Help me interpret my Galciator results!

[jsford]

These temps were taken with MBM, Ambient temp 76F
Tbird 1.33 GHZ @ 1.4 1.78v, Load temps were Sisoft Burnin
and Prime 95 running at once.

Thermo w/ Delta Idle 91F Load 102F
Glaciator Idle 104 Load 114
Case temp stays at 80 (120mm 130CFM blowhole)

Seems to me Thermo definitely cooling off the sensor better
but not necesarily the core. But you know what? IT QUIET IN HERE.
I can actually HEAR my 120mm monster over my CPU fan LOL.

I can live with 114, for this peace I now feel. hehe

 

[stroligo]

I've been doing that, one way or another, the last twelve hours.

The serious answer: The MBM temperature on the Glaciator is probably going to be at least 5C higher than actual CPU heatsink-juncture temperature. The Thermoengine, at least according to our test, is roughly the same. Joe and I are trying to figure out how we are going to convince the universe not to take the MBM numbers as Gospel truth, 'cause they aren't.

 

[deathdealer]

Ed, if what you say is true help me interpret this;
1.2 G AXIA 266 Tbird @1466 on a Iwill KK266 @1.85v. Readings after 15 of CPU warming utility.

MTS Glaciator
22' C ambient, 29' C MBT, 48'C CPU

Thermoengine w/ Delta 30 cfm
23' C ambient, 30' C MBT, 42' C CPU

If both are off by 5' C it still doesn't add up. Even with using a 20 C/W on the Glaciator, (84w X .20 = 16.8), I should be getting a reading of around 40' C (39.8' C actual), not 48' C. What's up here??

 

[stroligo]

I said the Glaciator (and Glaciator alone) would be at least 5C higher.

Per your calculation, case and motherboard temperatures have been thrown around interchangably, let me get back to you tomorrow to make absolutely sure I got this right.

 

[William]

ok, according to radiate, you are producing 84.2W of heat with that TBird. Your Ambient is 29C. We will use the average C/W oc.com in the test which is .15. So the best temp you can expect is (84.2*.15)+29 = 41.63C. That is the best to expect. But of course, are you using the best hermal compound, Artic Silver? Did you apply your goop correctly? Is your fan spinning at its rated speed? Every one of these can add to the higher than optimum temp.

 

[coolmojoe]

Temps still seem to be going up several degrees after installing the Glaciator and comparing results to baseline MBM temps taken with old HSFs.

It shouldn't matter that MBM isn't a perfect measurement of the actual core temp if it's at least a fair comparison of apples to apples in the same location near the core.

The Glaciator so far seems to perform like the best aluminum heatsinks paired with a quiet fan. I was expecting to see better performance from this design, considering the fan can't be upgraded and the expense of copper.

I hope I'm wrong, but fear I'm right

 

[William]

whats happening is that people are finding that temps average a bit higher than a top notch aluminum hsf but that they can overclock higher. This is just the copper effect of holding heat in more, but still being able to dissipate it away from the core. Just shows the Swifttech is still king.

 

[Oni]

Where can i pick up a copy of Radiate? I looked at the chart the other day, and they had my processor speed listed, but the voltage didn't go as low as mine (1.50) so I don't know how accurate my assumption was (It said at 1.7, its at 66 watts, so I used 63). Can somebody help me out?

 

[William]

Maestro (Jun 16, 2001 08:34 p.m.):
I sitll like the Fop 38 or Swiftech are better . The Glaciator is still too new to jump all over it.

Maestro

The 462 is still the king of hsf, but the Glaciator is better than the Fop, WBK, or ThermoEngine.

 

[Megahurtz]

Oni (Jun 16, 2001 08:29 p.m.):
Where can i pick up a copy of Radiate? I looked at the chart the other day, and they had my processor speed listed, but the voltage didn't go as low as mine (1.50) so I don't know how accurate my assumption was (It said at 1.7, its at 66 watts, so I used 63). Can somebody help me out?

Yes, you can D/L Radiate, here.

T

 

[seadog]

Ok, FWIW here's another Glaciator experience. Been running a Taisol copperbottom w/Delta black label at 1.45ghz. Idles at 37C, tops out in P95 at 43C - case temp 23C, ambient 26C.

