Cooling Heatsink fan replacement and static pressure

[Sandez]

Hey guys,

I'm needing to get a new fan to replace the GPU reference style fan on my GTX 970 as it's just making way too much noise and it's still throttling while gaming.

I've pulled the shroud off and I'm currently using a Noctua A9x14 at constant full speed but it's still throttling the GPU due to heat. I haven't installed it properly yet so I'm sure I'll get a little bit of improvement when I do, but I need to get a new fan regardless as the A9x14 will be going on the LP53 when that arrives. The fan is to just get me by until I get a new GPU when the next series of GPU's come along. I need to keep to the slim fans as 25mm most likely won't fit in my case.

I'm looking at these 2 Noctua fans:
Noctua A9x14 - 50.5cfm, 1.64 mm H2O
Noctua A12x15 - 94.2cfm, 1.53 mm H2O

My question is, will the slightly less static pressure but nearly twice the CFM be better for cooling the GPU or is static pressure the be all and end all for heatsink cooling?

 

[tinyitx]

The difference in static pressure (0.11 mm H2O) is not likely to make an impact. But almost doubling the airflow might help a bit.
You can try but it is really hard to say if the improvement is substantial. My reasoning is that pumping air at 50cfm is already a really high rate and since overheating still happens, perhaps this is a compound problem and there are other issues.

What I mean is, a typical reference style blower, IIRC, blows at about 10cfm and so 50cfm is five times the rate already. And this does not solve the problem. So, could it be something else?
(Granted that a reference blower has a much higher static pressure (easily 5 times more that your axial Noctua).)

Also, judging the arrangement of fins of the heatsink, it looks like it is not designed for air to blow directly down on it (by an axial fan) but by air being blown across (by a blower style fan). So, this factor might account for a bit.

Finally, even with a reference blower, overheating/throttling during gaming should not happen that easily. This leads me to ask you to look into 2 other factors:- bad airflow inside your chassis (especially hot air re-circulation) and thermal paste dried up of your 970. I cannot say anything further regarding the former and, for the latter, replacing with new thermal paste is a fairly simple procedure.

 

[GuilleAcoustic]

For heatsink, static pressure is the key. I'd go for the NF-F12 PWM that achieves 2,61 mm H₂O (1,83 mm H₂O with L.N.A adapter). If you do not like Noctua's colour scheme, it is also available in their black Chromax lineup.

In the 90x25 size, you have the NF-A9 PWM 2,28 mm H₂O / 1,53 mm H₂O (L.N.A)

 

[VegetableStu]

how about situations where that 140mm fan is used with a 120mm heatsink via some sort of funnel? (that was surprisingly difficult to google actually) o_o

 

[tinyitx]

https://www.frozencpu.com/products/...120mm_Fan_Adapter_-_Black_BP-FA140120-BK.html

This?

 

[VegetableStu]

yeah, that. I mean like how much airflow/pressure would be lost and all those

 

[Sandez]

The difference in static pressure (0.11 mm H2O) is not likely to make an impact. But almost doubling the airflow might help a bit.
You can try but it is really hard to say if the improvement is substantial. My reasoning is that pumping air at 50cfm is already a really high rate and since overheating still happens, perhaps this is a compound problem and there are other issues.

...

Finally, even with a reference blower, overheating/throttling during gaming should not happen that easily. This leads me to ask you to look into 2 other factors:- bad airflow inside your chassis (especially hot air re-circulation) and thermal paste dried up of your 970. I cannot say anything further regarding the former and, for the latter, replacing with new thermal paste is a fairly simple procedure.

Thanks tinyitx, it's kind of what I was thinking as well. I think I will go with the 120mm as I think it will give me slightly more coverage and if it sticks out past the heatsink, then a little extra airflow to hopefully remove more hot air can't hurt either.

Airflow wise, it's happened in 2 cases (albeit both small cases) even with the stock blower exhausting all the hot air, so while it probably has some impact now while using the axial, I don't think it's the main culprit. As for replacing the thermal paste, that is definitely something that I'm going to do. If the reference cooler can't control the heat then there must be something else amiss.

For heatsink, static pressure is the key. I'd go for the NF-F12 PWM that achieves 2,61 mm H₂O (1,83 mm H₂O with L.N.A adapter). If you do not like Noctua's colour scheme, it is also available in their black Chromax lineup.

In the 90x25 size, you have the NF-A9 PWM 2,28 mm H₂O / 1,53 mm H₂O (L.N.A)

I would love to be able to use a full sized fan but I've measured it today and it's only 24mm between the case and heatsink, leaving no room for the fan to pull in air even if I could get it to fit.

 

[eiselx86]

What RPM on the case fans? Is the VRM's cooled properly? Static pressure fans are the best from what I've read. Use HWiNFO and check the GPU sensors and what part is the hottest.

I have a hot gtx 1070 Katana and my conclusion is that pushing too much air towards it does not help. But giving it a dedicated fan at low speed does. Also negative airflow works good. Pushing the hot air out thru the back/top helps both CPU/GPU and hinders warm air to get trapped inside the case.

Intake: airflow for cool air
Exhaust: static pressure to direct air

 

[Sandez]

What RPM on the case fans? Is the VRM's cooled properly? Static pressure fans are the best from what I've read. Use HWiNFO and check the GPU sensors and what part is the hottest.

The main question was more or less to determine if the additional 0.11mm H2O was going to be more beneficial than an additional 43.7cfm. While I don't know if its strictly the better cfm winning in my case, but the A12 seems to be performing better. I think it's more likely due to the additional coverage of going from 92mm to 120mm.

I have a hot gtx 1070 Katana and my conclusion is that pushing too much air towards it does not help. But giving it a dedicated fan at low speed does. Also negative airflow works good. Pushing the hot air out thru the back/top helps both CPU/GPU and hinders warm air to get trapped inside the case.

Intake: airflow for cool air
Exhaust: static pressure to direct air

Just to clarify, I checked the numbers again and found the GPU was never 'throttling' from the stock clock speeds, but it always sat on 80oC when gaming. What I thought was the throttling was actually the card not 'over boosting' the clockspeeds due to the 80oC thermal threshold.

I had it in a Silverstone SG07 with 180mm top fan as intake, GPU was using reference blower, and both with case sides on and off, and still had the same issue. I've since moved the computer to a new custom case (similar to the Sentry) with no case fans. I'm currently running it without any side panels while testing.

I've since purchased the Noctua A12x15 and a PWM to GPU power adaptor so it can be controlled by the GPU. I've only done some short testing so far and it seems pretty positive as its reaching about 75oC in games while in 'over boost'. I still want to remove the heatsink and reapply the thermal paste to hopefully get some better temps, because when I put the case back together the temps will likely go up a little.

 

[eiselx86]

@Sandez
It would likely not be worth it nor make a difference with another fan. I agree, change the thermal paste and maybe new thermal pads on the VRM's.
GL