Excessive PSU noise

[Alec19]

Hey everyone,

 

I recently bought an EVGA 750bq as I needed a beefier power supply for the new RTX 2080 that I got (and it was on a big sale). After replacing my noisy CPU fans in an attempt to make my system quieter, I noticed that there was one remaining fan that was still quite loud. It appears to be my power supply as I still hear it when I turn all of the other fans off. Is there any way to make this quieter. I am not really interested in replacing it anytime soon. It is usually loud all of the time including at idle so I am not sure what the problem is.

 

For reference,

My case is a Thermaltake v200 RGB. It has pretty large feet and I have it on my desk right next to me so there is nothing obstructing the PSU intake at the bottom of the case. There is a small mesh filter but it really isn't fine enough to prevent airflow...or dust.

[Bombastinator]

14 minutes ago, Alec19 said:

Hey everyone,

 

I recently bought an EVGA 750bq as I needed a beefier power supply for the new RTX 2080 that I got (and it was on a big sale). After replacing my noisy CPU fans in an attempt to make my system quieter, I noticed that there was one remaining fan that was still quite loud. It appears to be my power supply as I still hear it when I turn all of the other fans off. Is there any way to make this quieter. I am not really interested in replacing it anytime soon. It is usually loud all of the time including at idle so I am not sure what the problem is.

 

For reference,

My case is a Thermaltake v200 RGB. It has pretty large feet and I have it on my desk right next to me so there is nothing obstructing the PSU intake at the bottom of the case. There is a small mesh filter but it really isn't fine enough to prevent airflow...or dust.

No joy I suspect.  Changing the noise profile of a fan means messing with a fan. And PSUs aren’t safe things to open. The way atx works is if the PSU has a heat fan curve it’s totally internal. Could maybe hit it with a vacuum to see if you could suck anything out of the grate. I personally wouldnt even open one to remove a foreign object that fell into the fan.  I have opened PSUs, but only after they had sat around unpowered for months. There’s some awful big capacitors in those things. I did learn that at least the model I had did NOT have a standard fan header on its fan port so there wasn’t any simple replacing of the fan, which might be a bad idea anyway. 

[Guest]

26 minutes ago, Alec19 said:

Hey everyone,

 

I recently bought an EVGA 750bq as I needed a beefier power supply for the new RTX 2080 that I got (and it was on a big sale). After replacing my noisy CPU fans in an attempt to make my system quieter, I noticed that there was one remaining fan that was still quite loud. It appears to be my power supply as I still hear it when I turn all of the other fans off. Is there any way to make this quieter. I am not really interested in replacing it anytime soon. It is usually loud all of the time including at idle so I am not sure what the problem is.

 

For reference,

My case is a Thermaltake v200 RGB. It has pretty large feet and I have it on my desk right next to me so there is nothing obstructing the PSU intake at the bottom of the case. There is a small mesh filter but it really isn't fine enough to prevent airflow...or dust.

How much RGB do you have in your build?  If you turn off all of your RGB and let the PC cool down, does the PSU fan slow back down?

 

[Bombastinator]

4 minutes ago, jonnyGURU said:

How much RGB do you have in your build?  If you turn off all of your RGB and let the PC cool down, does the PSU fan slow back down?

 

Oooh. Straight heat.  Worth checking.  Would require an inverted PSU case though, with the PSU sucking air out of a hot case.  If the PSU airflow is segregated the only thing that could force it is some sort of really heavy draw so the PSU itself is working hard.

Edited May 15, 2021 by Bombastinator

[Alec19]

4 minutes ago, jonnyGURU said:

How much RGB do you have in your build?  If you turn off all of your RGB and let the PC cool down, does the PSU fan slow back down?

The only RGB in the system is on the motherboard. I have a Gigabyte Z490 Aorus Master which just has a couple of accent lights on the logos. Turning these off did not have any effect.

[Bombastinator]

4 minutes ago, Alec19 said:

The only RGB in the system is on the motherboard. I have a Gigabyte Z490 Aorus Master which just has a couple of accent lights on the logos. Turning these off did not have any effect.

