SI-97 vs SP-97 vs CNPS7000A?

Jojo540 · Post by **Jojo540** » Wed Dec 01, 2004 9:01 am

wim wrote: if surface area (top) of the heatsink is only 92^2 mm^2 and the fan area is 120^2 mm^2, maybe at the perimeter (incidentally where there is max airflow) it is wasted on the si-97? or you are planning to build a sort of 'funnel' to pipe the 120 down onto the smaller top of the heatsink?

The mesured dimensions of the SI-97 fins are 92*96mm, diagonal 130mm, so with a 92*92mm fan, you don't cover all the surface of the fins, but only the middle of the fins.
With the 120mm I cover the whole surface of the fins and more, taking advantage of the perimeter best airflow on the heatest points (along the heatpipes), like on this pic:

wim wrote: can't remember but is the hub on 120 mm fan bigger than on a 92? if so you might have a bigger 'dead spot' right where it counts, and maybe not be able to get any better performance with a larger fan?

I haven't 92mm fan, but the hub on a PAPST 80mm is 36mm diameter, on a Globe fan 120mm the hub diameter is 44mm, so we can extrapolate than for a 92mm the hub diameter is between 36 and 44mm.

But, with my mod there is a space of 7mm (on blowing, and 10mm if extracting) between the top of the fins and the hub of the fan, instead of 4mm with a 92mm fan, see pic below:

I'm happy with that mod, because it's not definitive, I didn't alter fan + heatsink, just add an "interface".

So if the results of cooling are not as good as expected I can go back to 92mm fan.

sim`on · Post by **sim`on** » Wed Dec 01, 2004 10:58 am

the best of the three would be the SI-97, but i would (and did) buy neither of them. i just changed my thermalright SP-97 for a aerocool ht-101 tower-cooler, and i'm quite pleased with it.

the three coolers you proposed all ruin your airflow, plus they all tend to generate vibration noises, and it's quite hard to prevent them from doing this. the only cooler that offers better/equal temperatures to the ht-101 with similar noise is the SI-97, but the airflow-problem remains.

Jojo540 · Post by **Jojo540** » Wed Dec 01, 2004 11:16 am

sim`on wrote: i just changed my thermalright SP-97 for a aerocool ht-101 tower-cooler, and i'm quite pleased with it.

Do you use it fanless?

intx · Post by **intx** » Wed Dec 01, 2004 2:17 pm

sim`on wrote: but the airflow-problem remains.

airflow problem?

anyway, regarding jojo540's statement about using a 120mm fan.. what about the deadspot?

sim`on · Post by **sim`on** » Wed Dec 01, 2004 2:27 pm

nay, it's currently running with the stock fan at ~1000 RPM (really good fan once you've slowed it down a bit - even less speed is beyond my possibilities atm.), but i guess a semi-passive solution would work with some ducting. anyway, ducting or using PSU-fan for cooling should be a lot easier than with a "tried and true-layout" cooler, as the fins are already pointing to the right direction (unless you have a lian-li PC50?).

with an undervolted CPU, you could even manage to run it passive, as long as there is enough ventilation around... but it's constructed for being actively ventilated, and it's doing it's job quite well.

edit:

airflow problem?

the standard-layout coolers don't do anything for the airflow you ought to have in the case, on the contrary they disturb airflow by blowing into the direction the fins are pointed to. the result is al lot of mixed up airflow, just in the middle of your case. probably a lot of the air that was already blown through the cooler is re-used, so the cooling-efficiency isn't at the optimum.
a tower-cooler, blowing the air towards case or PSU-fan, supports a pointed airflow in the case, resulting in lower case- and CPU-temp, or at least air going where you want it to go

Jojo540 · Post by **Jojo540** » Wed Dec 01, 2004 2:44 pm

sim`on wrote:nay, it's currently running with the stock fan at ~1000 RPM

The 80mm stock fan?

sim`on · Post by **sim`on** » Wed Dec 01, 2004 3:18 pm

right, the tiny little 80mm stock fan. bigger fans wouldn't make too much sense, unless drop one fan and use a duct to a 120mm case- or PSUfan.

intx · Post by **intx** » Wed Dec 01, 2004 3:33 pm

sim`on wrote: the standard-layout coolers don't do anything for the airflow you ought to have in the case, on the contrary they disturb airflow by blowing into the direction the fins are pointed to. the result is al lot of mixed up airflow, just in the middle of your case. probably a lot of the air that was already blown through the cooler is re-used, so the cooling-efficiency isn't at the optimum.
a tower-cooler, blowing the air towards case or PSU-fan, supports a pointed airflow in the case, resulting in lower case- and CPU-temp, or at least air going where you want it to go

What temperatures are you getting with your current HSF (including ambient)? Or what is it's actual CFM rating and how many watts is your cpu processor running at?

I see your point after looking at the heatsink myself

I've always found that design easier to work with, airflow wise.

