Thermalright SI-97A for both K7 & K8!

pangit · Post by **pangit** » Thu Oct 28, 2004 4:08 pm

Hey, maybe we're guinea pigs, but if it performs well and helps extend the life of the Socket A platform, who's complaining?

But I agree with you, if they came up with something like the Aerocool HT-101 that performed as well or better, I'd snap it up. Ducting straight out the back or through a 120mm fan PSU makes perfect sense from an airflow perspective.

But Thermalright know what they're doing, we'll have to wait for the review to find out......

Cerb · Post by **Cerb** » Thu Oct 28, 2004 5:08 pm

mathias wrote:
DG wrote:Hmmm, could it be posible to fit a 120mm fan on the SI97, with an adaptor? The only issue would be the weight of the 120mm fan and adaptor....
I think that would work well if the 12cm fan was a case fan attached with a duct. Which would be a lot easier if this thing was vertical like every other fully heatpipe based heatsink, I think thermalright is planning to make one of those for P4/A64 and has created this strange intermediary to use it's socket A customers as guineapigs.

This is the first time I've been excited about a HSF since the AX-7. I'll happily be a guinea pig for Thermalright

.
I also agree. With the XP-120 and XP-90 out in full force, another socket would do well for a new design.

DG · Post by DG » Fri Oct 29, 2004 6:15 pm

Rusty075 wrote:but I can't seem to find any place that sells a 92-120 adaptor. (I suppose I could always home-brew something)

Here's one: http://www.frozencpu.com/cgi-bin/frozencpu/duc-13.html

tof · Post by **tof** » Mon Nov 01, 2004 10:15 am

MikeC wrote:120->92 adaptor -- that's silly. Going by the XP-90, a 92mm fan will provide plenty of cooling with the SI97 for any socket A/370 CPU.

Why silly ? 120mm will provide the same air flow, while running slower ?

Talz · Post by **Talz** » Mon Nov 01, 2004 10:36 am

Well the heatsink is obviously not designed for an especially easy conversion to 120mm. The way I see it a 120mm would be adding expense and significant hassle that should not be needed to get good quiet cooling for Socket A. You might get slightly better performance per db though, so if you want to try it I say go ahead, I'm just doubtful that the benefit will be especially worth the trouble in this instance.

DG · Post by DG » Mon Nov 01, 2004 4:29 pm

Another review, but it's in german: http://www.hartware.net/review_427.html

soujir0u · Post by **soujir0u** » Fri Nov 05, 2004 3:56 am

I just got mine today, first impression was it looks really nice. But the mounting mechanism does seem a bit dodgy and insecure, I'm really scared it's gonna fall out someday but I guess this being Thermalright it should be ok.

So far temperatures have definitely improved. My CPU is currently running at 210MHz x 10.5 multiplier at 1.47V. With the SLK800 and Panaflo L1A at 7V I used to get around 42 degrees C idle and 49 degrees load.

Now with the SI-97 and Papst 92mm at 7V (same amount of noise compared to the SLK800/L1A@7V), I get 39 degrees idle 44 degrees load. Might get better when the Arctic Silver cures.

wim · Post by **wim** » Fri Nov 05, 2004 5:37 am

old thread, but i only just saw it first time today

mathias wrote:

What on earth did they do that for?

...I think it could be modded by welding a passive cooler to the base. Like maybe an old orange zalman northbridge heatsink(NB32?), but with the pins bent out in all directions.

i had a think about this, and i half suspect doing it might actually make things worse. i don't know enough about the physics of heatpipes to be sure whether this reasoning is sound, but here's the idea: the small block at the base needs to get sufficiently hot to boil the liquid. if you add a large sink at that spot, some of the energy goes into heating up the sink and less into the phase-change action. when the phase-change process is not yet saturated, this might be detrimental (because the superconducting heatpipe is probably more efficient means of moving heat away from the CPU than your extra sink).

so 'what on earth did they do that for?' ...is there a chance it is actually made that way purposefully to be a better design? (i.e. better cooling as well as just looking cool) if somebody here knows lots about heatpipes please shed some light / point out any false reasoning.

mathias · Post by **mathias** » Fri Nov 05, 2004 6:15 am

wim wrote: i had a think about this, and i half suspect doing it might actually make things worse. i don't know enough about the physics of heatpipes to be sure whether this reasoning is sound, but here's the idea: the small block at the base needs to get sufficiently hot to boil the liquid. if you add a large sink at that spot, some of the energy goes into heating up the sink and less into the phase-change action. when the phase-change process is not yet saturated, this might be detrimental (because the superconducting heatpipe is probably more efficient means of moving heat away from the CPU than your extra sink).

If that was so, I would expect that less heatpipes would work better and a CPU producing more heat would run cooler.

