SPCR

Hi Mike,

wasn't this heatsink supposed to be drawn back by the manufacturer, or is this the revised version? Or have I just been dreaming / was it a different product?

Anyway, interesting product and review!

Yup, and then they told us they changed their mind. The testing was done like >2 months ago!! We never wrote it up before because of what they told us late Oct.

Interesting

are you going to have it installed in a p180 rig like in the scythe ninja review?
or is that Ralf's job?

the whole weight thing is scary though, probably because it has no extra heatpipe "braces" to counteract the forces of gravity....

So what's the MRSP? I'm guessing they'll be a tad more expensive than the Ninja, which seems to cost about the same as the now-obsoleted XP-120...

excellent review as usual! seems like a good HS, however im a bit worried about how it kind of "hangs" off the mobo like that. i dont remember seeing the ninja doing that
-btw, i was just looking through the recommended HS section and was wondering why the ninja isnt listed there? isnt the ninja better than the XP120? and yet that is still listed as the best?

MRSP Look it up yourself.

I think it's time to update the recommended heatsink list ... it's been about a year, which is why the Ninja isn't on it yet.

Thanks for the link. For the record, the MRSP is the same as for the XP-120, fifty bucks. Not bad.

Another fine review overall as usual. I have a few nits - that I hope you guys don't think are petty.

1. SPCR often uses the word "impedance" when SPCR writes about resistance to airflow. Impedance has a very specific meaning in electric circuit theory and acoustics. One way to look at impedance is that when you want to include the effect of reactance on current flow you have impedance. I have never seen impedence used to describe airflow. I suggest it is clearer and simpler to say "resistance" because SPCR is not claiming there is anything more than resistance to airflow.

2. SPCR simply says in the beginning that the fins are thicker than they need to be. SPCR simply says if they were thinner they could be further apart. Unless you are a thermal engineer - I don't know how you can simply assert this as if you know more than Thermalright's engineers. I assume that Thermalright could easily have built this same heatsink with thinner material (probably at lower cost). We know Thermalright builds well-designed no-bling products.

3. SPCR says most airflow will be around the heat pipes because of less "impedance." If you look at the side photo - it looks like air would flow much easier throught the middle. The heat pipes look like they would block a significant amount of airflow. There is much more open space in the middle of the high rise.

I have no quarrel with the data nor the conclusions. One of the hallmarks of SPCR vs. other PC enthusiast sites is the quality writing and credible analysis.

hofffam --

Thanks for the praise... to soften the picks...

1) I started using the term "airflow impedance" after I came across it in many fan/cooling engineering docs and websites in early SPCR days. I think the concept of reactance also exists in airflow -- turbulence is a kind of reactance, and impedance includes this. Try a web search for it -- it is a widely used, valid thermal engineering term.

2) No quibble -- I din't know exactly why Devon wrote this but left it in. I assumed it was about thinner material taking up (very) slightly less space. The fins are much thicker than the ones on the Ninja, for example

3) You could be right... but in other heasinks, we've often observed that less air flows through the center of the fins in this kind of design, probably because closer to the sides, the forced air has two exits -- the side and the far side (if you know what I mean).

Although the performance of this HS with a fan attached is almost as good as the Ninja, it misses the point in a key area. As far as I am concerned, the greatest feature of the Ninja's design is that it is non-directional, and is especially suited to fanless operation. This allows great flexibility for quiet PC enthusiasts, and pretty much renders exhaust ducting obsolete.

good review. Informative and thorough as usual.

Makes me feel even better about my ninja when i saw how topheavy this thing was while close to the same cooling efficiency.

Editor's Note: The reality is that the noise reduction with the Nexus 120 fan as you go below ~8V is pretty much inaudible in just about any system, so there's really no reason to run it that slow anyway. Of course, if you have a noisier fan...

I think the original writer meant that you do not need to attach the fan to the heatsink for it to work. While the fan might run at 9V, if it's not directly on the heatsink (and there certainly is more airflow resistance in real closed case than in SPCR test bench), so in effect it equals to less airflow through heatsink.

