[SOLVED] A couple of questions on air cooling in sub-10L SFF

[PiffPuff]

Hi guys,

could you please kindly help with the following questions:

1. Which CPU fan curve thresholds will you recommend for CPU cooler of a non-overclocked AMD Ryzen 5600X or 5800X?
2. To create positive pressure intake fan should suck in more air than the exhaust fan blows away - this is clear. However, is it right, that the volume of intake air increases even more with the increase of perforations in the case?
3. Is it important that the intake/exhaust fans are not directly opposite, so that the exhaust fan doesn't eat intake fan's air right away?

Thank you guys!

 

[Eximo]

1. Completely dependent on the cooler and your noise preferences, and your temperature preferences. Obviously a hard limit at 85C or so for the reasonable person. But noise and when you want the fan to go full speed is rather up to you. If you can keep it quite while it is a little warm, that is okay. If you want it cooler, set a more aggressive fan curve.

2. Yes? If the fan is unrestricted, it will have more pressure. If you mean that the rest of the chassis has more perforation, I suppose that could slightly increase the fan speed since there would be less resistance to push air out of the case, but I'm not sure that is beneficial, that would just mean air going out of places all over.

3. Kind of, you want the fresh air to reach your components, and go through them, if possible, but the important thing is to get fresh air into the system and exhaust out the heated air. So if they are in a direct line, not the end of the world.

 

[PiffPuff]

Thank you @Eximo , appreciate your help!

Regarding the CPU fan cooler curve: am I right, that there's only 1 mandatory ("manufacturer recommended") threshold, which is equal for both Intel and AMD and the same for any CPU model, and it it 85C?
As for the other thresholds: the noise level is obviously a judgemental thing to be evaluated when all components are already mounted, but what about the temp levels - are there any recommended ones? I mean I can setup thresholds in equal e.g. 20C steps (85C - 65C - 45C), but shall it be the smart approach? Am I not supposed to somehow adapt these thresholds to my use scenarios and if yes - how?

Finally, regarding dust filters. If the nylon filter manufacturer says it keeps 65% of the airflow (35% is lost), should I automatically decrease (in my mind) each fan's official CFM to 65% when choosing which fan exactly to buy? I mean it is 65% of the overall CFM that I should keep in mind when calculating the air volume needed in general and to have the positive pressure in particular.

 

[Eximo]

It isn't a mandatory threshold at all, if you want to shorten the lifespan of a CPU you can certainly pump voltage through it to the point it runs near TJMax. All silicon based electronics have the same rough limit in actual temperatures. Intel will officially shut off at 105C if I recall, but it will be thermal throttling well before that. 85C is a good maximum to stick to with a desktop, laptops tend to get a little warmer and there isn't much you can do about it besides replacing thermal compound and adding external fans. There have been times when AMD used a completely different measuring standard, but they have since stopped that practice. A lot of power electronics can have limits as high as 120C, any warmer and they start losing the characteristics we value them for.

If you are referring to setting up fan curves, generally want to start out slow and adjust it so that it only really starts to ramp up above 70C or so. Going to be constant spikes in temperature in a CPU that may reach 70C easily, this is normal. But if you set a linear curve you will find the fan is constantly ramping up and down.

Again, that is subjective. CFM is measured at maximum fan speed, which you would almost never want to run at due to noise. If you want to be scientific about it, you would have to measure the airflow in and out and test with and without filters.

 

[PiffPuff]

@Eximo thank you so much dude, you can't imagine how helpful your answers are. May I ask you a couple of last ones and we call it a day:

1. As fans' CFM is calculated at max speed (= noise level) looks like if low noise is crucial, it's better to go for the fan which CFM has a certain margin above the real demand, do you agree?
2. Maybe a dumb question, but the cooling system (be that air or AIO) gets all the information on the current temperature from CPU itself, right? I mean there are no temp detectors built in either CPU cooler or case fans and, therefore, there's no such a thing as "in-case temp", only the CPU temp, is that so?
3. If my assumption #2 is correct, case fans' curve can (and probably should!) be adjusted to CPU temp too. Shall the logic of temp/fan speed thresholds be the same as that CPU fans use (i.e. start slow, increase to fit desired noise levels (but not in a linear manner), start real ramp up after ~70C)? I'm a novice, but I suppose that if case fans are in general more silent compared to CPU cooler (because of bigger size and slower spin) it makes sense to set case fan's temp thresholds a bit lower vs. those for CPU cooler (e.g. 70C for CPU and 65C for case fans). My assumption is that in this scenario case fans will probably prevent warming up from 65C to 70C in a pretty silent way and the CPU cooler's curve (and extra noise) won't trigger at all. Does this logic make sense?

Thank you a lot!

 

[Eximo]

1. Yes, that is reasonable to a degree.
2. Correct, the CPU has multiple temperature sensors, for the total package and each core. The motherboard often also has a CPU socket temperature reading.There are system temperatures, which may be chipsets or VRMs.And on some boards you can add your own temperature sensors
3. That might work. I can think of a few ways to accommodate that. One would be to have the case fans be more aggressive than the CPU fans so that air flow is increased prior to the demand from the CPU. The opposite would be better for noise reduction. Going to be something you have to test, because you might end up in a weird oscillation mode.

 

[PiffPuff]

I can think of a few ways to accommodate that. One would be to have the case fans be more aggressive than the case fans so that air flow is increased prior to the demand from the CPU.

You meant "than the CPU cooler fan", typo - right?

The opposite would be better for noise reduction.

Ehmm.... if my comment above is true, "the opposite" will be "CPU fan more aggressive than case fans", but how can that be better for noise reduction, if case fans are usually bigger and spin at lower speeds? Possibly you meant that CPU cooler fan is hidden inside and for this reason there will be less noise...?

Sorry for this misunderstanding.

 

[Eximo]

Yes, I will correct that.

Size and noise are not the important factor, you are trying to propose a scenario where you are unbalancing the fans by CFM. That means the output of one will be greater than the other.

Also, nothing says the CPU fan has to be smaller than the case fans, only really true due to the form factor you are building in. Plenty of 140mm air coolers, even 150mm.

 

[PiffPuff]

@Eximo Noted, thank you so much for the explanations and patience!