It is still possible to use regular 'blow towards' heatsink design without screwing up airflow. josephclemente of SPCR does it quite nicely with his Zalman CNPS7000A, all while maintaining a decent look. However, it is more work, and may (likely) require case modding depending on your case choice.

The SP-90's review here said the fan could work both ways, but unlike advertised, there is a performance hit when it's blowing away from the HS. Still the performance difference is not large, so if desired you can still duct the air out of the case.

I do wonder, however, if the HT-101 or another tower heatsink was compared to the SI-97, ducted, which would perform better?

wim · Post by **wim** » Wed Dec 01, 2004 9:35 pm

Jojo540 wrote:so we can extrapolate

interpolate

Jojo540 · Post by **Jojo540** » Wed Dec 01, 2004 11:12 pm

wim wrote:
Jojo540 wrote:so we can extrapolate
interpolate

No, no I confirm, extrapolate

Useful blowing area of a 92mm fan: 6217 mm²
Useful blowing area of a 120mm fan: 9785 mm² - it's 50% more than 92mm fan

For that calculation i didn't extrapolate the hub diameter of a 92mm, just took the 80mm fan diameter.
The difference is impresive.

The problem of airflow is common to ALL heatsink with fan, except if you have an airduct, it's due to the ATX board layout.
I think this problem will be solve with the BTX boards.

sim`on · Post by **sim`on** » Thu Dec 02, 2004 7:23 am

What temperatures are you getting with your current HSF (including ambient)? Or what is it's actual CFM rating and how many watts is your cpu processor running at?

the stock fan does 29.6cfm at ~2500 rotations. however, i have no idea what the actual performance is, as i said it does ~1000 rpm with speedfan at 1% (mobo abit nf7-s, mCubed fan-control coming this christmas

).
the processor is a XP 2500+, when i run it at original voltage 1,65V, it ought to produce 68.3W maximum, my mobo reads:

idle: CPU 36°, Case 28°
load: CPU 52°, Case 35°

idle means: only speedfan, zonealarm, antivir, s2k bus disconnect, prime95 (turned off) running
load: speedfan zonealarm, antivir, s2k bus disconnect, prime 95 on "inplace large FFTs" torture test

normal room-temperature (no method of measuring). the problem is: the abit nf7-s temperatures are not precise at all. the temperature will differ if the PC ist booted normally, comparing to the temperature when being restarted. this is in no relation with the real temperature, as it even occurs if you check in the bios right after start/restart (you start, it shows CPU ~18°, you reboot, it shows 26°, you shut down, start: 20°...)

I do wonder, however, if the HT-101 or another tower heatsink was compared to the SI-97, ducted, which would perform better?

i guess, they would perform quite similar, if ducted perfectly the SI-97 would be a bit cooler i guess. but on the other hand, the ht-101 is a lot easier to duct.

for anyone who is interested in a comparison (german):
http://www.hartware.net/review_427_8.html

in this review the SI-97 is the definite winner, though.

EDIT: checked again and corrected the temperatures...

Jojo540 · Post by **Jojo540** » Thu Dec 02, 2004 1:45 pm

First results with the SI-97:

Testing configuration:
Antec SLK-3700 (damped) - 2 x 120mm Globe fan (in/front & out/rear)
LC power 420W - modded with 120mm Globe fan
Abit NF7-S, passive northbridge cooling, Zalman NB47J
Athlon XP 2600+ Thoroughbred (2083Mhz) - FSB 333Mhz - 1.65V
2 x Samsung PC3200 DDR 512Mo
ATI Radeon 7000
IBM 30Go - IBM 80Go

My SI-97 is modded with a 120mm Globe fan as shown a few posts before, blowing to heatsink as recommended by Thermalright.

The 3 fans, PSU, in & out are at 6V, ambient temp is 21°C.

On idle, only eMule, IE, Speedfan, NSW, NIS running, I got following CPU temps:
CPU temp: 31°C - with SI-97 fan at 12V
CPU temp: 34°C - with SI-97 fan at 7V
CPU temp: 38°C - with SI-97 fan at 5V

Tomorrow I'll remove the SI-97 fan for best air circulation and test it passive, I'll also try fan sucking air from heatsink.

ChucuSCAD · Post by **ChucuSCAD** » Thu Dec 02, 2004 5:14 pm

I have been sitting aside watching this banter and I really just can't stand it anymore. So for what it's worth here is my 2 cents....

For someone who was that concerned about methodology and "real comparisons", that's a pretty half-assed test you've done there:

*You're using a motherboard that does socket temps only.
*You're using a motherboard that is known to be inaccurate by as much as 10°. (and I can almost guarantee that you haven't calibrated your temp readings, right?)
*You're wasting an unknown percentage of the 120mm fan's CFM by blowing it off the side of the heatsink, so you won't even be able to make comparisons between the heatsinks on the basis of CFM/°C
*Knowing the ambient temp is nice, but without the case's internal temp there's nothing to base a dT number on.
*eMule as a loading program!?!?!? What if next test you have more porn being downloaded/uploaded? Or less? You have no real idea what wattage the CPU is drawing.