Thus is seems the Thermalright heatsink might require less optimal case to still provide enough cooling for CPU.

An excellent review. The photograph of the HR-01 sagging severely when the mobo was arranged vertically concerns me. Either the mobo was _seriously_ warped, or the bottom of the HS had come partly loose from the CPU's IHS.

One would think the vertical-mobo position would result in better passive cooling, since heated air can rise far more freely than in the horizontal-mobo position. Despite this, the review reported that the vertical-mobo position resulted in a higher CPU temperature, which also leads me to believe that the HS and IHS were not making firm contact.

Now, if you add a ~149gram fan onto the already-sagging HR-01, what would be the result? I assume the active-cooling portion of the review was done with the mobo safely horizontal?

A word about the photo that seemed to show the HS sagging horizontally off the vertical motherboard:

It wasn't sagging.

The motherboard itself isn't always bolted tightly to the wooden platform; usually there's only 2-3 screws holding it down, it's all that's needed because usually it's sitting flat on the platform, horizontal.

What you're seeing is the board itself pulled a bit away from the wood platform on that side, which makes the HS look as if it was sagging. If you follow the line of the PCB (motherboard) from the top edge downward, you'll see that it is not parallel to the wooden platform; there is a gap at the top and much less gap as you move down. This is probably because there was a holding screw in the middle of the motherboard and maybe one at the other end, but not at the top end. (There is enough tension in HS mounting clips that there's no way it could pull away from the CPU, and the HS is far too rigid to bend.)

We should have anticipated the reaction of readers & put in an explanation near the photo.... must do in the morn.

BTW, there is not ONE single heatsink ever tested on any SPCR test platform (in nearly 4 years) that can be run fanlessly under long term CPU load. Convection alone -- on the test bench -- just can't do it; all of these "passive, fanless" HS depend on the peripheral flow from other fans in a system... unless you are dealing with a very low power CPU -- like a Pentium M, Turion, earlier P3s, etc. -- in a very well ventilated, convection-optimized case/system.

Anyone know where I can get one of these in Canada?

Felger Carbon wrote:Despite this, the review reported that the vertical-mobo position resulted in a higher CPU temperature, which also leads me to believe that the HS and IHS were not making firm contact.

No, there was effectively no difference in performance. The CPU throttled in both the horizontal and vertical positions, it never stabilized enough to judge between the two.

there is not ONE single heatsink ever tested on any SPCR test platform (in nearly 4 years) that can be run fanlessly under long term CPU load.

To be honest, I don't think anyone is surprised by this. A couple of points:

1) Not many people run their CPU at 100% load 24/7.

2) Of course a hot, power-hungry CPU is fundamentally unsuitable for passive heat evacuation. Luckily, there are more and more efficient, low-power chips on the market these days.

3) As you point out, intelligent case and airflow layout can make passive CPU cooling a much more attainable goal.

4) As a result of this, as long as you are willing to construct your PC around the principles of low heat generation and efficient heat evacuation, passive cooling is a much more realistic proposition than that statement makes it sound.

I just noticed this, page 4 features a mis-typed link to AOpen's website: usa.apoen.com instead of usa.aopen.com.

Greetings,

jaganath wrote: 1) Not many people run their CPU at 100% load 24/7.

That may be true, but those of us who run Folding@Home (or SETI, etc.) do run 24/7.

MikeC wrote:
It wasn't sagging.

Mike, you do highlight in this review the cantilever effect, which I don't recall you doing to the same degree with the Ninja.
Although the HR-01 is approx 100g lighter than the Ninja, is more of it's weight distributed out at the fins compared with the Ninja?

Given the similarity in test results, purchase decisions are going to come down to things like price, compatability, ease of installation (attaching the HR-01 looks SO simple) and things like the cantilever effect.

I suppose the simplest way to pose my question is: Which of the two would you rather have attached to your vertically-mounted motherboard?

