Interesting. I can see the button getting pushed and popping out when being handled or poking about inside the case and unintentionally setting it up to fanless mode.
I'd prefer a slide switch or just let the PS electronics decide whether the fan is needed or not.
Gives a marketing-style definition of the two modes. My takeaway is that the two curves are the same above the transition point where the optionally fanless mode decides to turn on the fan.
I've looked at a number of Seasonic and enthusiast forum postings on this button and found no indication that it governs any other functional difference than the fan speed difference sketched on the web page I've referenced.
Because Gary Roberts operates a huge fleet of machines, is attentive to their failure modes, runs them a long time, and has bought a large number of bargain-basement supplies far below the majesty of the supplies approved by participants in this thread, I think he has seen and diagnosed a large number of power supply failures. I'd be curious, if he stops by to comment, in hearing what fraction of his failures were fan stoppages. Also any other identified failure types.
Yes, Gary's input would be valuable since he runs such a huge fleet. The large sample size would provide more weight than for someone only running a few hosts. Even if they have run them for years.
I've only ever needed to replace a power supply because they get fatigued over time (8-10 years) and the +12V rail falls below spec under my large ~1000W loads and I start getting intermittent task failures.
1. Not every PSU has peak efficiency at 50% load. That’s a broad generalization. In reality it varies from model to model. And some hold the efficiency ratings even at very high load percentages.
2. even models that have semi-passive modes, the fans will spin if the unit is at certain loads or temperatures. So if you are always over that limit you might not be aware that the mode exists unless you read it in the documentation.
3. San-Fernando-Valley has several high power multi-GPU systems (with many Titan Xs and Titan Vs, obfuscated by a single GTX1650) which can likely use even a 1600W adequately. I actually use a single 1600W to power my 7xRTX2080 system. Each power limited to 185W to keep the system under the limits. It’s an EVGA 1600 Platinum unit, that I bought used from a mining farm. Been running for about 3 years near continuously by me and unknown how long before that.
platinum and titanium models are certainly worth the cost for a 24/7 rig. Not only are there energy savings that payoff over time, the platinum and titanium units are almost almost built to higher quality standards in order to meet the specs. They last longer as a result.
I (as an amateur) would want to believe that all the well known manufacturers have found very good curves for the semi-passive mode these days. Maybe there were more problems a few years back. I remember reading that the first Corsair RM-series PSUs run too mild curve and got too hot until fan started spinning. Then those PSUs were failuring too soon. I had one of those old RM750's and not even in 24/7 crunching. It broke down too, though I can't know the real reason, but the case of that PSU felt often surprisingly hot when the fan still wasn't spinning. And it was in open computer case. To my understandind Corsair then changed the fan curve for the upgraded RM series models to start the fan spinning at lower temperature.
Oh, the orientation of the PSU naturally must be linked with that passive operating temperature too. Mine had the fan opening facing up. I thought that was good for its health, but maybe that wasn't enough over time. Many computer cases will suggest a PSU to be installed upside down, at the bottom of the computer case. Then I'd assume that heat would build up on the "bottom" of the PSU and circuit board perhaps even more. Even after the fan would ultimately start spinning.
About bearings on the fans, I see "magnetic levitation" is an option these days. That's interesting. Maybe those fans then woudn't degrade and get noisier over time. I read people have complained though that those ML fans in Corsair PSUs are noisier than regular fans with bearings.
I don't think the noise of the ML fans can be attributed to the bearing design. In my opinion the noise is coming from the "choppy" nature of the fan blades cutting the air. They spin at an aggressive rpm and at 100% fan speed, they really produce an annoying sound signature. Cut the rpm's down to the 70-80% range and they are not as obnoxious. They do move a whole lot of air at full speed though.
OK - So the new EVGA Supernova 220-P5-1000-X1 seems to be working as advertised, ECO mode and all!
WRONG - After a long day of crunching, decided to play a game and that apparently was the last straw, it tripped.
Now I'm returning the EVGA and moving back to Seasonic. Going with the Seasonic 1200 Platinum. If the trips continue then I will have to consider that the new EVGA 3080ti is causing the problem, as I've never had any problem until I installed the 3080TI....
Interesting. I can see the
)
Interesting. I can see the button getting pushed and popping out when being handled or poking about inside the case and unintentionally setting it up to fanless mode.
I'd prefer a slide switch or just let the PS electronics decide whether the fan is needed or not.
Seasonic describes the magic
)
Seasonic describes the magic push button as choosing between S2FC and S3FC fan control modes.