Just installed the Glaciator and saw idle temps at 42C. Reinstalled several times with no luck. Removed and checked base with 10x magnifier and noticed it was "textured" so I spent 2 hrs lapping the copper with up to 600 grit.

Reinstalled the hs. Now idles at 40C, crashes in P95 at 45C - case temp 22C, ambient 26C. With the Taisol, my max stable p95 was 1.45ghz, with the Glaciator I'm getting rounding errors at only 1.44ghz.

Nothing else in the system has changed. Antec sx830 - two 47cfm panaflos in front, two 34cfm fans in back, Enermax psu w/2fans

Now I'm not saying it's a bad hsf, but there is something going on that is preventing me from getting the C/W that this heatsink should be capable of providing.

Hopefully further review will shed some light on the problem.

 

[deathdealer]

William (Jun 16, 2001 05:49 p.m.):
ok, according to radiate, you are producing 84.2W of heat with that TBird. Your Ambient is 29C. We will use the average C/W oc.com in the test which is .15. So the best temp you can expect is (84.2*.15)+29 = 41.63C. That is the best to expect. But of course, are you using the best hermal compound, Artic Silver? Did you apply your goop correctly? Is your fan spinning at its rated speed? Every one of these can add to the higher than optimum temp.

Your reply is confusing me now. My case air temp/ambient temp is 22' C, now if you are figuring C/W based on MBT and not ambient (case air temp) why does Joe use the word "ambient" and not MBT (motherboard temp) in his explaination of the C/W formula? Even using MBT instead of ambient and using a 20 C/W rating the end # is 45.84' C not 48' C
Yes I used Artic Silver 2, MTS also included more with the sink. Yes it was applied as per instuctions from AS's website, ( I have installed heatsinks with thermal compound many times before). Yes fan was running at rated speed 5700 RPM, (it should be noted Joe's review mentioned a 6200 RPM fan in his final review, was that a typo?). Case is an Antec SX-1030 with (4) Sunon 42cfm fans so adequate case cooling is available. The numbers sure come up much less than was in the review. Still wondering about whats the deal on the differing fan rpm's?

 

[AZZKICKER]

here is the deal

the temp probe on the mobo its touching the ceramic

the thermoengine with its design cools the ceramic and the Gladiator donnt

so the temp on the probe is going to be lower for the thermoengine

dont mean the core is hotter or cooler

 

[Thelemac]

How are you getting your ambient temp? Are you putting a thermister close to your hsf? If not, then it probably not accurate enough to give you a good idea.

HOW you get your temps is probably the most important thing if you're going to be comparing things. For making sure your processor isn't going to be dying anytime soon, MBM works just fine. For comparisons, forget it. You can't possibly hope to get anything accurate enough to do a comparison between hsfs.

 

[Hoot]

There's a "fly in the ointment" of the reasoning behind this thread, somewhere. I can't pin it down, but my gut is telling me so.

Let me see if I can put this into words...

We are challenged with getting the heat out of the core. The majority of the heat gets out of the core (hopefully) via the contact it has with the heatsink baseplate. Obviously, the core is also in contact with the ceramic carrier it is embedded in. Therefore, some of the heat generated by the core will migrate into the ceramic carrier if that carrier is cooler than the core. So, the ceramic is in parallel with the heatsink baseplate as a thermal conductor. It has greater resistance to heat flow (hopefully) than the baseplate, so less heat flows through it in a given amount of time. Any path that helps, no matter how little, with getting the heat out of the core, is a good thing. Here it comes ;D

So, why is it so bad if the ceramic carrier is cooler with one HSF than another. I'm not talking about the impact that its temp has on measurement, but rather on performance. Personally, I'll take any help I can get, removing heat from the core.

If the ThermoEngine, because of its tapered base, cools the ceramic more than a wide base, so much the better. Somehow, I doubt the cooling effect on the ceramic is compromising the in-socket thermistor that much. I'm assuming you don't just have the thermistor contacting the ceramic off-center, as it comes with the board, but bent to make contact with the ceramic dead-center under the core.

Except during transition from cold to hot, or hot to cold, I find my in-socket thermistor to be within .5C of the reading I get from my drilled HSF base with a thermocouple sitting directly on top of the core. And, that is with a veritable hurricane of airflow going on inside my case. sure, there is a response delay, but once the core temp has stabilized, they are very close in readings.

