Thermalright Ultima 90 is out :)

[BonzaiDuck]

EEyore --

Ducting the cooler fins or the cooler itself will mostly get you an opportunity to avoid hanging a fan directly on the cooler. I still get slightly better temperatures with the duct for the cooler, but I think the advantage is much more pronounced for the mobo components.

But if ducting the cooler looks easy with your case, then attending to "low-volume/high-airflow" across the mobo components would be part of that design and just as easy. If you duct the cooler, you should then have a surface covering the motherboard, and getting the air underneath ported to an exhaust fan is equally easy.

I need to look at the ShinEtsu paste. The IC Diamond stuff, or even my custom mix of micron diamond powder and JEtArt CK4800, is worth about 2 to 3C degrees in improved cooling -- definitely 2C's-worth.

On the "convex-cylindrical" design of the base, I may have said that -- I'd talked to their tech-support about it -- their "deliberate design." They're just doing things that seem less than optimal and with the changes they've needed to make in mounting mechanisms after releasing the product initially for resale, my diagnosis is that they're not paying enough attention to testing innovations and they're trying too hard to rush products to market. They KNOW they've got an edge with their heatpipe design, the proportion between the mass of the base, the pipes and the fins. But note that all these troubles arise from changes in mounting mechanism and contact with the processor.

San Diego. Riverside. This is like the price of gasoline. It was $1.60 in 2002 or 2003, but to see a drop from 3.40 to $2.60, everybody acts like it was manna from heaven. My family is telling me "Gee, it's only going to be 97F tomorrow."

I dream of snowstorms in So-Cal in July through September, with 6-foot drifts at the lower elevations. Or perhaps we're due for some freak hurricane south of here that spins off piles of precipitation that are enougn to cause a big "Sonoran Bloom." Or one of those swirling cells that sits off the coast of Washington and Oregon, and spills spiral clouds and rain all over So-Cal, Nevada and Utah.

At least -- you're near the ocean . . . .

 

[Eeyore69]

BDuck,

I hadn't thought of eliminating the heatsink fan, which is currently a 120cfm, 120mm fan, and just ducting the intake and exhaust fans directly to the heatsink. This might help since the 120mm heatsink fan overlaps the heatsink on the top side which means some of the air provided by this fan is wasted for cooling. Reducing the ducting from 120mm on the case fans to 90mm at the heatsink would increase the air speed at the heatsink which should help cooling of the heatsink. And there would be one less fan. In addition, the Antec 900 has a side panel mounting postion for a 120mm fan which, of course, I have populated with an Antec Tri-Cool fan that I had hanging around. It is just below the HFS and blows directly on the video card and chipset. But I am wondering what it does to the total airflow; the airflow for the heatsink is direct and good but does the side airflow cause a disruption in the heatsink airflow? Ducting the heatsink would help prevent the disrupting airflow from the side fan from affecting the heatsink airflow. It would be nice to have a CFD program output that shows the airflow within the case with my setup.

I bought the Shin-Etsu because of a couple of reviews that I read. It did very well, is reasonable in price and fairly easy to get. Since I hadn't built a new system in quite a while I was/am still using some old Arctic Silver (not AS5 but the original AS!). I'll try the SE and see if it does anything but it will be hard to tell if it does so. If I was just changing the thermal paste it would be definitive but I am also going to lap the cpu at the same time which changes the dynamics of the thing.

I still don't think there is a major problem with the U90 mounting. It is designed with the shoulder screws/springs so that when the screw bottoms out the spring/s exert a certain pressure. If additional pressure is required/wanted then small washers or spacers can be placed under the springs. Or, if a more drastic measure needs to be taken, then shorter screws without the springs can be used to apply direct pressure on the retaining bracket. Of course this means that the screws may be overtorqued and apply too much pressure which could cause damage. In addition, on my motherboard, there are a row of short caps just below the bracket that hold the spacer that the retaining screws go into. Use too long a screw and the cap could easily be punctured and I am sure the warranty would not cover it. My main worry is that when I lap the processor that enough material will be removed so that the top surface of the processor is below the level of the surrounding cpu retaining bracket. Then the base of the heatsink would contact the cpu retaining bracket and not make contact with the top of the cpu. Then you are SOL!

Overall I am satisfied with the U90 and the cooling it provides. This morning after I sent the last email I stopped Einstein@Home and ran Orthose for ~10-15 minutes. The Orthos program didn't stress the cpu cores any more than does E@H since the temperatures were the same using either program. I figure 51C with an ambient of 100F at 4 x 100% loading is good enough especially considering that I still have to lap the processor and change the thermal paste.

Another interesting thing I read was a comment that BIGGER is always better with heatsinks. I can only say that that is partially right. There are other factors that come into play such as type of cooling (air, water or such) and the use of heatpipes. Give me a large heatsink with no heatpipes versus a smaller heatsink using multiple heat pipes and they will probably duel to a draw. Choosing a heatsink is a personal choice depending on case size, mounting position on the motherboard, overclocking requirements and such. For a slightly smaller heatsink I gave up a bit of cooling power which suits my needs.