I think a 280 gram, 92 mm, 4 pipe tower would be the perfect heatsink, and I hope this is just a prototype of that, and will be replaced, including for socket A.

wim · Post by **wim** » Sat Nov 06, 2004 11:51 pm

mathias wrote:If that was so, I would expect that less heatpipes would work better and a CPU producing more heat would run cooler.

how do you reach that conclusion?
more heatpipes means more capacity for phase-change action (i.e. that process isn't saturated as easily). with a big passive sink placed there (i.e. placed where the heatpipes are acting from) you're just taking potential work away from the phase-change system.

when you look at the zalman TNN case, notice that the heatpipe blocks at the heat source are quite small, and the large heatsinks are at the other end (the destination) of the heatpipes. same reason, i think..

mathias · Post by **mathias** » Sun Nov 07, 2004 8:26 am

wim wrote:how do you reach that conclusion?
more heatpipes means more capacity for phase-change action (i.e. that process isn't saturated as easily). with a big passive sink placed there (i.e. placed where the heatpipes are acting from) you're just taking potential work away from the phase-change system.

Well, I would think that adding a passive heatsink that can dissipate, say 8 wats would be similar to decreasing the procesor heat by that(although you'd have to compensate for the extra layer the heat has to go through, and the nearby fan). I'd think it might actually be slightly beneficial for the heatpipe process because the heat would be more evened out by the heatsinks ability to store heat.

wim wrote:when you look at the zalman TNN case, notice that the heatpipe blocks at the heat source are quite small, and the large heatsinks are at the other end (the destination) of the heatpipes. same reason, i think..

I think that's because the point is to get the heat out of the case.

tof · Post by **tof** » Mon Nov 08, 2004 12:36 pm

Wim, you appear to understand quite well Heatpipes...

I am still stuck with this idea that a HP works with gravity. It means if it's horizontal, it doesn't work well. If it's upside-down, it does not work at all !
That's why I've been a bit reluctant to buy HPipe-based HSink. They all appear to have been oriented for a MB that lies horzontal, but mine is vertical in a tower case.
When I look at the XP-90 or SI-97, I think of them on my cpu, with only 2 out of 4 HPipes active... a waste

But then, scpr reviewed the XP120, which showed only 1 degrees of difference between the different orientations... I am a bit puzzled...

Cerb · Post by **Cerb** » Mon Nov 08, 2004 2:11 pm

There are heatpipes going in the proper direction for a tower case. I wouldn't be suprised if the fine width (from one side's heatpipes to another's) is not too large for the heat to be conducted through (after a point, more fin becomes useless, but I don't knw how to figure that distance out based on material). If that is the case, then orientation on those heatsinks would not matter until the CPU got too hot for one HP's phase change abilities. When that time comes, I'd bet TR will have something better anyway

.

Tephras · Post by **Tephras** » Mon Nov 08, 2004 5:03 pm

To make heatpipes work in any angle I suppose they use heatpipes that some manufacturers refers to as "superconducting" and that is explained by Stealthgirl in this thread.

Michael_qrt · Post by **Michael_qrt** » Mon Nov 08, 2004 6:41 pm

tof wrote: I am still stuck with this idea that a HP works with gravity. It means if it's horizontal, it doesn't work well. If it's upside-down, it does not work at all !
That's why I've been a bit reluctant to buy HPipe-based HSink. They all appear to have been oriented for a MB that lies horzontal, but mine is vertical in a tower case.
When I look at the XP-90 or SI-97, I think of them on my cpu, with only 2 out of 4 HPipes active... a waste

But then, scpr reviewed the XP120, which showed only 1 degrees of difference between the different orientations... I am a bit puzzled...

You are still a bit puzzled because heatpipes don't work by gravity.

Heatpipes work by capillary action. The same type of thing that allows trees to shift gallons of water from their roots to their leaves maybe 100 feet in the air. I do not have an exact explaination of the process as used in heatpipes, but the heatpipes used in all good heatpipe coolers do not rely on gravity. Gravity is obviously still present and does affect performance slightly but it is not the dominant factor.

I'd say that the XP/SI series of coolers would actually still work well upside down. It would make an interesting test and demonstration...

wim · Post by **wim** » Mon Nov 08, 2004 8:18 pm

the water climbs..visualize this by dipping the end of a strip of absorbent paper towel into a glass of water. the top of the strip gets wet pretty quickly even acting against gravity.

PorBleemo · Post by **PorBleemo** » Mon Nov 08, 2004 9:41 pm

This looks interesting. Unfortunately the base is way too small for my liking. It looks like it could fall off of the pads and crush the core too easily.

Cerb · Post by **Cerb** » Mon Nov 08, 2004 9:54 pm

PorBleemo wrote:This looks interesting. Unfortunately the base is way too small for my liking. It looks like it could fall off of the pads and crush the core too easily.

Supposedly, it is coming with a shim.

PorBleemo · Post by **PorBleemo** » Tue Nov 09, 2004 7:28 am

Cerb wrote:
PorBleemo wrote:This looks interesting. Unfortunately the base is way too small for my liking. It looks like it could fall off of the pads and crush the core too easily.
Supposedly, it is coming with a shim.

Well, I'll wait and see if it does...

Post by **Rusty075** » Tue Nov 09, 2004 7:42 am

The sample sent for testing included a shim in the retail packaging.

