Thermalright XP120 cooler: The new King

Lliam · Post by **Lliam** » Mon Sep 13, 2004 2:56 pm

Good to see Ralf slipping his (?massive) record collection into the review.

Post by **Ralf Hutter** » Mon Sep 13, 2004 2:59 pm

icancam wrote:Now that's service! Thanks for being so responsive, so quickly, Ralf.

Hey, no prob. We only run a full-service operation here at SPCR. This ain't no fly-by-night operation like TH or AT.

acaurora · Post by **acaurora** » Mon Sep 13, 2004 3:01 pm

TH? AT? THAT? lol... *doesn't know what TH/AT are*.

Post by **Ralf Hutter** » Mon Sep 13, 2004 3:02 pm

acaurora wrote:TH? AT? THAT? lol... *doesn't know what TH/AT are*.

Tom's Hardware

AnandTech

acaurora · Post by **acaurora** » Mon Sep 13, 2004 3:05 pm

*speaks with an English accent*

OH... right right... quite right... righty-o.

Post by **MikeC** » Mon Sep 13, 2004 3:31 pm

BTW, Ralf's finding have now been integrated into the article, along with his vertical photo. (p.2 & p.4)

icancam · Post by **icancam** » Mon Sep 13, 2004 3:43 pm

Efficiency, thy name is SPCR!

RaNDoMMAI · Post by **RaNDoMMAI** » Mon Sep 13, 2004 4:03 pm

Ralf Hutter wrote:

This is the globe fan i have.

did you use a normal zalman fanmate to undervolt these? mine did not start under 6V, maybe i just have a bad pair?

~RaNDoM

msm_zgok · Post by **msm_zgok** » Mon Sep 13, 2004 4:15 pm

Awesome review. Now, I must get a P4/A64 and see the capabilities of the XP-120 myself.

alleycat · Post by **alleycat** » Mon Sep 13, 2004 4:44 pm

Thanks for the review, Ralf. As a point of reference, would it be possible to see what temps you get running the HS totally passive? I am guessing that the Globe @5V is moving the least amount of air out of all the fans tested, and perhaps explains why it is more efficient in suck mode - it is assisting rather than opposing the natural convection current. If that's the case, then this HS is creating a fair amount of 'breeze' by itself, and could be useful in some fanless applications.

mg1394 · Post by **mg1394** » Mon Sep 13, 2004 5:11 pm

Great review as everyone has pointed out.

I believe there is a new parameter that needs to be included in the reviews (and spec's) of some of these large (physical size) heatsinks.

While there are always flags and notes regarding what motherboards they are compatible with (I guess they share space with parts on non-compatible boards), there is no mention on whether they exceed the ATX footprint - namely the edge of the board near the CPU.

I have run into a couple of cases that I could not use a Zalman 7000 on due to it colliding with a nearby power supply. From the pictures of this current favorite, it looks like a large overhang beyone the notherboard exists. How about a measurement of this if it exists for a particular model? Thanks.

Edward Ng · Post by **Edward Ng** » Mon Sep 13, 2004 5:13 pm

There is a distinct possibility that this is why Thermalright did not list Ralf's board on the list of boards that are compatible; notice, in his images, that the XP-120 goes past the edge of his board, yet otherwise fits just fine with no need to bend capacitors etc...

dabl · Post by **dabl** » Mon Sep 13, 2004 5:22 pm

RaNDoMMAI wrote:
This is the globe fan i have.

did you use a normal zalman fanmate to undervolt these? mine did not start under 6V, maybe i just have a bad pair?

~RaNDoM

I'm interested as well in exactly how the Globe fan speed was controlled for the review.

--Don

icancam · Post by **icancam** » Mon Sep 13, 2004 5:35 pm

alleycat wrote:As a point of reference, would it be possible to see what temps you get running the HS totally passive? I am guessing that the Globe @5V is moving the least amount of air out of all the fans tested, and perhaps explains why it is more efficient in suck mode - it is assisting rather than opposing the natural convection current. If that's the case, then this HS is creating a fair amount of 'breeze' by itself, and could be useful in some fanless applications.

Good idea!

I'd like to see that taken one step further and eventually see a comparison between the XP-120 and the Scythe NCU-2000 in fanless mode. Would the convection currents created by the particular orientation of each heatsink's fins mean that the Thermalright would work best as in Ralf's original (horizontal) test set up whereas the Scythe would be expected to perform best in a vertical position (as designed)?