You can dream of snow storms all you like; I haven't been to the mountains and snow since I was a young boy! I was born in and currently live in Oceanside (30 miles north of San Diego); but I don't live along the ocean but about 4-5 miles inland. For quite a while I lived in Santa Ana (the OC) but had friends that lived in Riverside because of the high housing prices in OC; and this was 25-30 years ago. Now they are paying in high gas prices. If the temperature gets to 70F or less I begin to get cold. The best temperature is between 90 and 95F as long as the humidity is not too high. I don't care for cloudy or overcast days although I do like the rain; either sun or rain but not in between. Bring on the heat!

Eeyore

 

[BonzaiDuck]

EEyore --

You'll find some articles at OverClockers.com on the thermal resistance and thermal conductivity of different materials. AS5 chose silver particles because silver is second-best, with diamond way out ahead of everything. The earlier articles that dealt with this dismissed diamond because it was too expensive, but this has changed -- somewhat.

A tube of IC Diamond is about $5, and worth one application for each of a CPU and GPU, with enough extra in case you make mistakes. Compare this with AS5, such that you could get ten or so applications for the small $5 tube.

I was able to measure the improvement over AS5 two ways: my own concoction of Penn Scientific micronized diamond powder and JetArt CK4800 (10% diamond), and the IC Diamond (see HeatsinkFactory.com) which has the consistency and percentage loading (> 90%) of my concoction (it's almost as hard to spread than mine -- but slightly more manageable for the oils or base used). The reduction in idle temperature over AS5 was 2C degrees under a 33% over-clock for a processor rated at TDP = 105W. [This could very well push the TDP of the OC'd processor beyond 150W.] Their own tests showed an improvement of 5 to 6F(ahrenheit) degrees, but I can't remember the thermal wattage in the tests, since that would affect the improvement in temperature directly.

I did a quick web-search on ShinEtsu, and so far I only find that it is (likely) a silicone-based TIM. One article shrouded the material in mystery, suggesting a "secret formula." Certainly, there are all sorts of additives, including aluminum oxide and boron hydride in many silicon pastes and greases. I'd be interested if your tube of ShinEtsu reports general ingredients, for what is used in the material.

On the matter of ducting 120mm fan to the narrower 90mm aperture -- yes -- I recommend it. Figure it this way: an Xacto knife, any of various glues (I can recommend best and second-best), and a $5 2'x3' piece of foam-art-board, you mostly will consume a little time in the design and "manufacture." The black foam-board sold at Michael's Arts & Crafts seems to have the highest quality; Target sells panels made by Elmer's for $2, but I've discovered the foam in it is less dense and flakier. Lexan is good for the transparency and "pretty" appearance, but it takes a lot of time and filing to get the pieces to fit. A 12" x 24" panel of Lexan from Home-Depot will cost you $10. The foam-board has noise-deadening properties -- the Lexan doesn't.

I've discovered a hobby-shop brand of glue that is tops with foam-board, and they also make a glue for Lexan. The product is Zapo "foam-safe" for the former, and Poly-Zap (same company) for the latter. But someone at another web-site pointed at another Lexan glue which also fuses and welds the plastic, and it could be less expensive. The foam-safe is $10 per bottle, and the Poly-Zap is $5.

Second-bests are a silicon-adhesive-type glue from Elmers, and a volatile rubber-contact-cement sold by Michael's. Michael's recommended their glue, but it melts the foam and must be applied sparingly. The silicon-adhesive Elmer's could be replaced by any automotive clear silicon adhesive-sealant, and has good adhesion properties, although it has a rubbery-"give" to joints you'd like to see rock-rigid.

You can use both Lexan and foam-board, but fusing the Lexan to the foam-board requires use of the Poly-Zap -- the Zapo doesn't reliably fuse to the Lexan. And the Poly-Zap dissolves foam, so must be used sparingly. If it is to fuse the Lexan to the paper-surface of the foam-board, no problem at all. If you use it on a foam-surface, apply sparingly, or put the glue on the Lexan first and allow it to get tacky. The faster it sets with the foam, the better.

All the various colors and qualities of foam-board can be spray-painted with primer, enamel and clear-enamel. It actually makes the resulting project more rigid, but be careful -- again -- about excess paint on the foam edges -- it will dissolve some of the foam. Thin coats seem to work fine, and appear to seal the foam from any further deterioration from the paint.

Oh yeah. I discovered something else. You can repair or finish imperfections in the foam-board construction with Bondo-Hair fiberglass body plastic, or the Bondo without the fiberglass (easier to work with). But sanding the former takes time, and if you use a file, imperative to keep it flat to the surface. Was still surprised that filing did no damage to the foam board, though.

I need to take a photo of my ducting project in its most recent stage and post it -- even here. I think the last photos were in my "Motherboard Ducting" thread (which see), but don't do justice to the most recent incarnation. I was able to install an 80x15mm Zalman-OP1 fan in the duct extension covering the VGA card, and the assembly works just great.

