Thermalright IFX-14 - a first review

Cooling Processors quietly

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

sheninat0r
Posts: 76
Joined: Mon Oct 29, 2007 2:02 pm

Post by sheninat0r » Tue Feb 19, 2008 12:34 pm

johnniecache7 wrote: ...the Ultra 120 Extreme beats everything you can throw at it including the Ninja.
New heatsink review method: throw ninjas [the ones with shurikens and swords] at them until they break, whichever takes more hits is the best... :lol:

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Thu Feb 21, 2008 2:42 pm

halcyon wrote:Another review, this time from Madshrimps, although the noise measurements are wanting:.
explain, please.... :)

halcyon
Patron of SPCR
Posts: 1115
Joined: Wed Mar 26, 2003 3:52 am
Location: EU

Post by halcyon » Sat Feb 23, 2008 12:29 am

I didn't mean it in a bad way, but compared to the SPCR reviews and general measurement recommendations for industrial noise:

- high bg noise (36 dBA)
- no spectral characterisation
- no loudness estimates (only SPL)
- mis-characterisation of 'twice as loud' (based on 10 db(A) diff)
- 44dB(A) as a treshold of quiet (!)
- no measurement device noise floor, measurement position or other measurement setup characteristics explained

Even with all that said (written), the data is of course useful in comparative sense and does reinforce the impression from other reviews.

nyu3
Posts: 144
Joined: Wed Feb 18, 2004 6:07 pm
Location: Canada

Post by nyu3 » Sat Feb 23, 2008 1:07 am

My biggest problem with madshrimps is their graphs: they just lump everything into two big graphs. Would be nice to see the data organized in a more meaningful manner. For example, show how heatsinks perform with reference fans like SPCR and that german silent website.

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Sat Feb 23, 2008 1:35 am

uhm :shock: in the graphs ALL heatsinks are with reference fan the text above the chart explains all AND also tested with stock fan, comparing performance of reference vs default fan....

@Halcyon:

- high bg noise (36 dBA)
below 30dBA noise floor in a normal house.. I find that too hard to believe, sorry:) 36dBA here is dead quiet; can hear a car driving by slowly in the street... 1km away... if you pay attention. at night where meter drops below 36, goes toward 35dBA; it's like being in no-manslands:) not yet met anybody where noise floor was near 30dBA...

- mis-characterisation of 'twice as loud' (based on 10 db(A) diff)
come again? "As a rule of thumb a 10dBA. increase corresponds approximately to a doubling of perceived. loudness (eg 60dBA sounds twice as loud as 50dBA)"

- 44dB(A) as a treshold of quiet (!)
do note that 44dBA IS quiet in those charts, you're skipping the part where the noise is not at 1m, but at 1CM! noise drops quickly over distance;

- no measurement device noise floor, measurement position or other measurement setup characteristics explained
dBA meter is placed right at the edge of the case - with side panel removed and I showed a photo... what device used, etc.

Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

Post by Felger Carbon » Sat Feb 23, 2008 4:26 am

jmke, I am mostly on your side WRT the above. However, I need to take issue with you on home noise floors:
jmke wrote:not yet met anybody where noise floor was near 30dBA...
A long time ago I lived on the edge of Irvine Ranch in Orange County, California. I used a General Radio Type 1 one inch electret microphone and related gear to measure 23dBA in my bedroom at 3AM once. The self-noise of that equipment was 14dBA (note 1).

I now live where it gets cold - 0F in the winter is not uncommon at night. To keep the heating bill down, double-glazed windows are common here. A side effect is, outside noise is cut down considerably. I now own an Extech Type 2 SLM with an electronic noise floor of 26dBA, and in the winter (now), at night, I commonly measure less than 27dBA total noise floor. This indicates to me that my acoustic ambient noise floor is about the same 23dBA I knew before.

While I believe that 30dBA would be very quiet in an urban, high-density setting, those of us lucky enough to live in rural or semi-rural areas can have home noise floors well under 30dBA, esp at night.

Note 1: 14dBA was the specified self-noise of the 1" electret and preamp. I actually confirmed that in an acoustic chamber. Honest. The acoustic chamber was a little over a foot long and about 4.5" in diameter, made of brass, with a wall thickness of 1 1/2". It was just big enough to contain the preamp and microphone with no danger of touching the inside wall of the chamber. We (me and my employees) had to double-suspend, using bungee cord, this acoustic chamber for mechanical vibration isolation, along with the preamp power/signal cord. This was in the late 70's, early '80s. I wuz doing acoustic experiments a tad before MikeC started SPCR... :P

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Sat Feb 23, 2008 6:58 am

I don't live anywhere near the city FC, closest are Farms :lol:

tjl30
Posts: 7
Joined: Sun Aug 05, 2007 12:57 pm

Post by tjl30 » Sat Feb 23, 2008 9:56 pm

http://www.tomshardware.com/2008/02/20/ ... arts_2008/

Toms Hardware also reviewed this heatsink. From what I am reading this seems like the best choice for cooling my CPU.