Personally, I've never liked the idea of mounting a fan directly onto a heatsink because of vibration and the coverage gap caused by the fan's central hub. A fan mounted a minimum of 3 cm away overcomes the fan hub gap and, naturally, is not physically attached to the heatsink-processor-motherboard.

sorenbro · Post by **sorenbro** » Mon Sep 13, 2004 10:56 pm

I could imagine that the Globe has fairly similar characteristics to the Papst, as it runs at approx. the same rpm at those voltages. Also the sound described in the review could fit the Papst.

meglamaniac · Post by **meglamaniac** » Tue Sep 14, 2004 1:56 am

Very nice review, been waiting for SPCR to write that one up and I look forward to an XP-90 review too of course

Where are you guys buying these Globe fans?
I can't seem to find 'em, although this could be cos I'm a dirty foreigner (UK resident)

JvD · Post by **JvD** » Tue Sep 14, 2004 4:11 am

Really nice review. Only ONE thing bothering me. IS THIS THING TOOOO BIG FOR ME? You can clearly see that the XP-120 reaches out over the edge on the motherboard. I got a IS7 and in a Antec SOnata and I dont know if I can fit this heatsink. The plastic retention bracket (or what is callt, the one that holds the heatsink) is rotated the oter way, so the short side face the PSU.

Maybe you can add a couple oif extra cm up to the PSU. SO WILL IT FIT? Or do I have to pick the XP-90 instead?

Q to the reviewer (or anybody else having thios HS) how much bigger is it that this plastic retention bracket?

EDIT. From the picture above (and in the review) it seems as if the retention bracket has the same orientation. If the board is the Intel Rock Lake is shuld look almost EXACTLY the same. Se here:

How much does it reach outside the edge of the MB?

Post by **Ralf Hutter** » Tue Sep 14, 2004 4:46 am

alleycat wrote:Thanks for the review, Ralf. As a point of reference, would it be possible to see what temps you get running the HS totally passive?

I'd have to be a bit tipped over to try this. Sober I'm too chickensh!t to try passively cooling my out-of-production 2.4C.

My concern is based on extensive testing that I've done with much, much cooler running undervolted PIII Tualatins on TR SLK800 heatsinks. I don't remember the exact numbers, but there was a large difference in the idle and load temps (on the order of 20Â°C, IIRC) between not having any fan, and running a 5V L1A. It seems like just the slightest breath of air cools much better than no air at all + praying for good convective cooling.

(Send me a sixer of Guinness and I may try running the XP-120 fanless, but otherwise my common sense is telling me not to.

)

Post by **Ralf Hutter** » Tue Sep 14, 2004 4:52 am

Edward Ng wrote:There is a distinct possibility that this is why Thermalright did not list Ralf's board on the list of boards that are compatible; notice, in his images, that the XP-120 goes past the edge of his board, yet otherwise fits just fine with no need to bend capacitors etc...

My guess is, that along with TR's Compatibility Lists for their other heatsinks, the XP-120 data is not complete. I'm guessing that the chart listings come from data sent to them by users, and is not something they've generated themselves by obtaining one each of every motherboard made, and testing the fit. I suspect that as the XP-120 goes into wider distribution, TR's chart will have other mobos added to it, via customer feedback. (but then again, I could be wrong.

)

Spod · Post by **Spod** » Tue Sep 14, 2004 5:11 am

Has anyone seen if this will fit on the MSI K8N Neo2 Platinum? With an NV/ATI silencer*?

The CPU socket on this board is right next to the GPU, with perhaps 2-3 cm clearance. See this picture. It's a popular board, so I'm sure I'm not the only one interested in this.

*I've seen cards with the new silencers that have a metal plate on the back of the card to cool the RAM on the back. I'm not sure how far it sticks out, maybe 5 to 8mm?

I suppose what would be most useful to everyone is a measurement of how far the cooler overhangs the processor socket on each side - that would make it easier to guess at whether it would fit in any particular setup.

And thanks, Ralf, for all your efforts. I'm glad you used 120mm fans with this - what's the point in testing a heatsink with a "standard" 80mm fan if half the benefit comes from its ability to use something bigger?

Post by **Ralf Hutter** » Tue Sep 14, 2004 5:20 am

dabl wrote:
RaNDoMMAI wrote:
This is the globe fan i have.

did you use a normal zalman fanmate to undervolt these? mine did not start under 6V, maybe i just have a bad pair?

~RaNDoM
I'm interested as well in exactly how the Globe fan speed was controlled for the review.