Talz · Post by **Talz** » Tue Nov 09, 2004 9:57 am

I ordered one and it included a shim. The only comparison I can make is that when used with a coolermaster 80mm rifle bearing/led fan it beats a vantec aeroflowl with the tmd fan by 1-2 degrees. This is with both at 12 volts. For reference I find the coolermaster fan on a simliar performance level to the panaflo L1A. This means it's much, much quieter as the TMD fan was horrible, and cooler. Also note my tim application is suspect, I rushed it and didn't have anything on hand except for a piece of paper to smooth it out with.

I haven't messed around much with undervolting the fan because the current 120mm Antec (which is undervolted) is slightly louder than the 80mm and it's my office pc so most of the time ambient noise is at a fairly high level to begin with.

Mounting this thing while the motherboard was in the case was a pain though I think it would be easy with more room such as before mounting the board, or in a case/mobo combo that gave more clearance around the socket area to fit my hands into. My NF7 puts the socket right up next to the power supply when mounted in an AMB-3700. It is definately a capable cooler though, no complaints.

Metaluna · Post by **Metaluna** » Tue Nov 09, 2004 1:49 pm

Talz wrote:My NF7 puts the socket right up next to the power supply when mounted in an AMB-3700. It is definately a capable cooler though, no complaints.

Did you happen to notice how much the heatsink overhangs the edge of the board on the NF7? I'm wondering if this thing will fit on an NF7-S in an Evercase where the PSU is almost flush with the top edge of the motherboard. IIRC the clearance isn't much more than 1/4".

Talz · Post by **Talz** » Wed Nov 10, 2004 10:08 am

Actually it does not overhang at all, but you may end up removing either the power supply or motherboard to get that clip on. It is really seemed surprisingly secure once installed for anyone wondering, I would definately consider it safe for a lan party box, and probably even for shipping.

DG · Post by DG » Wed Nov 10, 2004 10:24 am

Talz wrote: It is really seemed surprisingly secure once installed for anyone wondering, I would definately consider it safe for a lan party box, and probably even for shipping.

That's very good news to hear.

pony-tail · Post by **pony-tail** » Wed Nov 10, 2004 9:39 pm

Actually it does not overhang at all, but you may end up removing either the power supply or motherboard to get that clip on. It is really seemed surprisingly secure once installed for anyone wondering, I would definately consider it safe for a lan party box, and probably even for shipping.

That surprised me ,I was a little concerned about it .
I am still waiting for the SPCR review - before I go out an buy one . Anyway Below Zero my local supplier is out of stock for a couple of weeks so I have time .

dan · Post by **dan** » Wed Nov 10, 2004 11:18 pm

hi,
one question i have is how this product compares with other heatpipe/heatsink solutions for socket a such as the aerocool dp-101?

also could those heat pipes hit the psu, my tisu motherobard pic is like this

http://www.soyousa.com/images/products/800x600/tisu.jpg

Cerb · Post by **Cerb** » Wed Nov 10, 2004 11:43 pm

dan wrote:hi,
one question i have is how this product compares with other heatpipe/heatsink solutions for socket a such as the aerocool dp-101?

also could those heat pipes hit the psu, my tisu motherobard pic is like this

http://www.soyousa.com/images/products/800x600/tisu.jpg

I can't answer on performance, but for thr weight, I would go for it.
As far as the PSU butting up against it...the best I can say if take the dimension in question, divide in half, and measure from the middle of the socket. I think it's about 1 1/2" (75mm?)

Talz · Post by **Talz** » Thu Nov 11, 2004 10:22 am

pony-tail wrote:
Actually it does not overhang at all, but you may end up removing either the power supply or motherboard to get that clip on. It is really seemed surprisingly secure once installed for anyone wondering, I would definately consider it safe for a lan party box, and probably even for shipping.
That surprised me ,I was a little concerned about it .
I am still waiting for the SPCR review - before I go out an buy one . Anyway Below Zero my local supplier is out of stock for a couple of weeks so I have time .

I want to point out this will vary by motherboard with socket orientation and placement. It is a large heatsink, but the design seems like it will have few flat out incompatibilities. The most likely problems I see are systems that have the socket mounted close to the power supply with a different socket direction than the NF7. I have no idea how many mobo/case combos this covers though, not something I've really payed much attention to before.

For instance the Soyo board pictured might have problems if the power supply is extremely close to the motherboard, but I would expect most cases to have adequate room.

Jojo540 · Post by **Jojo540** » Thu Nov 11, 2004 11:10 am

pony-tail wrote: I am still waiting for the SPCR review - before I go out an buy one . Anyway Below Zero my local supplier is out of stock for a couple of weeks so I have time .

Me too, so I hope the test will be finish before availability

Rusty 075 can you please include in your test some figures without fan, because I plan to use it without fan

dan · Post by **dan** » Fri Nov 12, 2004 12:05 pm

For instance the Soyo board pictured might have problems if the power supply is extremely close to the motherboard, but I would expect most cases to have adequate room.[/quote]

yeah, i have an aopen kf56, it costs less than most PSU's and it includes a PSU, anyhow, it is a industry standard mid-tower ATX with 3 5" bays.

the PSU is at the edge of the socket-side of the SOYO TISU motherboard.

it's good for undervolting/overclocking tualatins, has a nice sound chip.