Both Mad and Hart like, where possible, to test using a "standard fan". Hart uses the venerable Papst, while Mad switched just this year from the same Papst Hart uses to the GW NCB. Using a standard fan means that HSs can be tested under known airflow and known noise conditions. If a standard fan is used, we can forget about the ambient noise level, for instance.
Regrettably, Hart then "loses the recipe": while they test from a high RPM to a lower RPM, they aren't consistent. The lower RPM is sometimes 840, sometimes 780, and sometimes different yet again. While the upper RPM is commonly 1040, this is also sometimes variable. This makes it tough to use datamining techniques.
Mad, on the other hand, consistently uses "high" and "low" fan settings, which a highly placed mole assures me means 12V and 6V. Us SPCR folk know that means a 15dBA noise difference (for a good fan, and Mad standardizes on a good fan). So us SPCR folk know that if the sound level (at whatever distance) is 35dBA at high, the low sound level will be 20dBA or less, regardless of the ambient noise level. ("Less" because the 35 figure may also be too high because of the ambient, which includes the internal noise of the SLM.)
Datamining: a newish term for an old old procedure: if you have tons of data, you can sort through it to find other information that was irrelevant to the original purpose for compiling those tons of data. In this particular instance, I want to find hints as to which HSs will cool well at much lower airflow (and hence much lower noise) than was tested.
As an example, I've selected three pages from two of Mad's reviews: the 2006 HSF roundup and the recent 2007 roundup. Let's see what we can find from same-fan data:
from the 2006 review, page 11 (775 HSFs)
Ninja 55C 52.5C 2.5C 30.5C 8.2%
Translation: 55C = low, 52.5 = high, 2.5C is the difference. Since 22C is the ambient, the best (high) cooling is 30.5C above that. If the fan is turned down to low, the temperature increases 8.2% of 30.5C. Got that? OK, let's start over again:
from the 2006 review, page 11 (775 HSFs)
Ninja 55C 52.5C 2.5C 30.5C 8.2%
NPH Big 57 51 6 29 20.7%
from the 2006 review, page 3 (AMD HSFs)
Ninja 64 59 5 37 13.5%
Mine 64 59 5 37 13.5%
Shogun 70 60 10 38 26.3%
Tuniq Twr 69 57 12 35 34.3%
from the 2007 review page 10 22C ambient
Ninja 55 52.5 2.5 30.5 8.2%
NPH Big 56.5 52 4.5 30 15%
Katana 2 60 53.5 6.5 31.5 20.6%
Andy-Sam 63.5 54.5 9 32.5 27.7%
Tuniq Twr 56.5 49 7.5 27 27.8%
K Cross 67 56.5 10.5 34.5 30.4%
The 2007 table simply repeats the Ninja data from the 2006 review. Mad often recycles data. If I wuz real careful, I'd triple-check my data and computations over a 3-day period. Instead, I'll make a nitpicker maybeso happy by not triple-checking.
The above sez if I'm interested in cooling at low airflow, I should look at the Ninja. Zounds!! Brand new info!! (cough). Well, my problem is, a Ninja - and I own 3 of the original version and one Rev B - won't fit in any of my big-fan cases. I'm limited to small towers - Freezer 64 Pros - or shelf- or flower-type coolers.
For some time now, I've been reading HSF reviews looking for A) good cooling, and B) little falloff as the fan is turned down. My conclusions?
My latest #1 computer is in the Mio 5601 case. I'm using a 220mm sleeve-bearing fan, 600RPM nominal @ 12V. I have an Enzotech Ultra-X with a SilenX "11dBA" 120mm fan. Both fans run at 4V, and I use a custom start circuit to make sure they get running at that low voltage. The airflow at 4V from the big fan allows me to run my PSU with no fan - passive cooling. I selected the Enzo because it'll fit in my case, has excellent cooling, and with 1.8mm fin spacing on its 120mm x 120mm cooling grid, provides excellent cooling for my CPU with 4V on the fan.
"0dBA" is not a total absence of sound - that would be "minus infinity dBA". 0dBA is simply the accepted limit of hearing for a normal human being. My computations say that the big side fan and the 120mm CPU fan each produce less than 0dBA at 4V. Nobody can measure this, so nobody can prove, or disprove, that using actual measurements.
But you could get a good idea by progressively turning down the fan voltage while holding the fan closer and closer to your ear (the 0dBA is at one meter). If you don't hear the bearings "go crazy" at some voltage, then my figures are probably close.
My next #2 computer will be in the Aplus CS-188A case, which the Enzo won't fit - the case is too narrow. I'll transfer my GeminII HSF that has the phoney-baloney "120mm X2" stuff discarded and is being cooled with two Scythe 100mm fans, as low as I can make them run (they're due for a custom starter circuit so I can further reduce the voltage).
I'll use one of my Xclio cases to experiment with medium-fancy ducting and the Dominator HSF as a passive HS. I've already proved the Dominator works passively (and posted my measurements here); now I want to see just how well I can make it work. [The Dominator has a 140mm x 140mm cooling grid with the fins spaced almost 2.4mm.]
That leaves another Xclio case, already DIYed, quickly available for other tests that might arise in the future. For example, if somebody figures out how to mod the new MaxOrb HSF with a GW NCB fan, I'd want to give the MaxOrb a try.
There's always something new and interesting.