--Don

Fan speed was controlled by inputting a known voltage (5 or 7V) from a Zalman Fanmate, or directly from the motherboard fan header (12V). Yes, most of the Globes didn't start at 5V (with the thermal probe in the circuit) but they did start at 5.5-6V. They all start at 5V with the thermal probe shorted, but they run about 5-15% faster.

shadestalker · Post by **shadestalker** » Tue Sep 14, 2004 5:23 am

With regard to the weight of the sink / fan combo and possible stress on the motherboard and mounting hardware:

Is the recommended fan included in the listed product weight?

Would a Zalman fan bracket allow mounting a fan off the HS and still get adequate airflow?

It would be nice to see a bolt-through mounting option for people concerned about that hunk of metal hanging off the board.

Post by **Ralf Hutter** » Tue Sep 14, 2004 5:36 am

Spod wrote:
I suppose what would be most useful to everyone is a measurement of how far the cooler overhangs the processor socket on each side - that would make it easier to guess at whether it would fit in any particular setup.

Probably the most useful method would be to measure how far the installed heatsink overhangs the Intel Retention Bracket itself (the included K8 bracket has the exact same dimensions, 3.000" x 3.500"). The footprint of the bracket is 88.9mm x 76.2mm. The "heatpipe side" of the heatsink overhangs the retention bracket by approx 19.5mm. On the side opposite the heatpipes, it overhangs the bracket by 34.3mm. On each of the other two sides, it overhangs the bracket by 20mm. Using these dimensions, you can extrapolate where the heatsink will end up on your own board.

Post by **Ralf Hutter** » Tue Sep 14, 2004 5:38 am

shadestalker wrote:With regard to the weight of the sink / fan combo and possible stress on the motherboard and mounting hardware:

Is the recommended fan included in the listed product weight?

No.

shadestalker wrote:Would a Zalman fan bracket allow mounting a fan off the HS and still get adequate airflow?

Don't see why not.

shadestalker wrote:It would be nice to see a bolt-through mounting option for people concerned about that hunk of metal hanging off the board.

The XP-120 is really light. It's safely under Intel's max heatsink weight for S478. I wouldn't worry about it's weight unless you're installing one of those aluminum Panaflos, or perhaps an all-steel fan.

bryson · Post by **bryson** » Tue Sep 14, 2004 6:57 am

Given the performance of this heatsink, was there any thought to see what happened with no fan? (or at the very least relying on a case fan remote from the CPU that would provide other airflow benefits).

Calculon · Post by **Calculon** » Tue Sep 14, 2004 8:23 am

Thermalright wrote: Our Heatpipe construction was also redesigned to eliminate any gravity effect and to unleash the capability of the XP-120's performance to operate at 100% in any direction you mount it.

Good heatpipes have an internal wick-structure to transport the fluid back to the hot end using the capillary force. Of course having the hot end at the bottom is always a bit better.

How heatpipes work.

hero · Post by **hero** » Tue Sep 14, 2004 8:27 am

Looks like someone had the cajones. I didn't think fanless would be an option.

DrCR · Post by **DrCR** » Tue Sep 14, 2004 8:32 am

Ralf Hutter wrote:Yes, most of the Globes didn't start at 5V (with the thermal probe in the circuit) but they did start at 5.5-6V. They all start at 5V with the thermal probe shorted, but they run about 5-15% faster.

Oh, so don't cut off the probe (they would be disconnected), but connect the wires together?

Also, where would the Nexus "Real Silent" 120mm fan if in? And which Papst model would, according to sorenbro, be much like the Globe? Oh, and unlike the shorted Globe and most Papsts, the Nexus won't go below ~7V right?

Thanks!

DrCR

__________

Post by **MikeC** » Tue Sep 14, 2004 8:33 am

Ralf Hutter wrote:(Send me a sixer of Guinness and I may try running the XP-120 fanless, but otherwise my common sense is telling me not to. )

Methinks this is wise. None of the Thermalright coolers ever did well without forced airflow. Quite simply, the fins are too close together for natural convection to be effective. Just compare it to the fanless-specific Scythe Heatlane NCU-1000 or 2000. Also, the airflow path through the fins is not omnidirectional (as in the cylindrical HS types, the orig Zalman flower HS or Alphas).

Trip · Post by **Trip** » Tue Sep 14, 2004 8:39 am

Panaflo 12cm weighs 270g for a total of 640g if anyone is wondering.

Thermalright XP120 cooler: The new King

SUPER!