I came here from Illinois (snow and seasons) at age 10, then left to work in Wash DC at 24. I recall summer days when I was in junior-high-school that reached 107F, but I'm convinced that -- statistically -- it is hotter here now than it was then. I'd rather wear a comfy goose-down parka or Polar-Tec clothing and keep the thermostat tuned to 62F, than walk around in bermudas and floppies and a tee-shirt while sweating a lot. I guess the mid-Atlantic "grew" on me. Or maybe I've got some mental disease, like the homeless person who wears six layers of clothing in summer heat. [Just joking.]

URGENT HEADS-UP -- In another thread, read my exchange with "MrDudeMan" and his remarks on lapping the heatspreader. He knows his S***. I think I will leave mine "as is."

 

[Eeyore69]

BDuck,

I too never found any good info on the Shin-Etsu thermal paste and do believe that it is silicone based. Many of the reviews rated it above the Arctic Silver varieties, and since I needed to replace my old, original Arctic Silver with something more up to date, I decided on the Shin_Etsu.

Now that you mention the foam material for creating ducting I seem to remember reading something by you concerning ducting. The foam material seems like it would be easy to use and, since we have a Michael's close by, I may pick some up and fool with it. It is much easier than using acrylic. I buy cutoff pieces of acrylic from a local plastic supplier for $1 a pound and use IPS Weld-On 4 to make small articles; but I wouldn't want to try it for a ducting project. I don't have the time or patience! If you ever do finish your ducting project and have pictures please post them for all of us to see and get ideas from.

I once worked for a small skin/scuba dive manufacturer in OC. One year the DEMA (Dive Equipment Manufacturers Association) trade show was at McCormick Place in Chicago in February. At night it was 18 below and not much warmer during the day. One week there was enough for me. Never again! California is for me.

I read the thread concerning lapping of the cpu. I don't totally agree with MrDude. For one he says that the extra temperature isn't what destroys the processor but that the electromigration in the semiconductor is the culprit. But it was always my understanding that the increased temperature of a processor causes increased leakage currents. These increased currents causes increased electromigration. The higher the temp, the higher the leakage currents and consequently the higher the destructive electromigration. The higher temps seem to be a direct cause of the early demise of processors. Also, he talks about the convex case preventing the expansion of the internal components of the processor. If so, why wouldn't a perfectly flat case against a perfectly flat heatsink base do the same thing? And if the case was convex and only contacting the heatsink base at one point then I would think that there is the possibility of expansion in the non-contacting areas that wouldn't happen with flat to flat areas in total contact. But that is only my take on the situation. I have lapped 6 or 7 processors with no problem and definite heat reduction. I even did one of the fpga "flip chip" epoxy processors which, at the time, hadn't been done. And I will continue to lap processors and heatsinks. I just want you to understand that I am not totally crazy when it comes to lapping; I am not one of THOSE that removes the cover and exposes the dies themselves. Now that is totally crazy (LOL)!

Again, let us see some pictures of your mods when you have the chance,

Eeyore

 

[VinDSL]

Originally posted by: Eeyore69
BDuck... blah, blah, blah...,

LoL!

You obviously don't get it... took me a while myself!

BDuck?!?!? It should be BDuct...

Get it?!?!?

duct (dukt) pronunciation
n.
1. An often enclosed passage or channel for conveying a substance, especially a liquid or gas.
2. Anatomy. A tubular bodily canal or passage, especially one for carrying a glandular secretion: a tear duct.
3. A tube or pipe for enclosing electrical cables or wires.

This guy (I assume) is obsessed with air ducks -- I mean air ducts.

My gawd, what has led him/her down this path?

I digress...

 

[BonzaiDuck]

Quack!

VD -- you thirry thavage!! I'm not entirely Looney Tunes!! Ducting works.

All OEMs duct their machines, or many, such as Dell do. It's economically feasible, because they can pick both the mobo and the computer case for mass production. Those of us here who build our own tend to forget the possibilities, and mobo manufacturers can't anticipate ducting design options for 100 different cases, nor can case manufacturers provide ducting parts for 100 different motherboards.

The earlier AMD or P4 machines were the basis for this article in 2003, although the thermal power of the components was in our ball-park:


John Cinnamon's 'Cheap As Free' Air cooling

Another article by Bill Leggett

Look at the statistics on temperature improvement. That's how I got started with this.

I can envision using ducts with water-cooling, so that no motherboard components are missed in the equation. A lot of water-cooled systems cover both the VGA card(s) and the processor, but leave the chipsets and Mosfets to the air currents without maximizing air-cooling potential.

I'll get those pictures on my latest progress. The experimental duct-box I mentioned in the "MOTHERBOARD DUCTING" thread deployed a puller fan on the VGA cooler (VD would call it the "ThermalWrong HR-03"). I've replaced it with a smaller pusher fan. The puller (92mm) version kept my 8800 GTS card at 42C both load and idle ("load" measured with game play). The pusher (80x15mm) version idles at 42C and maxes out at between 47 and 51C with 3DMark06 and my TrackMania United racing game. Here -- the VGA temperature is the Everest Ultimate GPU temperature.