And another thing! (ala John Madden) I would be more concerned about the impact the HSF baseplate has upon a drilled and embedded thermocouple. In HSFs, with good heat spreading characteristics like the MC-462A with its 3/8 inch thick copper slab. With the thermocouple surrounded by that copper, as well as in contact with the top of the core. That copper has to be wicking away some of the heat which is registering in the thermocouple junction. A junction that I might mention, should not be soldered, but welded, to avoid having the lead/tin junction interacting with the dissimilar metals.

Now, where's my 4-legged turkey...

Hoot

 

[Randy]

William (Jun 16, 2001 08:39 p.m.):

Maestro (Jun 16, 2001 08:34 p.m.):
I sitll like the Fop 38 or Swiftech are better . The Glaciator is still too new to jump all over it.

Maestro

The 462 is still the king of hsf, but the Glaciator is better than the Fop, WBK, or ThermoEngine.

William, with all due respect, no one has proven to me that the glaciator is a better HSF than the ThermoEngine in fact all the posts I have been reading the results seen in real world conditions have been quite the contrary. I am in no rush to go out and buy a Glaciator over the ThermoEngine I already am getting good results with just to be disapointed with higher load temps. Myself, the only upgrade I could see that would be feasible would be to go to the Swiftech MC462-A if I was looking for a better/more expensive high end cooler.

 

[Thelemac]

Randy (Jun 17, 2001 01:19 p.m.):

William (Jun 16, 2001 08:39 p.m.):

Maestro (Jun 16, 2001 08:34 p.m.):
I sitll like the Fop 38 or Swiftech are better . The Glaciator is still too new to jump all over it.

Maestro

The 462 is still the king of hsf, but the Glaciator is better than the Fop, WBK, or ThermoEngine.

William, with all due respect, no one has proven to me that the glaciator is a better HSF than the ThermoEngine in fact all the posts I have been reading the results seen in real world conditions have been quite the contrary. I am in no rush to go out and buy a Glaciator over the ThermoEngine I already am getting good results with just to be disapointed with higher load temps. Myself, the only upgrade I could see that would be feasible would be to go to the Swiftech MC462-A if I was looking for a better/more expensive high end cooler.

You want a good comparison? Here ya go:

Thermoengine
Glaciator

If you're going by articles using the insocket thermistor as thier temperature monitoring device, here is an article for you.

 

[deathdealer]

Thelemac (Jun 17, 2001 12:18 a.m.):
How are you getting your ambient temp? Are you putting a thermister close to your hsf? If not, then it probably not accurate enough to give you a good idea.

HOW you get your temps is probably the most important thing if you're going to be comparing things. For making sure your processor isn't going to be dying anytime soon, MBM works just fine. For comparisons, forget it. You can't possibly hope to get anything accurate enough to do a comparison between hsfs.

It was never my intention to be doing comparisons of HSF's, I leave that to those better qualified and I rely on that information as a guide when purchasing items. I was trying to figure out why my results were so poor with the Glaciator. BTW I reran a test, this time I put a 20' box fan next to my open case going full speed, that managed to stabilize my "ambient temp" (motherboard) at 24' C throughout the 15 min test run. This is a 1 C difference from the same ambient as the review. Results of that test had the CPU at 46' C nowhere near the 39' C as in the review. It should be noted in all fairness that fan speed from my bios reads the Glaciator's fan speed at only 5500, 200 RPM's less than the 5700 RPM rated speed. Bad fan?
My question is this: the review of the Glaciator was done with a MB identical to mine and a watt load the same as I have and more or less the same air temp yet there's a 7' C (12.6' F) difference between the two MBM temps (mine & the reviews). MBM temps from the various boards used in the review tests were listed in the review so variances in MBM reporting on the 2 Iwill KK266's (the reviews & mine) should not understandably be so far off. You asked is my thermister near the sink, if it was away from the sink I would get lower readings, no, and consequently would have never posted here. Am I wrong to have expected to have had results similar to those posted in the review? I expected to find a C/W of around 18-20 as stated in the review, even with ridiculously exaggerated air flow (a 20" box fan) the best I could achieve was a C/W of 26. Unfortunately now all it seems I have is an expensive copper paperweight since its performance is much less than my "old HSF.