Right now I am looking into replacing my stock CPU and GPU cooling.

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Sun Feb 24, 2008 5:44 am

do note that the huge size and high cost are not the best features ;)

nyu3
Posts: 144
Joined: Wed Feb 18, 2004 6:07 pm
Location: Canada

Post by nyu3 » Mon Mar 03, 2008 10:40 pm

jmke wrote:uhm :shock: in the graphs ALL heatsinks are with reference fan the text above the chart explains all AND also tested with stock fan, comparing performance of reference vs default fan....
Look at the graph below for example. Fan types: stock, pabst, ncb, delta1, delta2; voltages:high, low, lower, auto. This is too much information for one graph. Needs to be organized in a useful manner.

http://www.madshrimps.be/?action=getart ... rticID=680

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Post by MoJo » Wed Mar 05, 2008 11:11 am

The photo at the bottom of this page is very interesting:

http://www.xsreviews.co.uk/reviews/cpu- ... t-ifx-14/4

(ImageShack mirror:)

Image

I'm not sure I blame TR, perhaps they want their heatsinks to be slightly convex so that they make better contact with concave Core 2 IHSs :(

Considering the price of the thing, you would expect it to be flat though. Since lapping practically seems to be a requirement for the hotter Core 2 chips anyone bothering to splash out on one of these will most likely want it to be flat. Either that, or the mounting springs should apply enough pressure to flaten the IHS out a bit anyway.

3x3cUt0r
Posts: 2
Joined: Wed Mar 05, 2008 11:07 am
Location: Colombia

Post by 3x3cUt0r » Wed Mar 05, 2008 11:21 am

That Heatsink is too big and too heavy for my taste, i bet it won't even fit on my case.

FartingBob
Patron of SPCR
Posts: 744
Joined: Tue Mar 04, 2008 4:05 am
Location: London
Contact:

Post by FartingBob » Thu Mar 06, 2008 2:29 pm

Apart from aesthetics and the backplate (which i believe you can now buy separately), is there much reason why someone would choose this over the TRUE, for OCing or silencing? It costs less and is smaller.

Chris Beard
Patron of SPCR
Posts: 58
Joined: Fri Aug 03, 2007 3:16 am

Post by Chris Beard » Fri Mar 07, 2008 7:43 am

FartingBob wrote:Apart from aesthetics and the backplate (which i believe you can now buy separately), is there much reason why someone would choose this over the TRUE, for OCing or silencing? It costs less and is smaller.
TRUE doesn't really cost less. These are around £35 now. The reason I chose it is the 140mm fan mounting. It has temps around the same as the TRUE.

It does need lapping though - oh man does it need lapping.

In terms over overclocking I believe if you strap 3 fans to it then it does better than the TRUE with a 2 fan setup. That's all a bit... loud for these forums though.

halcyon
Patron of SPCR
Posts: 1115
Joined: Wed Mar 26, 2003 3:52 am
Location: EU

Post by halcyon » Sun Mar 09, 2008 2:24 am

Jmke,

I'm sorry if I came off as harsh. I should not write in a hurry. Let me walk through what I meant:

jmke wrote: - high bg noise (36 dBA)
The noise level of a recording/monitoring room should be set with noise criteria contours (NC) or equivalents like NCB or NR.

While 36dBA fall within the range NC range of 25 (c. 38dBA wide-band) and as such is what would be considered as quiet in an average urban residential room, it is at the upper limit of what can be considered sufficient for a recording environment. (ref: Acoustical Noise Control, Doug Jones, in Handbook for sound engineers, 3rd ed, Glen Ballou, ed).

This means that the total system noise of the recording environment + the system clearly increases the systematic error in the measurements and sets the lower limit of what can be accurately measured.

BTW, now that I read page 4, I'm a little confused. Is 36dBA bg noise with the system on, without the system on or the same in both situations?
jmke wrote: below 30dBA noise floor in a normal house.. I find that too hard to believe, sorry:) 36dBA here is dead quiet; can hear a car driving by slowly in the street... 1km away... if you pay attention. at night where meter drops below 36, goes toward 35dBA; it's like being in no-manslands:) not yet met anybody where noise floor was near 30dBA...
That is often true for A-weighted levels. However, I was trying to refer to the recording environment noise control levels, not perceived loudness levels. But you are of course right, it matters that if one is in a room perceived to be quiet, then sound sources above JND in spl will be heard and that for this the bg level if perfectly adequate for subjective assessment.

BTW, we all should be using ITU-R 468 as the weighting for noise sources, but that's another issue altogether.
jmke wrote: - mis-characterisation of 'twice as loud' (based on 10 db(A) diff)
come again? "As a rule of thumb a 10dBA. increase corresponds approximately to a doubling of perceived. loudness (eg 60dBA sounds twice as loud as 50dBA)"
Yes, 10dB is the rule of thumb often used in engineering. How ever, psychoacoustics is the science that measures this.