This weekend, I'd installed the thermal sensors for my Striker Extreme board and moved the fan plugs to the pinouts provided by the mobo, so with a sensor on the VGA card and another on the Northbridge, I've been able to monitor the temps, control the fan speeds thermally, and get an idea of the NB temps at idle and load. No matter what I do -- 2 x ORTHOS, games, 3DMark06 -- anything -- the chipset temperature measured at the base of the heatsink just stays rock-solid at 40C (idle or load). My guess is that chipset temperatures are really 10 or 15C higher than that, but this isn't any different than the difference between TCase and Core TJunction temperatures.

 

[BonzaiDuck]

[Anyway, VD, you've "crossed the line" with your innuendo. I've been even-tempered with you so far, but you're starting to **** me off. I can see how some other people here reacted the same way.]

 

[Daverino]

Originally posted by: BonzaiDuck
[Anyway, VD, you've "crossed the line" with your innuendo. I've been even-tempered with you so far, but you're starting to **** me off. I can see how some other people here reacted the same way.]

Don't worry BD. I think it's safe so say that someone running a four year old processor with a marginal overclock might not be the best authority in enthusiast cooling. . . Personally I appreciate the work you put into your posts here.

 

[BonzaiDuck]

Davarino! You seem to be relatively recent addition to our group, so welcome aboard.

I appreciate your remark. Thing is, it is becoming almost obsessive-compulsive for me -- a sort of outlet. I should even worry at the moment that my steak is burning on the grille.

 

[VinDSL]

Originally posted by: BonzaiDuck
[Anyway, VD, you've "crossed the line" with your innuendo. I've been even-tempered with you so far, but you're starting to **** me off. I can see how some other people here reacted the same way.]

Bah! I was just in a whimsical mood...

Don't start acting like a ****y old queen or we'll out a dress on you!

Dude, I've been called every name in the book, on this web site - mostly by "lifers" - some of them mods now, you know?

AnandTech Forums isn't for the faint of heart! You know that.

Anyway, I apologize if my jest offended you...

We ALL appreciate your hard work and excellent posts - I'm sure! :thumbsup:

 

[aigomorla]

Originally posted by: BonzaiDuck
[Anyway, VD, you've "crossed the line" with your innuendo. I've been even-tempered with you so far, but you're starting to **** me off. I can see how some other people here reacted the same way.]

ROFL... its like watching 2 siblings fight.

Both are fairly knowledgeable with air products, and how to tweek air products.

I would most definitely come to you guys if i needed a solution with air.


But i think you guys need to jump on water and see what monsters you guys can create.

 

[Daverino]

Originally posted by: BonzaiDuck
Davarino! You seem to be relatively recent addition to our group, so welcome aboard.

I appreciate your remark. Thing is, it is becoming almost obsessive-compulsive for me -- a sort of outlet. I should even worry at the moment that my steak is burning on the grille.

No, I've actually been here for a while. I mostly stick to P-180 threads because I own one and can help with questions. Well that and telling VD not to threadcrap all the ThermalRight threads with messianic Tuniq bullshit.

Personally, I too am disappointed with where TR is going, as I own a SI-120 and it works well for me. I was planning a new build sometime after the turn of the year and would stick with TR if I was convinced that everything was kosher on their end. When the news of base warping on the Ultra-120 came out, I was dubious at first because TR has such a great reputation for QA. So when they said it was by design, I took it for granted that they knew what they were doing. Now I'm not so certain, especially with issues on the Ultima-90 as well. Now with the cost and fuss of putting in a TR cooler, I'd consider water-cooling as an alternative.

And so it goes in the boutique world. It took a year or so for AMD's fan base to dwindle, but for something as esoteric as high-end cooling only a few bad pieces of press can be a killer. Hopefully TR can turn it around.

 

[BonzaiDuck]

Take a look at EEyore's post on another thread with his Ultima 90 experience. Also, we've had a very interesting discussion on "Is this lapping job adequate?" -- dealing with TR's decision to make the convex base.

The argument there was that there is an Intel spec for pressure on the heatspreader, that the cores are mounted with a substance that is supposed to inhibit their thermal expansion, and that the pressure spec is also supposed to inhibit the expansion.

So lapping the base might compromise the spec for pressure on the heatspreader. I just don't think that less than a millimeter (what SVC told me they remove from the base of the U-120-Extreme) -- will make all that much difference. I suggested you can shim the springs with washers -- either nylon or metal -- and metal would not be a risk, since the shim would go between the scissors bracket and the springs, anyway.

Again, I can interpret TR's behavior (as a company) to growing pains and staff turnover, but then, MrDudeMan ("Is this lapping job adequate?") may also be correct. Then again, there are several heatsinks by other manufacturers with flat bases and spec'd to work with a C2Q.

TR is feeling exuberance over their ascendance or dominance in a very competitive market. This is a specialty market: a good majority of computer users will only buy OEM machines and never open the box, so people such as us aren't leaving any of those companies much wiggle-room to stay in business or prosper. Only so many companies will be able to serve our sparse enthusiast population. Some will fail and bail and work on some other type of product. So the exuberance is causing them to spin off products for release faster than they can do thorough testing and work out the bugs, in addition to the intense competition.