LEDs draw very little power.  Would need a whole lot of them to make any difference for power draw.  Multiple light strips or some old school fluorescents or something.  Looking at the case it actually has a PSU basement so not an inverted case design unless you inverted it.  Cases with PSU basements usually have a dedicated filter for the PSU on the bottom of the case.  Putting one on, say, shag carpet or having a “wow that’s dirty” air filter could obstruct PSU airflow

[Guest]

Ok.  So no RGB fans or anything?  Hmm..

 

Yeah.  I always ask because of this:  

 

https://forum.corsair.com/v3/showthread.php?t=196372

 

Yeah.  Might just be a really loud PSU.  If you look at the Cybenetics report for it, it has the worse possible noise rating. 

 

[Alec19]

9 minutes ago, Bombastinator said:

LEDs draw very little power.  Would need a whole lot of them to make any difference for power draw.  Multiple light strips or some old school fluorescents or something.

The system is drawing 70 Watts at idle which seems a little high to me but is still pretty low compared to its maximum output.

8 minutes ago, jonnyGURU said:

Yeah.  Might just be a really loud PSU.  If you look at the Cybenetics report for it, it has the worse possible noise rating.

That's not great news. I originally bought it because it was cheap and I needed something quick. I suppose I should have done more research. Either way, it was much quieter when I originally put it into my system. It was silent for a couple of days after I did that fan upgrade.

[Guest]

4 minutes ago, Alec19 said:

The system is drawing 70 Watts at idle which seems a little high to me but is still pretty low compared to its maximum output.

That's not great news. I originally bought it because it was cheap and I needed something quick. I suppose I should have done more research. Either way, it was much quieter when I originally put it into my system. It was silent for a couple of days after I did that fan upgrade.

Ooh!  Here's an idea!

 

I learned about this problem way back with the Chicony made Corsair RM series....  Mount the PSU in there so the fan is pointing up, not down.  If they're using PSU 101 design, the thermistor is on the PCB near the SR.  If you mount the PSU fan up, the heat will rise away from the thermistor and the fan should run slower.

 

[Alec19]

1 minute ago, jonnyGURU said:

If you mount the PSU fan up, the heat will rise away from the thermistor and the fan should run slower.

That makes sense. I will get working on that in a couple of hours. In the meantime, I looked into that 70 Watts of power draw at idle and found that 65 of those Watts are coming out of my GPU at 1% usage. That seems very high to me although I am no expert when it comes to graphics cards. (I am in the balanced power plan and the GPU is a EVGA 2080 Black)

[Alec19]

I just went through the process of flipping my power supply. After booting up, the power supply was completely silent. However, when I started to put a load on the power supply, the fans started up as loud as usual.

 

I also fixed that really high idle power on the GPU. I had selected the prefer maximum performance option for the power management mode in the nvidia settings.

[Spotty]

24 minutes ago, Alec19 said:

I just went through the process of flipping my power supply. After booting up, the power supply was completely silent. However, when I started to put a load on the power supply, the fans started up as loud as usual.

I would say that it's just a loud PSU. Like Jon mentioned it received only a Standard rating on Cybenetics noise testing, the lowest they offer.

Cybenetics noise testing shows it starting at 1500RPM and 41dBA at 10% load, which is pretty loud for a power supply.

http://members.cybenetics.report:5050/d/cybenetics_OPr.pdf

 

Also apparently failed ripple tests for 5V and 3.3V, which I never noticed before.

 

Funnily, if you look at the product page on EVGA's website for the EVGA BQ 750W, EVGA's marketing describe it as "Ultra-Quiet" and "Quiet and Intelligent Auto Fan for near-silent operation"

Spoiler

EVGA's team testing the BQ 750W

 

[Alec19]

19 minutes ago, Spotty said:

Funnily, if you look at the product page on EVGA's website for the EVGA BQ 750W, EVGA's marketing describe it as "Ultra-Quiet" and "Quiet and Intelligent Auto Fan for near-silent operation"

I saw that when I looked back to their website in an attempt to find their noise rating. I should have checked the Cybenetics website before buying it. The EVGA marketing team seems very proud of that "Teflon Nano-Steel fan".