"Half-Loudness corresponds to a 6-dB attenuation rather than the 10 dB suggested..." (ref, Psychology of Hearing, 5th ed., Brian Moore, 2003).

There is considerable variance based on spectral distribution, preceding sounds, masking and other factors.

I agree that my choice of words 'mis-charaterisation' was not the best one, but I think we should consider that in a quiet setting with no significant masking sounds, a 6dB increase in the SPL of a wide bandwidth sound is likelier to approximate the perceived doubling of loudness.

jmke wrote: - 44dB(A) as a treshold of quiet (!)
do note that 44dBA IS quiet in those charts, you're skipping the part where the noise is not at 1m, but at 1CM! noise drops quickly over distance;
I'm afraid I have misunderstood you here and did indeed miss the page with the photo. :oops:

So you measure at 5 cm and then consider the level of perceived loudness of that sound 1m away from the origin (within the 38dBA bg noise room)?

Now I understand.
dBA meter is placed right at the edge of the case - with side panel removed and I showed a photo... what device used, etc.


Can't find data for SL4001A. What is the self-noise, measurement range, accuracy and error?

Again, my apologies if I came off as harsh, but having missed your page 4 (my bad) and not understood your measurement/listening distance setup properly, I came up with the conclusion that the setup was a bit different than what it actually was.

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Sun Apr 06, 2008 10:45 am

hey Halcyon, I've missed your reply, never got a mail, although I'm subscribed to the thread :/

anyway, the meter we're using is http://www.reliabilitydirect.com/soundl ... -AR824.htm
not very high end, we use it especially for comparative measures with the same unit, for comparison purposes, we've found the unit to perform on par at 35dBA+ levels to more expensive models. Hence I changed my measurements to record at closer distance from the CPU heatsink, to get repeatable results, and usable for comparison. But you also know that noise drops off quickly over distance. So a 5dBA difference in the chart at ~5cm will not be as noticeable ~1m away.

I dugg around deep to find out the noise floor levels; as this meter is limited at 30dB, in A level it never goes under 35.5dBA in this room, dead of night, but it's the meter, as it turns out :roll: ; also other models which cost me around double ($200) had similar outcome.

With a $7000 dBA meter I recently got a chance to use, measured by trained professional (not me;)) the noise floor of the current test room without any equipment running was 22.2dBA; with one machine running (not the CPU test station) it was at 32dBA at ~1m (yes that machine is loud(er) but still very quiet for most people).

a bit sad :cry: that I'm not capable of buying the $7000 unit, will look into obtaining a higher end model though, also building a noise box, but all in planning stages.

Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

Post by Felger Carbon » Sun Apr 06, 2008 6:01 pm

halcyon wrote:
jmke wrote: - mis-characterisation of 'twice as loud' (based on 10 db(A) diff)
come again? "As a rule of thumb a 10dBA. increase corresponds approximately to a doubling of perceived. loudness (eg 60dBA sounds twice as loud as 50dBA)."
Yes, 10dB is the rule of thumb often used in engineering. How ever, psychoacoustics is the science that measures this.

"Half-Loudness corresponds to a 6-dB attenuation rather than the 10 dB suggested..." (ref, Psychology of Hearing, 5th ed., Brian Moore, 2003).

There is considerable variance based on spectral distribution, preceding sounds, masking and other factors.
A long time ago, when I too had direct access to expensive SLMs and 1" GenRad electret microphones, I remember that the textbooks stated that at normal sound levels a doubling of noise level corresponded to 6dBA. Only at very low sound levels did doubling correspond to 10dBA.

As I say, it was a long time ago - in the late 1970s. I may have misremembered. JMKE, I now own a $300 Extech 407738 SLM, with a self-noise of 26dBA. In the dead of winter, at 3AM, I measure 26.9dBA in my storage bedroom and even now, 27.9dBA is not unusual in my front room when there's no traffic.

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Mon Apr 07, 2008 12:43 am

^^ I'm getting ~22dBA here with $7000 meter ;)

Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

Post by Felger Carbon » Mon Apr 07, 2008 7:59 am

jmke wrote:^^ I'm getting ~22dBA here with $7000 meter ;)
In ~1978, with a then-equivalent of your $7000 meter, I measured 23dBA in my bedroom on the edge of Irvine Ranch in Orange County, CA, at 3AM (my favorite time for testing noise floors). No double-glazed windows then, which I do have now. That 23dBA is the lowest I've ever measured in one of my homes, but then it was the only one where I had the proper equipment at hand (as you just had).

If I was a really nice person, JMKE, I wouldn't point out that you (before the $7000 SLM) asserted that sound levels such as 22 (or 23) dBA in home settings just weren't possible... :P

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Mon Apr 07, 2008 10:01 am

I still know what I said back then, and will look into it further before I burn all my previous statements ;) yes below 30dBA which is dead quiet will not be obtainable everywhere, let alone in homes.

I'm a nice person too, claiming the below 30dBA, not below 20dBA ;)
anyway, in the run for my own $7000 meter at the moment, in 2 days maybe the owner of a more expensive dB meter

Post Reply