Also, with such a limited market, and the cost of design and manufacture, you wonder how they were able to release the U-120-Extreme, the IFX-14 and the Ultima 90 -- all in a four-month time-frame. They are furiously trying to grab more market share by attracting more customers, but it's costing them more for the "product differentiation." Since some of their products have such a good track record sustained over some three years, you'd hate to see them start making bad decisions or spending too much to sustain either "breakeven" or profitability, and then fall by the wayside for the very specialized, limited market of customers.

But if they do, somebody else will replace them. That's a dead certainty, and just a matter of time -- as long as people like us are buying these sorts of products.

[Man, I'm so glad to be retired and out of the rat-race, just thinking about "theory of the firm" and perfect competition. What a reeee-leeeF!!]

 

[Eeyore69]

Even allowing that the lapping is going to lessen whatever the pressure might be with your heatsink and processor combination, this heatsink is not made to NASA specs; thank goodness for that or heatsinks would cost $1,000,000 each. And, I am sure, that Intel's spec is not a single number but has to be a range for the pressure requirement. Your combo could be on the high end or the low end but it would be an unknown value. If you lap the cpu and heatsink you can add shims of some sort but how would you know what amount of pressure increase the addition of the shims would cause or even if they are needed?

That is why you have to find a manufacturer that you can trust until they cause you not to trust their product. I have had three different TR heatsinks on various processors and have had good luck with them even though some of the things they do don't agree with my thoughts. I didn't like the convex/cylindrical shaping of the base of the Ultima-90 so I lapped it. I have yet to lap the processor (Q6600 G0) since I am still burning it in (I installed it ~2 weeks ago) to see if it will fail; in another couple of weeks I will lap the processor also. Even running with only the processor having been lapped I seem to be getting decent temperatures from the 4 cores. When installed it I overclocked it to 3.2G and began running Einstein@Home 24/7 and 100% loading on all cores. At 70F ambient the hottest core runs at 41C at 100% loading. At 70C ambient the hottest core idles at 29C. Even a few days ago, when the ambient in SoCal was ~100F, the hottest core at 100% loading was 51C. And I only had the 120mm 120cfm fan running at ~7-8v which is just over 1/2 speed. So I continue to think that Thermalright and their U90 are ok! Although I am somewhat worried about the temps since many other people have much higher temp readings on their Q6600s. I use CoreTemp to get the readings as do many others but the difference in comparative temps still worries me.

I always read and listen to what people are saying about this or that but I usually don't let someone elses opinion overwhelm mine unless what they are saying is really powerful.
So when you listen to what people have to say take everything with a grain of salt. There are always THOSE people, you know who you are (not you BDuck), that speak from ignorance, self-importance or just plain spite.

By the way BDuck, what is your reasoning for covering/ducting the whole motherboard as I saw you do in another thread? I would have thought that ducting certain components (heatsink, chipsets, memory, video cards and such) would be sufficient?

Eeyore

 

[BonzaiDuck]

If you look at Cinnamon's OverClockers article, you can see how he boxed in all the PCI cards and built a fan into the box.

I don't think that's necessary -- or not for me. I bunched up the deployment of all the cards so they're nearest the VGA. [I've yet to take some photos of the duct-box, and eager to show]. As you would find from that "Motherboard Ducting" thread, the box over the CPU cooler is rectangular. But you wouldn't need to build it the way I did, if the duct between the 90mm Ultima and 120mm exhaust fan is fairly well sealed, and only convenience would dictate a pusher fan. If the fan is mounted on the cooler, then you wouldn't need to duct on that side. If, as I did, you have a fan pulling air from behind the hard disks, then extend the duct to it but let it be leaky.

The bottom of the CPU duct serves to cover the entire mobo from the upper edge down to the VGA and chipset sink, and the VGA duct extension runs from the top of the CPU duct-box over the top of the VGA and down the other side. In that latter panel, I've installed a fan whose hub just sits on top of the heatsink-base attached to the GPU. It sucks enough air off the 3Ware RAID controller that I don't need to worry about that. The eight-port and 12-port versions of the controller card have fans, but my 4-port card just has a heatsink, and the VGA fan is pulling plenty of air off the backside of that card.

That leaves an AverMedia PCI-E tuner card. If it were an older Hauppauge card, I'd consider covering it, but these new M780's have really tiny tuners and don't heat up. All that's left after that is a sound card, and that's hanging over a bottom intake fan.

Since the southbridge is obstructed by some of the expansion cards, I'm not going to run a "tunnel" over it, and it's connected by a heatpipe to the NB heatsink. The NB heatsink is getting a lot of air drawn across it moving down the CPU-side of the VGA card.