So, in summary, you've got an unknown and variable number of watts being dissipated through the heatsink by an unknown CFM, and results that are being reported by just about the most inaccurate way possible.

I understand that you wanted the results faster than waiting for SPCR to do them, but you could have at least read a couple of the reviews here, or better yet the Cooling Methodology article so you'd understand the how's and why's behind testing heatsinks in a meaningful way.

Hopefully the real review will be up soon.

chucuSCAD

wim · Post by **wim** » Thu Dec 02, 2004 7:20 pm

heh..unfortunately..i agree

Jojo540 · Post by **Jojo540** » Thu Dec 02, 2004 8:58 pm

ChucuSCAD wrote:
For someone who was that concerned about methodology and "real comparisons", that's a pretty half-assed test you've done there:

You haven't read my previous posts (even the title of the topic where you are replying), I'm not doing a general test, but a comparison between 3 heatsinks in my rig, with my current use of it.

I'm not comparing on charge, but on idle.

I think you don't know what is the difference between a review and a comparison.

A comparison is relative, a pure test is absolute.

So to compare I must test those 3 heatsinks in same real conditions with the parts I own:
- same motherboard, even if you don't like it it's the one I own
- 120mm fan, even if part of the flow is out of the fins, it will help cooling northbridge, ram etc....it it's the one I own
- eMule is not a loading program, I just inform of what I where using, when I'm on IDLE, it it's the one I run.

ChucuSCAD wrote: What if next test you have more porn being downloaded/uploaded?

In order to keep polite I prefer not comment this sentence, which reveal us your use of eMule

Post by **Rusty075** » Thu Dec 02, 2004 9:35 pm

I think the point he was making Jojo, was that even for the purposes of your "comparison", your methodology is flawed. The peculiarities of the procedure mean that it will not be accurate, or meaningful, to anyone, including you.

Your comparison won't even be relative if the methodology has poor reproducibilty. If you're running eMule, you're not idling, and your CPU wattage can vary wildly (50% or more). Why not use one of the simple CPU loading programs to get more consistant results?

Rather than circling your defensive wagons and going into flame mode, perhaps you should look at what he and wim said, and look for ways to improve the testing. Launching into a personal attack against someone who took the time to try to help you will not earn you friends. (it's also against forum policy, consider this a friendly reminder)

As I commented before, you're a bit new here, and probably not as familiar as Chucu and Wim are about the drive this place has for empirical data over conjecture.

sim`on · Post by **sim`on** » Fri Dec 03, 2004 1:43 am

a short remark by myself (before some other admin starts to flame me, too

): my idle/load "test" is quite messy, too. i don't have the time nor equipment to test or compare SP-97 to HT-101, but i can assure you, that my personal feeling is, that the HT-101 does the job a lot more quiet, giving approximately equal temperatures. the temperatures i mentioned above do vary from day to day, as (like i said) the mobo does not read CPU-diode, and generally gives messy values, varying from bootup to bootup.

acutally i only wrote something about temps because i got asked

ChucuSCAD · Post by **ChucuSCAD** » Fri Dec 03, 2004 6:41 am

Jojo540 wrote:
ChucuSCAD wrote:
For someone who was that concerned about methodology and "real comparisons", that's a pretty half-assed test you've done there:
You haven't read my previous posts (even the title of the topic where you are replying), I'm not doing a general test, but a comparison between 3 heatsinks in my rig, with my current use of it.

In responce to that... well quite frankly it does not matter what the original topic was, as always they tend to stray a bit... I also have carefully read each post and stand firm that your method is flawed.

Jojo540 wrote: I'm not comparing on charge, but on idle.

I think you don't know what is the difference between a review and a comparison.

A comparison is relative, a pure test is absolute.

In order to make an accurate comparison or a test you have to have STRICT testing conditions to duplicate load. a system at idle does not consistantly load/unload, depending on how you look at it, a cpu to be accurate for a test.

chucuSCAD

intx · Post by **intx** » Tue Dec 07, 2004 6:49 am

another review, more geared towards quieter fans:
http://www.techniz.co.uk/modules.php?na ... ent&id=151

mathias · Post by **mathias** » Tue Dec 07, 2004 7:02 am

intx wrote:another review, more geared towards quieter fans:
http://www.techniz.co.uk/modules.php?na ... ent&id=151

The boxshot is very informative, first time I heard there would be at least three more heatsinks in the SI-** series, I think one that would fit XP's and A64's would be much better, though I still need something that'll keep the heat away from my PSU, I'm hoping one of these SI's will be a tower.

JanW · Post by **JanW** » Wed Mar 02, 2005 10:20 am

MikeC wrote:The test data for the SI-97 is in my hands and what's required is for me to pull it all together into a review article. This should happen within a week.

I'm sorry to bump this very unfortunate thread. Is the SI97 review still on the agenda? Or have I missed it

?