Interesting heatsink, however unless it can be shown that HR-01 is more effective than Ninja in passive configuration (say in P180 with both fans running exhaust), then Thermalright kind of dropped the ball here. So far, from the limited testing SPCR has done, the performance appears to be neck to neck, however Ninja is much more versatile than HR-01, so it's really a no-brainer on which one to buy.

OK, taking Felger's suggestion elsewhere (the modded Z9500 thread?), I checked the balance of the HR01, the NInja and the Zalman 9500 by hanging each from a string -- a barewire Panaflo fan lead, actually -- and molving the position of the wire till it balanced.

Keep in mind that the center balance point by itself is NOT the only factor in the cantilever effect -- the height, the total weight AND the concentration of weight in the base all have an effect. (The last item is not easy to quantify without taking the HS apart.)

The results:

HR01 -- 16cm tall; 525g; center of gravity at at 7.5cm above bottom of base
Ninja -- 15cm tall; 665g; center of gravity at 6cm above base.
Zalman 9500 -- 12.5cm tall; 530g; center of gravity at 6cm above base.

The Nija is definitely more base-heavy and feels it. The HR01 feels more top-heavy than its center of gravity point would suggest; the fins are definitely heavier, and the base lighter than in the Ninja. But since it's lighter overall, the total cantilever force is probably about the same as the Ninja. Adding a fan will change this somewhat -- depending on where exactly the fan ius mounted.

The Zalman 9500's weight is about evenly distributed top/bottom, but it's much shorter and weighs the same as the HR01, so the total cantilever effect is smallest. Also, the 530g weight is with a fan -- while the other weights are without.

As to which HS I'd rather have in my own system, the answer would depend on availability, price, socket type (the HR01 will not go on a 775) and how often I intend to experiment -- ie, remove & reinstall. In this last matter, the HR01 beats the Ninja hands down. With either of these HS, if I was transporting them any distance at all (mounted in a system), even in a car, I'd make sure the top was supported or the case laid down with the MB horizontal -- and my gut instinct says it'd be more important to do this with the HR01 than with the Ninja.

MikeC wrote: BTW, there is not ONE single heatsink ever tested on any SPCR test platform (in nearly 4 years) that can be run fanlessly under long term CPU load. Convection alone -- on the test bench -- just can't do it; all of these "passive, fanless" HS depend on the peripheral flow from other fans in a system... unless you are dealing with a very low power CPU -- like a Pentium M, Turion, earlier P3s, etc. -- in a very well ventilated, convection-optimized case/system.

3 years back I build a Folding@Home machine, Duron 900Mhz, cheap ALU heatsink with a 80mm fan, the system was installed outside of a case, without any other case fan or airflow available except for the CPU fan. After a few days I checked up on the status of the machine remotely and saw the CPU temperature at 76°C; but the machine was still running; So I ran over to the system to see the CPU fan had stopped, and so the CPU could be cooled.. passive

[quote="MikeC"]OK, taking Felger's suggestion elsewhere (the modded Z9500 thread?), I checked the balance of the HR01, the NInja and the Zalman 9500. [/quote]

Mike C's data:
HR01 -- 16cm tall; 525g; center of gravity at at 7.5cm above bottom of base
Ninja -- 15cm tall; 665g; center of gravity at 6cm above base.
Zalman 9500 -- 12.5cm tall; 530g; center of gravity at 6cm above base.

I just measured the Shogun, less fan but with mounting screw/springs, at 577g. It balances 5.4cm above the base. The NCU-2000 is 478g, 3.3cm. If we define the product of the weight times the CG distance above the base as the Cant, measured in Kg-cm, I come up with this table:

Ninja = 3.99Cant, no fan
HR-01 = 3.94Cant, no fan
9500 = 3.18Cant, with fan
Shogun = 3.12Cant, no fan
NCU2K = 1.58Cant, no fan

Thermalright recommends the 1.5" thick Panaflo M 12cm fan for its SI-120. The CG of that fan would be 4.75" above the HS base, and the fan weight is speced at 270g. So the **fan alone** would have a Cant of 3.76, higher than the entire 9500 _with_ fan! That's why I believe it was imprudent of Thermalright to recommend that fan.