I'm really pleased with this setup. The chipset temperature is rock-solid at 40C with ambient between 75 and 80F. Right now, the ambient is low (NOW I can wear my Polar-Tec!) -- and the thermally controlled fan in the VGA duct has dropped from 2,500 rpm to about 1,400. The chipset thermal sensor has also dropped the exhaust fan pulling air from above and below the motherboard from 2,000 to about 1,000 rpm !!! So the chipset temp stays at about 38C. I'm sure I could bring it down more, but that's fine, and the 80mm fan is under no strain, either. Nor are my ears, but that wouldn't even be the case for the Zalman OP-1, even if it were cranking up to 2,900 rpm.

This sucker is REALLY, REALLY QUIET -- for seven fans!! I really wanted to reduce the number of fans and cover the holes with Lexan plates I've already cut and drilled, but the 140mm jobs are only spinning at 1,400 rpm -- noiseless. The Striker Ext. mobo spec is for a maximum of 7 Amps @ 12V for the cumulative fan plugs. I'm using seven of the eight plugs on the board, and the maximum current draw is (0.52A x 2) + (0.32A x 4) + 0.20A = 2.52A, for a total power consumption of maybe 30W. With the exhaust fans spinning at below half speed because of our "cool-air reprieve" tonight, it's probably less than 2A or 24W. That's like having two 1A fans -- one for intake and one for exhaust -- running at full speed.

As you may have seen, you can do this sort of thing in different ways, like cutting slots in foam or Lexan rectangle which slides under the lowest fin on the CPU cooler. Given the case, the cooler, the motherboard, the VGA and other considerations, it may be wise to pop open a six-pack, enjoy the night air, and think about it first in terms of airflow, ease of assembly (and ease of removal!), and hopefully you won't need to experiment or modify too much.

I also thought that this task was easier with the coolers like the SI-120 with the fan blowing down on the mobo. And I'm also thinking it might be feasible to share the two exhaust fans by opening up a channel between them -- between the two duct components in my design -- between the fans and the CPU cooler -- but then you foul up the thermal control you have by keeping them separate. IF they weren't thermally controlled, I'd think it would pull a bit more air from the CPU cooler without compromising the draw from the motherboard too much. But I think I'll leave that alone.

 

[BonzaiDuck]

The other post was already too long, but I want to elaborate, since you're interested in pursuing this. The ram modules can serve as a part of the duct, in that they block off enough of the airflow and increase the speed of it, and they're also cooled considerably by being an obstruction -- my guess is you'd just want maybe 3/8" clearance between the mobo panel and the top of the RAM modules. This will also keep the air flowing up and over the VGA card, down onto the chipset, past the base of the CPU cooler, and past the mosfets -- if it's being drawn from a ducted fan just above the upper edge of the mobo.

But all this is tempered by where those exhaust fans are placed, or how many of them you have. If you only have one, then there is an option to duct the air off the CPU cooler and force it up into the PSU (sealed fairly well, to keep warm air from getting back into the case.) So the rear exhaust fan could then draw air from below a mobo duct-panel and off the mobo.

I just don't like pushing warmer air through the PSU, but these PSUs today are much more efficient, and if the air is moving fast enough, even the warmer air will do more to cool the PSU. I'm assuming the PSU has a 120mm fan pointed down at the Ultima and drawing air out of the case, but these things often have vents on the case-center side where the wires are connected, and I can imagine building an entire duct-box around a PSU so air only goes one way through the PSU to avoid recirculation.

There are other possibilities, even noisier ones, where you're pulling air off the CPU cooler with a right-angle duct and pushing it out the side-panel, using a single exhaust fan to pull air off the mobo components. you'll just have to figure it out, given your case, number of exhaust fans, etc.

Here, I'm speculating, but I think it's in the ball-park and even center-field: The TR coolers are so efficient that ducting the other motherboard components like chipset and mosfets is a slightly higher priority. I think the advantage is better that way. But if you have an exhaust fan to do it, having a 120mm fan pull air from a 90mm heatsink is really going to keep the temperatures at a minimum -- as long as the duct between them is fairly snug and not too leaky.

There's got to be some way to do all this with one exhaust fan, and you could also have a right angle duct on the intake side of the tower CPU cooler which turns down on the mobo components and draws air off of them. It's just another complication for the design and building of it. Patience with foam-board, keep the Xacto knife sharp, make straight cuts, and try to fold a piece where there's an opportunity instead of using two pieces with glue.

 

[Eeyore69]

In the Antec 900 the PSU is bottom mounted like in my Antec P-180 case. The front 120mm intake fan is mounted at the top and blows air directly at the 120mm fan on the U90 and the backside of the heatsink is directly in line with the 120mm exhaust fan. If you look through the front fan you can see the heatsink fan, heatsink and the rear exhaust fan; all in a direct line. This is why I thought I could build two ducts. One would take the intake air and direct it to the U90 and another would then connect the heatsink to the exhaust fan at the rear. Since everything is in line the ducts should be straight, square cross section but tapered for the transitions from 120mm to 90mm and back to 120mm. This would be a push-pull setup with the two case fans and eliminate the requirement for the U90 fan.

