PSU Fan overspeeding for no reason on a brand new Seasonic PSU

[Bytorane]

So I recently got a brand new PSU, the Seasonic Vertex GX 1000W, sometimes I have long sessions, this one is 26 days and 6 hours, usually Windows gets buggy and the desktop starts showing erratic behavior, browsers would crash, if don't reboot in 2 months, but none of that yet. Today I came back after a few hours to notice unusual fan noise, computer idling, temps normal, voltages normal, all fans normal, it seems to be the PSU fan going "berserk", noticably louder ... I had some chores to do so I delayed the investigation for 3-4 hours, I come back again and just like 15 minutes ago it's even louder now.

Some kind of a PSU malfunction, or at least PSU fan controller going bad?

 

[Vya Domus]

Hard to tell why it's doing that, the fan speed is based off some temperature probe or maybe something measuring the amperage, so possibly the sensor is defective. If it bothers you return it while you can.

 

[eidairaman1]

VERTEX GX

Seasonic VERTEX GX PSU combines gold-rated efficiency with robust performance for a superior computing experience. Ideal for gamers and professionals.

seasonic.com

Contact Seasonic on their hysteresis of the hybrid fan controller, or get an exchange.

 

[Eskimonster]

That PSU should get a ticket. RMA ticket that is.

 

[dirtyferret]

Some kind of a PSU malfunction, or at least PSU fan controller going bad?

could be a bad fan controller as the fan should not be full tilt when the PSU is idle unless you have the PC in a hot box. I have zero tolerance for PSU that don't work perfect once you get them outside the box. Return to the seller as a non-working unit and get an exchange (or refund).

 

[Bytorane]

It's too loud that I can't sleep, I have to turn off the computer. I'll try if rebooting helps first.

It's not bearable even so, too distracting to be able to work ... perhaps if I was playing a game ... maybe, but nah.

It's as loud as those Crysis sessions I had back in ~2008-2010 when I used custom tweaks and mods to max the heck out of the graphics settings, with those old hot 40nm GPUs.

 

[Shrek]

Even in sleep mode it is loud?

 

[Bytorane]

Even in sleep mode it is loud?

Haha, I can't sleep! I was suppose to go to bed an hour ago and I had to close down 50 tasks and work stuff, took me a while.

UPDATE: A hard shutdown power cycle fixed it for now. Plain reboot did (expectedly) not make any effect tho.

Maybe I hit some kind of a bug in the fan controller/sensor logic, after 26 days and X hours and Y minutes ... hmm

We'll see how long it lasts, but first I need some sleep.

EDIT: Not hard shutdown - and I did sort of expect a reboot wouldn't have any effect, since it doesn't really cut power the same way as a shutdown.

 

[ty_ger]

I don't know the details of their solution, but microcontrollers do crash sometimes. Or it could be a duration-based overflow that wasn't considered and handled properly when they designed the fan control firmware logic. Hard to say. If it happens repeatedly at random, I would RMA it. If it happens at a consistent x-day period, I would notify them of the bug and ask for an exchange for corrected version or a refund.

 

[Bill_Bright]

fan speed is based off some temperature probe or maybe something measuring the amperage,

For sure, speed is (or should be) controlled by temperature. It would not be amperage. Loads on PSUs are constantly changing so if by amperage, the fan speed would constantly change too, even when temps are still nice and cool - and that would drive users crazy. At least it would drive me crazy.

So either the probe is faulty (they are very cheap, low-tech devices) or the intelligence (circuit or code) used to interpret the sensor's reading and control the fan speed is faulty.

Either way, I agree with the suggestions to contact Seasonic for an RMA.

UPDATE: A hard shutdown power cycle fixed it for now. Plain reboot did not make any effect tho.

If by "hard" you mean you completely removed mains power by unplugging the PSU from the wall or you flipped the master power switch on back of the PSU, then that makes some sense. A simple reboot does NOT totally remove all voltage as the ATX Form Factor standard requires all ATX PSUs maintain +5Vsb standby power output, unless unplugged or that switch is set to off. A "hard" ("cold") reboot removes all power and thus ensures any [possibly corrupt - or stuck] setting is reset.

if don't reboot in 2 months

FTR, I never reboot unless some update or maintenance issue requires it. I do sleep, however. So when I am done computing for the day, I just let my computers go to sleep. I will typically keep my email handler running and if I am working on a Word document or Excel spreadsheet, I typically leave them open too. I do, however, exit all browsing sessions before stepping away from the computer for the day.