Adding a fan to the Cant is a scalar process; it isn't necessary to actually mount the fan. All you have to know is the fan's actual weight and how high above the base it's CG would be _if_ it were actually mounted. My measured weight of the Sunbeamtech 12cm (my favorite) is 109g, the Thermaltake A2330 13cm [you _vill_ mount this on a Sonic Tower!] is 134g, the Yate Loon 149g. The spec on the EU favorite Papst is 175g.

The fact that the Ninja tops the table in its current form suggests that perhaps all of its weight is not located in its base.

Hmm. I was looking forward to the HR01, but after looking at the mounting, it seems a lot of motherboards will cause the heatsink to be installed in a suboptimal position. I would have thought that it would be best if the tower was facing the rear exhaust fan, but a lot of mobos would mount the heat sink like in the review passing most of the heat towards the psu in tower cases.

Its a shame, it would have made for one nice passive heatsink that could maybe be mounted in a modded Q-Pack. Oh well.

Thanks for the feedback and data guys, top effort

Looks like the deciding factors have now been reduced to price, availability in Oz and what I like the look of...

Cheers

jmke wrote:

MikeC wrote: BTW, there is not ONE single heatsink ever tested on any SPCR test platform (in nearly 4 years) that can be run fanlessly under long term CPU load. Convection alone -- on the test bench -- just can't do it; all of these "passive, fanless" HS depend on the peripheral flow from other fans in a system... unless you are dealing with a very low power CPU -- like a Pentium M, Turion, earlier P3s, etc. -- in a very well ventilated, convection-optimized case/system.

3 years back I build a Folding@Home machine, Duron 900Mhz, cheap ALU heatsink with a 80mm fan, the system was installed outside of a case, without any other case fan or airflow available except for the CPU fan. After a few days I checked up on the status of the machine remotely and saw the CPU temperature at 76°C; but the machine was still running; So I ran over to the system to see the CPU fan had stopped, and so the CPU could be cooled.. passive

Sure... in idle. Scythe's notes on their NCU fanless HS series emphasize that CPU stress testing with them would be considered abuse. I agree that CPU stress testing is not really that relevant for most desktop PC users; rather, it's an attempt to induce a worst-case scenario and create a consistent test methodology for comparing HS performance. If you are a casual user, not a gamer, there are many HS/CPU combinations that can run w/o a fan on the HS... but not if you put long term stress (100% for >5mins) on the CPU. Most users never do this, except maybe in games.

[quote="pawstar"]Hmm. I was looking forward to the HR01, but after looking at the mounting, it seems a lot of motherboards will cause the heatsink to be installed in a suboptimal position. I would have thought that it would be best if the tower was facing the rear exhaust fan, but a lot of mobos would mount the heat sink like in the review passing most of the heat towards the psu in tower cases.[/quote]

Consider plan B. On a mobo that will face the HR-01 toward the PSU, a Sonic Tower will be facing the rear exhaust fan. Further, some EU test sites report that while the ST provides mediocre performance for a huge HS when a fan is mounted _outside_ the two towers, if the fan is sandwiched in between its performance is currently the best available for air cooling for tower coolers. Also, the fan eats a lot less mobo real estate when mounted that way.

So: save $10 over the HR-01, get better cooling performance, and probably have an easier time mounting the fan, all while pointing the HS in the correct (or at least the desired) direction.

For those of you who in the past have turned down the Sonic Tower because it was pointed the wrong way, let me introduce you to the HR-01, which for you will be pointed the right way.

Real estate considerations: the Sonic Tower is 112x112x150mm high. The Ninja is 110x110x150mm high, but the fan _must_ go on the outside. With the ST, the fan can go on the _inside_, reducing the footprint while providing superior cooling. Sometimes plan B is best.

Full disclosure: I ordered a Sonic Tower yesterday.