I also have a 120mm side panel fan directly above the 2 chipset heatsinks. Again the duct could be straight with a 120mm cross section at the fan and maybe a rectangular cross section to distribute the air on both heatsinks.

There is also another 120mm intake fan on the front panel at the bottom along with the 200mm exhaust fan on the top rear of the case.; and, of course, the 120mm thermally controlled fan in the PSU. These fans should take care of clearing the case of the rest of the warm air generated by the rest of the components. Even the video card since it is only an nVidia 8400GS with a tiny fan. It doesn't produce much heat.

Now I am anxious to pick up some foam. I am going out tomorrow and will take a look and see what I can find. By the way, does the foam come in different thicknesses? I want some thin stuff so that the finished ducts will be as flexible as possible. Hopefully I can make them fit without needing fasteners to hold them; hopefully I can bend them and wedge them into position.

I finally found an Intel data sheet on the Q6600. I don't know how complete it is since it is open to public view. But the data sheet shows a flatness spec for the top of the processor case but nowhere is it mentioned about putting a crown or concave surface on the top. From the information on the data sheet the top is supposed to be flat within .05mm. If you ever see an Intel data sheet with mechanical specs on the Q6600 that say different let me know.

I had better quit now. I don't want to give VD the opportunity to give me a new moniker; EeyoreDuct maybe.

Eeyore

 

[VinDSL]

Would someone please wake me up when TR Ultima 90s are $12.00 AR @ TigerDirect?

Thanks!

 

[BonzaiDuck]

I think the foam board only comes in half-centimeter thicknesses, or 1/4" thick. It wouldn't be flexible like vinyl. You would strip of a 1/16"-wide strip of the outer paper layer to facilitate straight bends.

Well -- it will be interesting to see what you come up with. It's a small cooler, and I'd suggest you try and be equally mindful to draw cool air through narrow apertures over the mobo components and get that exhausted quickly as well.

 

[BonzaiDuck]

Anyway, VinDSL, you already got your Tuniq, and it at least matches the performance of the Ultima90 . . . .

 

[BonzaiDuck]

I promised EEYore to post some photos of my revised, proto-type duct boxes.

Anyone who had looked at the photos in the "Motherboard Ducting" thread would've seen the very first pass at this. I replaced the upper [trapezoidal] duct-box with a frame of black foam-art-board and one panel cut from Lexan. From the proper angle, you can't really see it very well for the transparent panel.
[I took the photos in a hurry today, there wasn't enough light in the room, so at least the first one is slightly blurred -- so sorry.]

"Chrome Lightning" with side-panel removed (photo slightly blurred)

Again, the trapezoid-shaped duct is a hybrid of foam-board and Lexan, so you can see how both materials can be used optionally for the construction.

Upper duct almost invisible
[two photos showing upper box]

All of the foam-board surfaces were primered, painted silver, and then coated with clear enamel. And there have been minor revisions and modifications so that the ducts fit properly in the case.

The lower duct contains a fan which sits right on top of the heatsink covering the VGA GPU. The fan cabling has a connection that can be made after the duct is in place:

Fan plug connection

Ducts removed and placed on the floor

Again, the lack of light in this room really obscures the trapezoidal box, already hard to see because of the clear panel.

This huge old Compaq ProLiant server case had a cross-brace [photo shows it next to a red pocket-knife on the floor] which bolts into the case from front to rear. Because of case-mods using U-channel aluminum and modder's mesh with four quarter-inch-thick slabs of aluminum as supports for the double-wheeled casters, the brace is an insignificant factor now in the ProLiant's structural integrity -- I can take it or leave it. But the lower duct was modified to make a perfect, snug fit with the cross-brace, which holds it securely so that it cannot move around.

The upper, trapezoidal duct-box, better designed than the one made with white foam-board, has a frame which also fits snugly over the upper exhaust fan, so there isn't any need for velcro, threaded screw-taps, snap-in nylon stand-offs or anything else to hold these ducts in place.

Trapezoid

Again, sorry about the blurred image -- here's another shot:

Trapezoid from different angle

I put strips of PaxMate rubber-foam on the edges to assure a good seal.

VGA Fan

The VGA duct extension contains an 80x15mm Zalman OP-1 fan, as I mentioned. I'd apparently misplaced the Zalman fan guard, and only found it after commandeering an old ThermalTake fan guard to use in its place. I'm more concerned about wires getting caught in the fans than anything, but the duct comes close enough to my RAID controller that I put a strip of PaxMate over the lower edge of the metal fan-guard.

Keep in mind that several parts of this project are going forward, and the foam-board -- at least initially -- was intended to just be used to proto-type the assembly. Someone else might construct a neater, prettier duct, or forge ahead more swiftly in putting one together from Lexan panels. But my priority here was to get that fan on the VGA GPU to see how it brought the temperatures back down, and the paint allows me to take my time while fiddling at leisure with the Lexan panels.

But since the case-side panel window is mirrored with Mylar window-film, you can't seem much of the interior without turning on the lights, and I'm moving toward a more subtle approach to "bling-bling" with this. So I might just leave the ducting alone for a long, long time -- maybe until I'm ready to dump this project and build another.

