Accurate Core Temps

[jbd]

Great article. I'm planning to monitor core temp using the same technique. I opted for the Enermax HDD Cooler w/ 2 thermal sensors (I'll use one for core and one for case temperature). It's cheaper than buying 2 individual sensors plus you get the plastic bezel to mount the LCDs in a 5-1/4 inch bay.

I wondered just how accurate this method would be, thanks for providing the answer.

[Keldogg]

If it wouldn't be too much trouble. Can you tell us what article you are talking about?

[Hoot]

Probably my article from the front page here. Here is the link:
http://www.overclockers.com/tips491/

I managed to recover the two missing pictures after it got published. You can figure out where they go.

Hoot

[Hoot]

Here's the thermistor strip. You can see the excess Capton on the tip, that needs to be trimmed back.

Hoot

[zoopa_man]

I'm thinking of doing the exact same thing. I'm just having a hard time leting go of the money. ?? I'v been looking for a setup like this for a while and the price on this isn't that bad considering the led displays are allready installed in a 5 1\2 cover. Sorry dremel tool not this time :( As accuracy goes I'm thinking this is going to be a lot better than the system temp given by the thermal probe that's built into my motherboard. Getting a faster more accurate reading is my whole reason for looking for a new temp probe setup. Temps mean life or death to me and I need a good setup. So I need someone to kick me in the head and tell me what to buy.

Great article. It helps A LOT thank you.

[coldmop]

I got a Soltek mobo that came with the same sensor but no read out, nice touch back when I got it. One issue though. Remember the talk of swirling air currents of the Sorb giving false readings, well this technique (vs. the drill method) may allow additional cooling of the probe that varies by heatsink design. Thermo-engine comes to mind. I always thought a sheet of paper with a die cutout installed between the cpu and sink would negate that factor or at least prove it a non-issue.

[Hoot]

Trust me. with only a .015 inch slit for the air to get into, You will not be troubled by collateral cooling of the sensor. Besides, I spec'd the sensor in tandem with a drilled type K thermocouple and their readings were identical, to the tenths of a degree.

Hoot

[coldmop]

Fair enough, just a though because I observed some differences, although only tried the Sorb relative to itself (with and without paper). The rest is speculation.

[Devil24]

What about mobo chipsets? same procedure?

Thanks for any replies...

[Hoot]

On mobo chipsets, it is a little more interesting since they dont have a raised core to butt up against. After I lapped the heatsink on my Northbridge and applied arctic silver, it has never been warmer than skin temperature when I checked it, so I have not worried about monitoring the temperature continually.

Hoot

[stool]

It was while attempting the same procedure with the onboard thermistor that I fried my 1g TBird. The thermistor must have been just thick enough to prevent proper seating of the HSF, and within 30seconds, zap. Needless to say, I am a bit leery of attempting this.

[Colin]

Oh poopoo! I know how it feels Stool. Can't wait for the Palomino.

[Hoot]

I thought I had covered that clearance issue adequately in my article. If you two veterans didn't catch that, then I'm worried other, less experienced folks will not also. Wish I had not lost the accompanying pictures. They would have had more merit embedded in the article, than in this thread.

Hoot

[sammy]

I liked your article so well that I ordered one. Haven't gotten it yet.
I can't tell from your pics if you cut back the capton to the T/C metal or not.
I'm assuming it has metal to metal contact with core.
Do you push the T/C in after the HSF is installed?
Maybe not, since you used hot glue, but if it's a temporary installation, could you use a stretched rubber in the .015 clearance to hold the T/C in place with the HSF installed?

I was trying to thin the probe end of a indoor/outdoor digital temp indicator, and broke, separated the T/C weld joint. Would you happen to have a way to fix one, or good place to order just the thermocouple itself?

Thanks - Sammy

[Hoot]

Let me tell you all about it...
I originally clipped off the excess capton on the tip of the strip, very carefully with a sharp pair of diagonal cutters. I didn't want to crush the thermistor, so I only cut it so close. The fact that there was still a small amount of capton left, on the tip, got to bothering me, so I put a fresh blade in my X-acto knife and tried to slice it off. The Jewelers loupe kept falling out of my eye and I tried to do it without it.
Crunch!
If anyone has one of these units and the display is bad, let me know. I have a display without a thermistor.
I ordered a new one and it should be here by Tuesday.
In the mean time, I am back to using the one I drilled the hole in the HSF base for. Dumb Dumb Dumb!

Yes, to be absolutely safe, you should push it in to butt up against the core with the HSF already in place. That way, you can rest assured that it will not impede the mating of the HSF base and the core. You can tell if you are making it all the way to the core by retracting it and looking closely at the tip. There will be a smidgeon of AS on the tip from where it exuded out around the perimeter of the core. Put it back in and once you feel confident that you have it where you want it to be, put a dab of hot glue on the zip cord, right where it passes out from beneath the edge of the HSF. I tacked mine to the mobo. Another fellow had a good idea of taping it to the ceramic to immobilize it. You would want to use a tape that is resistant to heat making the adhesive break down. IE, not Masking tape, not Duct tape, Not Scotch tape. This goes without saying, but make sure the tape you use is very thin. You only have .015 inch to spare under there.

