Fans!!!! The most complete and comprehensive array of tests and benchmarks - Page 16

Quote:

Originally Posted by Cyclops 
It's all well and good that some fans like Delta AFB1212GHE-CF00 can bring the tempertures down to 38.9C (In my test suit), but it does that by generating over 80 DB of noise which is unbearable in a typical household.

The delta temperature is halved when you double the flow rate, which is why there is little benefit to very high RPM fans when your using a heat sink with a low pressure drop. Here's an example:

Let's assume 0 pressure drop (ventilating resistance, or restriction) to make it easier.

Scenario A:

Fan A @2000RPM: 50CFM, 2*C ΔT, 40dB

Fan A @4000RPM: 100CFM, 1*C ΔT, 55dB

You lower the ΔT by 1*C, but at what cost (in terms of noise)? Well, doubling the RPM will increase the noise by 15dB. Definitely not worth it.

Scenario B:

Fan A @1000RPM: 25CFM, 10*C ΔT, 25dB

Fan A @2000RPM: 50CFM, 5*C ΔT, 40dB

This time your ΔT drops by 5*C. You may have increase the noise by 15dB but Fan A @1000RPM was very quiet, and now the fan is at a tolerable noise level. So with little cost in terms of noise, you have lowered the ΔT by 5*C.

Scenario C:

Fan A @500RPM: 12.5CFM, 20*C ΔT, 10dB

Fan A @1000RPM: 25CFM, 10*C ΔT, 25dB

ΔT is now 10*C lower, but the fan is still very quiet. Quite a large gain for almost no cost in noise.

Equations used for the above:

CFM = 3.16 x W / ΔT (°F)

Assuming specific heat and density for sea level.

CFM₂ = CFM₁ (RPM₂ / RPM₁)

N₂ = N₁ + 50 log₁₀(RPM₂ / RPM₁)

Where: N = noise (dB)

Assuming speed and noise vary in proportion to the fifth power.

Note: ΔT is the difference between the inlet and outlet temperature of the heat sink and fan/s combination.

Keep in mind that +10dB is a doubling in noise. Let's add 40*C to the ΔTs since the heat transfer between the CPU and heat sink is not 100% efficient. Based on the above and using your rating, the lower the RPM, the better the fan.

Quote:

Originally Posted by Cyclops 

Fan A produces 40 DBs and can keep the delta temperature at 50 C. Fan B produces the same amount of noise, but it's delta temeprature is 45 C. Therefore, Fan A's noise to temperature rating is 9 ( [40+50] /10) and Fan B's rating is 8.5. We can conclude that fan B is performing better than Fan A because it's getting a smaller/lower rating (Lower is better).

The bold part seems to be the flaw. If noise is constant, then you can compare the fans directly. But with the noise being different for each fan, I believe you have to account for the perceived noise level (+10dB is a doubling in noise). The fans must be compared "apples to apples".

For example:

Fan A @50dB: 2500RPM, 8*C ΔT

Fan B @29dB*: 1250RPM, 20*C ΔT

*Measured 50cm from the intake rather than the standard 1m

If Fan B was measured at the same distance as Fan A, it would be 6dB louder (when the reflection sound to nearby walls is ignored).

So now it's:

Fan A @50dB: 2500RPM, 8*C ΔT

Fan B @35dB: 1250RPM, 20*C ΔT

Based on your rating (after adding 40*C to ΔT):

Fan A: 9.8

Fan B: 9.5

Fan B is better. Let's see what happens when we double the RPM of Fan B to match that of Fan A:

Fan A @50dB: 2500RPM, 8*C ΔT

Fan B @50dB: 2500RPM, 10*C ΔT

Based on your rating (after adding 40*C to ΔT):

Fan A: 9.8

Fan B: 10

Fan A is better. How can this be !? Well it seems that the lower the RPM, the more of an advantage the fan has in these zero pressure drop scenarios. Once you add some ventilating resistance, I suspect there will be a certain RPM for the heat sink you are using that will give the fans an unfair advantage when compared using your rating. And that doesn't even account for the fact that dB is not linear.

You may have already known that fans have a "sweet spot" in terms of pressure drop, but in case you didn't here is a graph that shows it.

So can you please stop rating the fans this way? And make sure any replacement equation you use makes sense.

Hopefully I didn't make any mistakes in the above. If I did, feel free to point them out with evidence to back it up.

Edit: Make sure you read the changes in the latest edit.

Quote:

Originally Posted by Art Vanelay 

Quote:

Originally Posted by nawon72 

Fan A @50dB: 2500RPM, 8*C ΔT
Fan B @50dB: 2500RPM, 10*C ΔT

Note: I first removed the +40*C from earlier.

Based on your rating (added 40*C to ΔT yet again):
Fan A: 9.8
Fan B: 10

Why did the noise from fan b change to 50dB?

This is why:

Quote:

Originally Posted by dhughesuk 

This is intense, I love it and kudos to the hard work gone into everything on here. From certain charts, and please correct me if I'm wrong, it looks like Push/Pull configurations have an indistiguishable gain and sometimes no gain at all at a very significant doubling of power. Push/Pull configs look good but I'm sure I've seen in other places that it does make a difference. If I am reading it right and your tests are showing that push/pull makes very little difference, could this be down to the setup of your test rig? can the nature of the rig affect this?

Fans in series increases the static pressure, so it is most helpful when there is a lot of ventilating resistance. See this diagram. So yes, "the nature of the rig" can "affect this".