Summarizing again the saga of GPU temperatures, the stock 8800 GTS cooler showed 70+C degrees after 20 minutes of gaming at about 75F room ambient. Replacing the stock cooler with the TR HR-03-Plus without ducting, and with a fan blowing on the cooler, the temperature regime was more like 47C idle, 55C load, if I remember correctly. Using the first duct revision and a 92mm fan integrated into the duct design, we had both idle and load temperatures shown at between 42C and 43C, but the duct and orientation of the CPU cooler was not optimal.

When I built the duct to cover the VGA card, it showed idle temperatures around 43C and load values between 51C without integrating a fan. But with the pusher fan built into the duct, the idle-to-load spread is "second-best" at 44C - 47C. Keep in mind I've replace a bulkier 92x25mm puller fan with an 80x15mm pusher.

Even so, putting that fan in the duct keeps my RAID controller very cool, and this seems to be the best overall tradeoff to optimize performance over the CPU and VGA coolers while keeping the Northbridge chipset at a rock-solid 40C -- at either idle or load stress. And since the NB and SB are connected together through a common heatpipe, this also cools the SB indirectly in addition to the NB, Mosfets and other mobo components. In fact, the air pulled across the motherboard is also whizzing across the CPU (TR-Ultra-120-Extreme) heatsink-base.

It couldn't be much better for air-cooling.

 

[VinDSL]

Originally posted by: BonzaiDuck
Anyway, VinDSL, you already got your Tuniq, and it at least matches the performance of the Ultima90 . . . .

LoL! Save yourself some money...

Here are deals on last month's world's best cooler:

Thermalright Ultra-90-775 $22.95 (save 49%)

Thermalright Ultra-90-K8 $15.95 (save 65%)

Or, just wait! This will be the Ultima 90 in 6 months...

 

[BonzaiDuck]

Like I said earlier -- there are only so many enthusiasts in the market, the companies compete to survive in that market, and if there are too many companies catering to the enthusiast-consumers, some of them won't be able to meet there production costs, and will therefore fall by the wayside.

Any given company can be a "multi-product" company. ThermalRight is not one of those, as far as I can tell -- Sunbeam is. So some companies have more of an incentive to innovate and differentiate their products. Sunbeam sells fan controllers and other stuff; TR seems to be solely in the heatpipe-cooler business.

And like "High Noon," "The Gunfighter," and other stories -- you can try and remain "King of the Mountain" or "Fastest Gun in the West," but there's always going to be somebody who thinks, tries, or succeeds in being bigger, better and faster.

 

[VinDSL]

Originally posted by: BonzaiDuck
And like "High Noon," "The Gunfighter," and other stories -- you can try and remain "King of the Mountain" or "Fastest Gun in the West," but there's always going to be somebody who thinks, tries, or succeeds in being bigger, better and faster.

LoL!

How about the "Fastest Gun in Portugal"?

Please take note of which CPU cooler he used...

Anyone beat him with a TR Ultima 90 yet?

 

[BonzaiDuck]

My best guess is that with the same VCore, mobo and other test-bench choices -- AND with the same room-ambient, which he doesn't report, he should be able to get nearly the same over-clock with the Ultima 90. He just doesn't have an Ultima 90 to test, and most people wouldn't push their VCore above 1.45V -- let alone 1.6V. So there aren't any other "forum bench-tests" that would show it, or if there were, they're hard to find.

Pedro pretty much says it himself: as in "Kids, don't try these voltages at home 24/7."

The BIOS temps give it away, showing an idle value at something like 52C degrees.

That's the TCase value -- at idle. My experience with the BIOS temperatures are that they may vary from CoreTemp and other Windows utilities, but not that much. They're using the sensors that the manufacturer installed in the board, and the manufacturer chose the BIOS.

Since TCase is likely to be 10 to 15C below the highest core temperature, then the core values would be well into the 60's Celsius at idle. If they're well into the 60's at idle, then at load you'd expect them to be well toward the throttling limit, and at a point where running them 24/7 (as he says) would shorten the processor life to near-nothing.

Given the Anandtech tests that included Tuniq, and as I've said before, the thermal resistance of the cooler even with a 2,500rpm fan-draft on it, would have to be somewhere between 0.095 and 0.11 C/W. So it would be impossible that the highest load temperature at a VCore exceeding 1.6V would be anywhere under 70C and it would probably be closer to 80.

The G0 is -- so far -- the most over-clockable of the Q6600 processors, but it really doesn't prove anything that to get it to 4 Ghz, one has to push the VCore to 1.6V.

The proof of the pudding would be how far he can overclock it by holding the VCore within 5 to 10% of the 1.35V retail-box maximum spec.

What I don't understand, though, is "What is VinDSL trying to prove?" Is he an "agent" for Sunbeam? I'd love to have a Sunbeam Tuniq -- if it would outmatch the Ultra-120-Extreme in a COMPARISON review, but Pedro doesn't give a comparison review. I'd love to have a Tuniq if it didn't have jagged edges on the fins.