This page:
https://knowledge.seasonic.com/article/29-what-is-seasonic-s3fc-smart-and-silent-fan-control
Gives a marketing-style definition of the two modes. My takeaway is that the two curves are the same above the transition point where the optionally fanless mode decides to turn on the fan.
I've looked at a number of Seasonic and enthusiast forum postings on this button and found no indication that it governs any other functional difference than the fan speed difference sketched on the web page I've referenced.
Because Gary Roberts operates a huge fleet of machines, is attentive to their failure modes, runs them a long time, and has bought a large number of bargain-basement supplies far below the majesty of the supplies approved by participants in this thread, I think he has seen and diagnosed a large number of power supply failures. I'd be curious, if he stops by to comment, in hearing what fraction of his failures were fan stoppages. Also any other identified failure types.
Yes, Gary's input would be
)
Yes, Gary's input would be valuable since he runs such a huge fleet. The large sample size would provide more weight than for someone only running a few hosts. Even if they have run them for years.
I've only ever needed to replace a power supply because they get fatigued over time (8-10 years) and the +12V rail falls below spec under my large ~1000W loads and I start getting intermittent task failures.
1. Not every PSU has peak
)
1. Not every PSU has peak efficiency at 50% load. That’s a broad generalization. In reality it varies from model to model. And some hold the efficiency ratings even at very high load percentages.
2. even models that have semi-passive modes, the fans will spin if the unit is at certain loads or temperatures. So if you are always over that limit you might not be aware that the mode exists unless you read it in the documentation.
3. San-Fernando-Valley has several high power multi-GPU systems (with many Titan Xs and Titan Vs, obfuscated by a single GTX1650) which can likely use even a 1600W adequately. I actually use a single 1600W to power my 7xRTX2080 system. Each power limited to 185W to keep the system under the limits. It’s an EVGA 1600 Platinum unit, that I bought used from a mining farm. Been running for about 3 years near continuously by me and unknown how long before that.
platinum and titanium models are certainly worth the cost for a 24/7 rig. Not only are there energy savings that payoff over time, the platinum and titanium units are almost almost built to higher quality standards in order to meet the specs. They last longer as a result.
_________________________________________________________________________
I (as an amateur) would want
)
I (as an amateur) would want to believe that all the well known manufacturers have found very good curves for the semi-passive mode these days. Maybe there were more problems a few years back. I remember reading that the first Corsair RM-series PSUs run too mild curve and got too hot until fan started spinning. Then those PSUs were failuring too soon. I had one of those old RM750's and not even in 24/7 crunching. It broke down too, though I can't know the real reason, but the case of that PSU felt often surprisingly hot when the fan still wasn't spinning. And it was in open computer case. To my understandind Corsair then changed the fan curve for the upgraded RM series models to start the fan spinning at lower temperature.
Oh, the orientation of the PSU naturally must be linked with that passive operating temperature too. Mine had the fan opening facing up. I thought that was good for its health, but maybe that wasn't enough over time. Many computer cases will suggest a PSU to be installed upside down, at the bottom of the computer case. Then I'd assume that heat would build up on the "bottom" of the PSU and circuit board perhaps even more. Even after the fan would ultimately start spinning.
About bearings on the fans, I see "magnetic levitation" is an option these days. That's interesting. Maybe those fans then woudn't degrade and get noisier over time. I read people have complained though that those ML fans in Corsair PSUs are noisier than regular fans with bearings.
I don't think the noise of
)
I don't think the noise of the ML fans can be attributed to the bearing design. In my opinion the noise is coming from the "choppy" nature of the fan blades cutting the air. They spin at an aggressive rpm and at 100% fan speed, they really produce an annoying sound signature. Cut the rpm's down to the 70-80% range and they are not as obnoxious. They do move a whole lot of air at full speed though.
Nice
)
Nice article;
https://www.anandtech.com/show/2624/debunking-power-supply-myths/3
The article emphasizes my
)
The article emphasizes my point that you will never hear the power supply fan over the noise of the fans cooling the components it is powering.
OK - So the new EVGA
)
OK - So the new EVGA Supernova 220-P5-1000-X1 seems to be working as advertised, ECO mode and all!
Gandolph1 wrote: OK - So the
)
WRONG - After a long day of crunching, decided to play a game and that apparently was the last straw, it tripped.
Now I'm returning the EVGA and moving back to Seasonic. Going with the Seasonic 1200 Platinum. If the trips continue then I will have to consider that the new EVGA 3080ti is causing the problem, as I've never had any problem until I installed the 3080TI....