seasonic psu: fanless mode in the x-series

[giannife]

Hi, in the really interesting review of the seasonic x650 psu:
http://www.silentpcreview.com/Seasonic_X650
I read that:
"On the test bench with ambient room temperature at 23°C, the fan in the Seasonic X-650 started at about 340W load. "
Is this value different for each model (the new ones are X-560-660-760-850) or it is the same (more or less)?
In theory it should be a percentage of the maximum output power, but in practice there should be some differences, for example in x-850 should have bigger heatsink, so the value should be even higher.

The same load should be valid also for the corsair psu:
"Actually, the new Corsair AX750 and 850 are based on the X-760 and 850, not the old models. "
viewtopic.php?p=521972

thank you very much
Gianni

[giannife]

I received an answer from seasonic:

Basically the fan starts to operate at room temperature 25C, and >20% load.
So the fan should starts to run in different W load for different models respectively.
560W*20%=112W
660W*20%=132W
760W*20%=152W
850W*20%=170W
Please note that this can be affected by actual ambient temperature and other factors.

so if you consider that:

On the test bench with ambient room temperature at 23°C, the fan in the Seasonic X-650 started at about 340W load.

340-132= 208 W, so the values for fanless operation (outside of the pc case) should be:
560W =320 W
660W = 340 W
760W =360 W
850W =378 W

do you agree, or do I need to take into account something else?

[MikeC]

Such linear extrapolation is unrealistic. The fan in all PSUs is controlled by temperature, not load. The info from Seasonic (marketing or tech support) is misleading in that it refers to 20% load -- that's not the way the circuit runs. The thermal sensor reaches a certain temperature, at which point, the feed to the fan exceeds the start requirement, and then the fan starts spinning.

So what would affect the thermal sensor's temperature??

Heatsink size -- maybe. It's passive before the fan turns on though. At a given power level, the same amount of heat is generated in all these different models (because they have the same efficiency). So regardless of the size of the heatsinks, the air temperature in the PSU is probably going to be the same... because the cooling effect of convection alone inside the PSU is probably not going to vary much with heatsink size. Different story if there was a fan blowing all the time -- where the bigger heatsinks would definitely produce lower internal temps. The key here is the exact positioning of the sensor -- if it is mounted on a heatsink, and if that heatsink size varies with PSU model, then you might see slightly different start points for the fan.

But I don't think so. I think they will all start the fan up at around the same load. At least until/unless the heatsinks are a LOT bigger than in the X650.

This was the case with the Corsair AX850, which is basically a Seasnoic X850 w/o PWM control for the fan (usues a straight voltage controller) - http://www.silentpcreview.com/Corsair_G ... wer_Supply

[giannife]

Thank you very much for your very competent answer

I'm still not sure whether I should buy the x-660 or the completely fanless x-460: the fact that it doesn't have a fan at all scares me a little bit

[MikeC]

Thank you very much for your very competent answer

I'm still not sure whether I should buy the x-660 or the completely fanless x-460: the fact that it doesn't have a fan at all scares me a little bit

I think you've answered your own question. Why buy a product that will worry you when the alternative will be the same in every way 99% of the time? It's not until the load gets to ~250W in a hot environment or nearly 400W in a cooler one that the PSU will make enough noise for it to become higher than ambient in most rooms.

[Compddd]

Thank you very much for your very competent answer

I'm still not sure whether I should buy the x-660 or the completely fanless x-460: the fact that it doesn't have a fan at all scares me a little bit

I think you've answered your own question. Why buy a product that will worry you when the alternative will be the same in every way 99% of the time? It's not until the load gets to ~250W in a hot environment or nearly 400W in a cooler one that the PSU will make enough noise for it to become higher than ambient in most rooms.

This is the reason I went with the X-760 over the X-460

[Zoiberg]

Any possibility that we will have a review on this new X-series?

Regards

[dhanson865]

I received an answer from seasonic:

Basically the fan starts to operate at room temperature 25C, and >20% load.
So the fan should starts to run in different W load for different models respectively.
560W*20%=112W
660W*20%=132W
760W*20%=152W
850W*20%=170W
Please note that this can be affected by actual ambient temperature and other factors.

so if you consider that:

On the test bench with ambient room temperature at 23°C, the fan in the Seasonic X-650 started at about 340W load.

340-132= 208 W, so the values for fanless operation (outside of the pc case) should be:
560W =320 W
660W = 340 W
760W =360 W
850W =378 W

do you agree, or do I need to take into account something else?

I agree with MikeC on why extrapolation wouldn't be useful here but even if it were I think your math is wrong. You seem to be assuming that you can just add 208W to the numbers provided from seasonic. I'm not sure why you think 208W is a constant.

If I were trying to extrapolate based on the review + seasonic info and your premise of scaling I would do something like

340/650 = 52.3% so

52.3% of 560W is 293W
52.3% of 660W is 345W
52.3% of 760W is 397W
52.3% of 850W is 444W

These numbers are likely useless but at least they assume the percentage would hold true. I'm having a hard time figuring out the logic you used to scale the numbers.

Now if in some hot box they figured out that the worst case scenario turned on the fan at 20% load on a 550, 650 and 750 (which were the wattage on the first generation X series models) and 20% of 750 was 150W we get

150/550 = .27
150/650 = .23
150/750 = .2

and then we could assume that their worst case test for the new series has the fan coming on around 150w for percentages of

150/560 = .26
150/660 = .23
150/760 = .20
150/860 = .17

or you could assume they knew about the higher loads coming out and based it on a test of an unreleased higher wattage model or any number of other assumptions.

I highly doubt each unit will turn the fan on at exactly the same wattage but I also doubt that they'll scale at a set percentage. I'd be curious to know if you had 3 of each newer wattage to test (to average out some sample variance) for a total of 12 PSUs how that curve would map but as a consumer I'm happy to just buy on price and am not going to worry about that particular bit of data.

[giannife]

In my post I tried to combine the marketing fluff and the experimental truth, so I added an hypothetical factor that took in account the passive heatsink - convection dissipation (208w, from the experiment and marketing) and a linear factor (from marketing). That it is a wrong approximation, anyway less wrong than a completely linear scaling (which I was trying to avoid).
Anyway a very interesting starting point for the next, surely compelling and intelligent - as always - reviews.