Hoot

[sammy]

Hoot - thanks for your experience - doesn't make me feel quite so bad about crushing mine.
I've got to find a place to just buy the T/C tips.

This might sound like a rubber band fetish, but I do find a lot of maybe unconventional uses for them.

I could never keep a jewelers loupe in place, even with bags. I use the part of a small magnifier in a stand used for things like stamps, etc., and unscrew the magnifier part, which is similar to a loupe AND then I used a rubber band to hold it on a pair of glasses - spectacles type. End up with a focal distance of only about 5 inches

I was thinking that once you push the T/C probe under the HSF and against the core, you could stretch a piece of rubber band enough to squeeze it between the HSF and T/C, and when you relax it, it would hold it in place without undue force on the HSF.

[Hoot]

Personally, I would not put anything between the baseplate of the HSF and the CPU ceramic surface that even remotely requires compression to function, such as rubber. I would not want to fight the retainer clip or springs to keep that baseplate in firm contact with the core, even slightly. There are virtually limitless options for immobilizing that sensor. It's just a matter of finding what solves your particular challenge with a minimum number of opportunities for failure.
Too bad I didn't measure the thermistor value before I trashed it. I do know it is less than the 10k ohm one that radio shack sells. Besides, the R.S. one is too thick to fit. Perhaps an email to Enermax will get you a replacement unit. The replacement one I ordered is actually a dual unit already mounted on a 5-1/4 bay escutcheon. It costs $20.00 plus s/h from KD Computers. Considering single ones go for $14.99 plus s/h, it seemed like a good deal.

Hoot

[CrystalMethod]

Radio-Shack is supposed to carry those thermistors. Well, the one I bought my thermometer from, had them anyway. Maybe things are different up here in Canada. So I bought an extra one just in case. Only thing is, I was in debate about cutting off the original thermistor and just using the spare from the start to bypass all the cutting/ whittling/ shaving down of the one that comes on the thermometer. The spare comes with no insulation what so ever. But being the sucker for punishment I am, I spent an afternoon shaving down the original. Having an old slot1 celeron, I had no problems sliding it in. But I did notice something, apprently I didn't buy the same thermometer as the rest of you. The ones Hoot has pictured, only came with a thermistor encased in metal, at my RS. Oh well, the only difference I geuss is I'll have to find a niffty way to mount it. Thanks for the great article Hoot! Solved my problem of wondering it the Asus PC-Probe really is 10°C off. In my case it is, not exactly 10°, but somewhere close. all this time I've been trying to get temps under 30°@ idle, and I've already been there all this time. Sitting at 27.7°C right now.

[sammy]

Hoot - your caution about using rubber bands is certainly understandable. I just usually don't like to use tape unless there's no good alternative, and you're right, there are countless other retention methods.
I tried a couple of different sizes and I think the force is negligible.
I just got my enermax today and was pleasantly surprised at how thin it is.
I trimmed the tip down pretty easy using fingernail clippers that worked great.
I'm thinking of a way to heat and flatten the heatshrink tubing, rather than cutting it back, and without damaging something of course.

[Hoot]

That's a good idea. I've "molded" heated heatshrink many times. The callouses on my fingers from soldering actually protect them quite well. ;D

Here's what you do. You got both leads coming out of the strip with heatshrink on them. Then you have a second piece of heatshrink over both of the smaller pieces. Remove the bigger piece. then, heat the two smaller ones with a cigarette lighter or match until they're soft. Not too much, so as to unsolder the leads. While they are hot, squash them and hold until they are cool,in the same plane as the strip.

With the HSF in place slip in the strip and see if the squashed heatshrunk leads will fit. if they don't, then it's back to trimming them back. If they do, you're all set. Whattaya think?

Hoot

[CrystalMethod]

You're into pain aren't you Hoot?

[sammy]

I was going to look for a couple of small pieces of metal(but it's dark outside and got distracted) to heat and press heatshrink between them with pliers or a clamp. Like you said, sure don't want to get it too hot!

I was wondering if you had cut back the outer heatshrink tubing on the one that you broke to see what under it. Are you saying there's another piece of heatshrink under the outer one?

I looked at the radio shack online catalog and they do have some thermistors that look like they would work if they're not thick. Will go to local Shack first chance I get.

I mike'd over 0.05" across the heatshrink. I'll bet you're right about this causing some smoked cores.

[Hoot]

If you can wait a few days, I will have one of those Enermax sensors and will measure the resistance. That way, you will know which thermistor to buy. The smallest generic thermistor I could find in the different component catalogs is still 1.5mm thick. Still too much.

Hoot

[Hoot]

sammy (Jun 12, 2001 11:55 p.m.):

I was wondering if you had cut back the outer heatshrink tubing on the one that you broke to see what under it. Are you saying there's another piece of heatshrink under the outer one?