 

27 minutes ago, Spotty said:

Also apparently failed ripple tests for 5V and 3.3V, which I never noticed before.

Yikes. My electronics knowledge is not exactly super strong, but I know that isn't good.

 

I am still within warranty, although out of NewEgg return policy, I went to EVGA support to see if they would allow me to trade up to a better model and pay the difference. Or even just give me a refund. I really do not know if that is something that they do. If trading up is an option, what is a good EVGA power supply that is quiet and has two EPS connectors?

[Bombastinator]

9 hours ago, jonnyGURU said:

Ooh!  Here's an idea!

 

I learned about this problem way back with the Chicony made Corsair RM series....  Mount the PSU in there so the fan is pointing up, not down.  If they're using PSU 101 design, the thermistor is on the PCB near the SR.  If you mount the PSU fan up, the heat will rise away from the thermistor and the fan should run slower.

 

70w?  That’s barely more than an incandescent table lamp. A heater is 1500w

[Bombastinator]

1 hour ago, Alec19 said:

I saw that when I looked back to their website in an attempt to find their noise rating. I should have checked the Cybenetics website before buying it. The EVGA marketing team seems very proud of that "Teflon Nano-Steel fan".

 

Yikes. My electronics knowledge is not exactly super strong, but I know that isn't good.

 

I am still within warranty, although out of NewEgg return policy, I went to EVGA support to see if they would allow me to trade up to a better model and pay the difference. Or even just give me a refund. I really do not know if that is something that they do. If trading up is an option, what is a good EVGA power supply that is quiet and has two EPS connectors?

“Teflon nano-steel?!” It’s a fricken fan. Teflon is a really soft plastic. It’s nonstick because it’s unreasonably flat and has some unusual valence bonds. Other than that it’s just not very strong plastic though.  “Nano-steel” could be steel with an extra small crystal structure I suppose, which is better known as “cheap steel” the expensive stuff has really big crystals that run in particular directions. They do wacky stuff like spray it onto supercooled pipes and stuff.  Turbofan blades were single crystal steel at one point.  some marketing guy pulled real hard on that one.  He may have made the bells ring. 

[Elisis]

1 hour ago, Bombastinator said:

Teflon is a really soft plastic

Uh... Its weathering resistance is actually quite large, so that's not entirely true. 

 

1 hour ago, Bombastinator said:

It’s nonstick because it’s unreasonably flat and has some unusual valence bonds.

It's nonstick because of the fluorine atoms on both sides that repel literally everything. It's no more flat than polyethylene, PVC, or most other polymers either.

[Guest]

6 hours ago, Bombastinator said:

70w?  That’s barely more than an incandescent table lamp. A heater is 1500w

Doesn't matter if the thermistor is located within a mm of the heat source. 

 

Just sounds like a very bad design.  Later today, guess I'll have to closely study a tear down so I can learn from their mistakes. 

[Guest]

Ok.  Couldn't find a 750W tear down but I did find Aris's 850W tear down.

 

I'm going to do "weird flex" now.  But it's Sunday and I'm in a good mood....

 

Nothing really screams out at me as terrible.  Only thing that jumped out at me as a questionable choice was the main switchers.  Lower voltage and current and higher RDS(ON) then I would want to use.  So maybe they made the same mistake with the 750W.  I'm not too familiar with MagnaChip other than they're Korean.  Maybe there's a big price jump to using a 600V, 12A+ MOSFETs.  The other thing is it's HEC.  LOL!  I mean.. they're a good OEM but they NEVER use an MCU for fan control.  Getting them to tweak the fan curve on the CX-F using their rocks & sticks analog control was infuriating and took way longer than it needed to.  So here, it could be that the engineer was aware of the potential for thermal runaway on those PWM FETs and set a really aggressive fan curve to make sure it didn't happen. I mean, the heatsink looks big enough and it looks like the MOSFETs are TO-220 with thermal insulators, so it's just a matter of keeping that heatsink cool at all times.

