02 | 4:02:13 PM | 1335 MHz | 37° C | 35° C

Check that jump out. That's when I started CPUburn. Now once I stopped CPUburn...

| 1/18/02 | 4:02:54 PM | 1335 MHz | 38° C | 36° C
| 1/18/02 | 4:02:53 PM | 1335 MHz | 38° C | 38° C

If I had more volts on this thing, the results would be much more dramatic, but I think this is sufficient.

Also, if you actually look at the socket of the Abit ST6...there is no thermistor.

Sklathill said:

Huh...what? The ST6 DOESN'T read from the diode? I'm fairly sure it does. [...]

Click to expand...

Hm. I must admit your results look good. Thanks for the insight.

The MBM page for Abit also says "P2 Diode" as recommended setting for the ST6. Could be that they just use a diode that is not onboard but in the socket (which would also explain the apparent lack of a thermistor). But I must admit it sounds a bit far fetched.

I should have done some research before suggesting taking a soldering iron to a completly innocent mobo .

zoopa_man said:

[...] And I need to find a place where there's a chip that i can see. those suckers are small !

Click to expand...

Indeed. They have the dimensions in some pdfs at the max site. I think I'd have to order at least five of them smaller 166Xers to make sure at least one makes it thru - and doesn't get lost in the carpet .

Thanks for the idea though...I can't believe I understand this stuff. I'm smarter then I think. Whooo Maybe college is worth the 8000. [/B]

Click to expand...

Heh . I hope this realisation is of more value to you than the amount of time I made you waste on reading up on all that stuff (assuming the ST6 really reads from the diode out of the box).

Of course that SMBus thermal monitor can be very handy if you want to check temps on other stuff (water temps...).

LOL @ lost in the carpet.


I'm sure i'll find something to measure. Like you said, maybe water temps or GPU temp on my GeForce. RAM temps, Clock Gen temp, Hard drive temp, Hey starting to sound like a digi Doc 5 for a fraction of the cost. Hmmmmm Things could get interesting if I can solder a few of these babies up with out loosing them in the carpet. Hehehe Thanks for the projct. This is fun and educational if nothing else.