Yes, the two strip conductors eventually transition to bare wire and are soldered to the zip cord. The bare areas have small pieces of heatshrink over them. Then, a second larger piece is shrunk over the smaller two. This is probably for mechanical stability. It can be removed.

Hoot

[mj3zero]

Hi - I followed your guide (heres my photos http://www.mj3zero.pwp.blueyonder.co.uk/thermo/page_01.htm )

What I found was that at 1533 ,1.96 volts my Tbird with MC462/Delta is maxing at 60.4 'c doing a long stint of BurnK6.

This is with an ambient temp of 26'c.

I tried it at 1333 , 1.85 (same settings as the overclockers MC462 review) and it maxes at 50'c doing RC5.

So - is my MC462 knackered ?

How do overclockers test - is it in a closed case or an open system ?

My result so far show that the Abit Kt7a socket thermistor is between 7 to 10 'c below real temps.


any suggestions ?

mj

[Hoot]

First off, I want to compliment the process you used to arrive at the final solution. Good photojournalism.

Something is definitely amiss with your MC-462A though. With an ambient of ~25C, the resultant CPU temp is way out of whack. Do you absolutely have to use that shim? General wisdom has it that they cause a slightly higher temperature. If it interferes with a tigh mate between the MC-462 base and the core, that would explain the extreme temperature. You do not have to worry about the MC-462 damaging your CPU. Do try again without the copper shim. You can increase the accuracy of the KT7A in-socket thermistor substantially by bending it slightly upward so that when the CPU is in the socket, the head of the thermistor rests firmly up against the bottom of the CPU. Before you lock the zif socket the CPU should ride slightly higher because it rests upon the lifted thermistor head. You push the CPU down into the socket and then lock the zif lever. I put a small dib of AS on the in-socket thermistor head, right where it contacts the bottom of the CPU, for even better thermal conductivity. Using that approach, the in-socket thermistor reads within .5C of the on-die thermistor once the temp has stabilized. During transition, either up or down, the in-socket thermistor is a few degrees behind the on-die thermistor, due to thermal propagation delay through the CPU ceramic. Lastly, make sure the Delta fan is blowing down into the HS, not sucking out. It makes a difference on the MC-462.
As for testing open vs closed case. When I am characterizing the performance of an HSF, I test with the case open, since I am not interested in characterizing the cases performance. When I am characterizing the entire systems cooling, I run case closed because that is how it will be used.

Hoot

[mj3zero]

Hi again,

my rig is now running wihtout the shim (forgot to mention that) and the temps are exactly the same. I did the socket thermistor bend / AS trick too.

I definitely think there is something amiss with the MC462 -

I have tried a Thermoengine /Delta and got load temps (socket temps) of 45'c with an ambient of 22 (case open though) and with a PEP66 w/Delta I got a load temp of 48 (case open) and an ambient of 22. This was at 1533, 1.96v. The MC462 gave the same load temp as the TE but with a higher ambient (26) , again with the case open.

What I have noticed is that my motherboard is warped slightly under the zif socket. Although I have made sure that the MC is sitting ok by taking it off and checking the AS2 residue trace.

I am going to try out the TE w/Delta tomorrow and see what sort of "real" temps I get with that.

cheers for the help
mj

btw- did you see the pic of the thermistor glued to the Tbird? Notice the fingerprint ? I nearly stuck my finger to the chip :) Just got it off in time and than had to scrape the super glue off with a razor - what a dummy I am :)

[Hoot]

We have all left the occasional "lasting impression" on our glue work. ;D

I went back and looked over your pictures closely. The shoulder washers between the brass standoffs and the kt7a look different than the ones on mine. They seem to be thicker. If that was a problem, I don't think you would have gotten such a good dispersion of the AS on the core in the pictures though.

I am stumped.

Hoot

[mj3zero]

Hi again,

the shoulder washers that came with mine are slightly too "small" for the holes in my board and move about slightly. Are yours the same or were they "snug" ? Out of interest - how "hard" do you have to press he screw/spring down to get it threaded? On mine it only takes a movement of a few mil's to get the screws threaded - whereas if I use a couple of extra washers it is more of a "press" -BUT the temps are lower by a couple of degrees.


I read your mod'd MC462 post and I replied to it - the gist being that there does appear to be something wrong with the thermal transfer on my MC. The pins do get warm but not hot , whereas the base and the motherboard area is "toasty". Have you heard of anyone else having high temps with this cooler ? There have been a couple of reports similar to mine here in the UK (at the overclockingstores forum).

cheers for the help
mj

[Hoot]

On my KT7A, the shoulder washers fit just right. Not too loose and certainly not snug. I tried adding washers to the spring tensioning screws and more than 2 got no improvement in temps. I am currently using one extra washer on each. If you bent that in-socket thermistor up like I mentioned, you should be able to get sufficiently accurate readings from it. I would remove the thermistor you worked so hard to install, clean everything up and see what the in-socket thermistor reads. This is just to test whether it could be something you are causing with the slip-in thermistor arrangement.

Hoot