[IIIIIIIIII]

25 minutes ago, jonnyGURU said:

The other thing is it's HEC.  LOL!  I mean.. they're a good OEM but they NEVER use an MCU for fan control. 

Most of the OEMs don't? I don't think I've ever seen use of an MCU for a fan controller except for Sirfa, CWT, and SuperFlower.

[Guest]

10 minutes ago, IIIIIIIIII said:

Most of the OEMs don't? I don't think I've ever seen use of an MCU for a fan controller except for Sirfa, CWT, and SuperFlower.

That's quite a few different OEMs, isn't it?  CWT alone makes about 70% of the PSUs out there. 

 

Like I said, it's not IMPOSSIBLE to make a PSU with an analog fan controller quiet.  It's just a pain in the ass and drags things out.  If you're like EVGA and don't have engineers, your own testing labs, etc., you just let the OEM do "whatever".

 

Analog ICs are a pain because you can't actually program the IC itself.  You change the values by swapping out resistors until you get what you want.  It's such a pain.

 

MCUs are sooooo much easier.  Take the current RM850x, for example.  After Aris pointed out it was louder than the previous and even louder than the 1000W, fixing it was a matter of just hooking a bunch of thermistors up to all of the internal components and tweaking the fan curve on the fly by changing lines of code.  The part that takes the longest is booking and getting reports back from an acoustic lab so we can show that the reduction in RPM actually makes an impact on the acoustic noise.  And then.... you have to wait for super busy Aris to test it so he can publish the numbers for you all to see!  LOL!

 

I wish all PSUs used MCUs for everything.  But it's "too expensive" for our cheap customers.  And now we have a HUGE MCU shortage.  Can't get anything from NXP, Atmel, etc.  They're all like 9 month lead time!  But when you replace PWM controllers and supervisor ICs with MCUs.. OH MY GOD!  You find something wrong during testing, you just add or change some lines of code and try again. 

 

 

[Bombastinator]

7 hours ago, Elisis said:

Uh... Its weathering resistance is actually quite large, so that's not entirely true. 

 

It's nonstick because of the fluorine atoms on both sides that repel literally everything. It's no more flat than polyethylene, PVC, or most other polymers either.

It was developed because lubricants are often near useless in vacuum.  Teflon can last longer than traditional lubricants there. Still not real long mindyou, but a lot longer. It’s still pretty soft.  It was a 1960’s wonder material but the 60’s were a long time ago.  Still pretty soft as plastics go.  Just a lot harder than mineral oil.  Weathering resistance and hardness are not the same thing. “Nano” has become a buzzword without meaning.  Maybe that one should go in that awful tech terms list.

[IIIIIIIIII]

22 minutes ago, jonnyGURU said:

Like I said, it's not IMPOSSIBLE to make a PSU with an analog fan controller quiet.  It's just a pain in the ass and drags things out.  If you're like EVGA and don't have engineers, your own testing labs, etc., you just let the OEM do "whatever".

Is it possible even if the semi-fanless mode is included? I've seen a lot of users complaining about the fan flickering badly on some PSUs.

[Guest]

Anyhoo.... "Teflon Nano-Steel Bearing" just means they're using a sintered steel bearing rotating in a teflon sleeve.  Doesn't matter what the bearing is if it's not properly controlled. 

 

[Juular]

1 hour ago, jonnyGURU said:

Ok.  Couldn't find a 750W tear down but I did find Aris's 850W tear down.

650/750W are completely different, still HEC tho. Jeremy did a review back then, but the site is dead. I found these tho :

https://coolenjoy.net/bbs/review/580610?p=4

https://www.ixbt.com/supply/evga-650-bq-review.html#n4

Looks pretty much like a CM MasterWatt / Corsair CV650.

[Guest]

3 minutes ago, IIIIIIIIII said:

Is it possible even if the semi-fanless mode is included? I've seen a lot of users complaining about the fan flickering badly on some PSUs.

That's actually when Corsair started using MCUs.  You have to be able to program a hysteresis into the fan curve so the fan doesn't "twitch" when temperatures are in between "fan off" and "fan on".  You can't do that with an analog controller.  Well... you can, but then you might as well use an MCU.