While it is common for a Windows or security update to require a reboot, it is not uncommon for me to go 1 or 2 months without rebooting too. And that has not been a problem with any of my W10 or W11 systems here.

 

[Bytorane]

If by "hard" you mean you completely removed mains power by unplugging the PSU from the wall or you flipped the master power switch on back of the PSU, then that makes some sense. A simple reboot does NOT totally remove all voltage as the ATX Form Factor standard requires all ATX PSUs maintain +5Vsb standby power output, unless unplugged or that switch is set to off. A "hard" ("cold") reboot removes all power and thus ensures any [possibly corrupt - or stuck] setting is reset.

I'm sorry, I am familiar with the terminology, I was in a hurry. I did not mean a hard/cold shutdown, but a plain shutdown soft-off (S5) - so it was kept plugged in, without doing any draining of capacitors.

I would have tried a cold hard shutdown if that didn't work tho, but I didn't have to.

 

[freeagent]

Send it back, shouldn’t hear it much if at all..

 

[natr0n]

Look in bios for psu idle (it shows full/auto/idle) throttle setting under power options. might help

 

[freeagent]

My system is doing 550w right now and all I hear are my system fans, I have never heard my Vertex at all..

 

[Bytorane]

The marketplace where I bought it doesn't have any more of these models in stock, so it's probably going to take some time, I'll ask, but also it's back to normal for now, so I can't prove it anymore, I unfortunately forgot to take a video and audio of it (but it probably wouldn't be necessary to start RMA).

It's definitely something to note to at least report to the manufacturer, could be a serial issue, affecting a certain batch for example.

 

[Bill_Bright]

so it was kept plugged in, without doing any draining of capacitors.

I was not talking about "draining of capacitors". That is an old, too often suggested and rarely ever needed practice these days since all PSUs are required by almost countless US, UK, EU and other international electronic and fire safely laws and regulations to bleed off residual voltages within just a few seconds of power removal. The ONLY time draining of capacitors might be needed is with a seriously damaged PSU. Otherwise, all residual voltages will quickly decay to safe levels so no manual bleeding is needed.

The +5Vsb standby voltage I was talking about is an actual output voltage ATX power supplies are required by the ATX standards to supply to multiple points on the motherboard, including various I/O ports (USB and network), the front panel power button and even our system RAM to keep data alive while the system is simply shutdown, the power cord is connected to the wall, and the master power switch is set to "on" or "|". Again, nothing to do with capacitors.

By doing a "cold" reboot (unplug from wall or set master power switch to "off" or "0", you remove all incoming AC voltage which quickly terminates all outputted DC voltages, thus releasing any potential settings that may be holding some erroneous or corrupt value in place.

 

[Solaris17]

How are you monitoring everything? This same exact phenomenon happens to my 4090 (liquid x). I use Aida64 to track everything every once in awhile I will wake it from sleep or something and my GPU temp starts to creep. It’s really not, everything is cool to the touch but it thinks it is. The fans will ramp and go wild and nothing will stop it but a shutdown.

Iv always chalked it up to a sensor conflict on the i2c bus because it is definitely something to do with Aida and my specific system.

 

[dirtyferret]

The marketplace where I bought it doesn't have any more of these models in stock, so it's probably going to take some time

Why keep a faulty PSU when you can send it back for another brand? I understand if it's past the send back date to the seller and you have to RMA the unit which means you out a PSU unless you have a spare as that's a different story.

 

[Bytorane]

Why keep a faulty PSU when you can send it back for another brand? I understand if it's past the send back date to the seller and you have to RMA the unit which means you out a PSU unless you have a spare as that's a different story.

I'm just tired of swapping so many PSUs so many times for in the past month, hehe. I still have the old one at hand so I won't be affected by a RMA that much, but I'd still rather wait until the seller can give me a good ETA that doesn't last for like a month or what.