How to reduce server power supply fan noise?

See the bottom of this post for pictures. If you look closely, you can see the power supply fan wires in one of those pictures.

I bought a legacy server with a 560 W power supply. It's dual fans are too loud, apparently running 100% speed all the time. Might be okay for flight simulation immersion, but No.

I suppose the easiest solution would be to put a resistor in series with the power leads.

Another possibility would be to sever all but the sensor leads. And connect the two power supply leads to the motherboard fan controllers. Apparently the motherboard has settings for different fan voltages at the fan outputs, and doesn't care about system temperature.

But the power supply needs to know that the fans are running, or it shuts down the system. I suspect that slowing the fans would produce a fault, just like removing the fans does. Do the sensors look for a fan speed of at least a certain RPM? I guess so, just double checking.

How difficult might it be to determine and satisfy what signals the power supply controller is looking for from the three wire fans?

Ideally, I would like to replace the power supply fan with a system board fan. But those power supply fan connectors need satisfying.

Thanks.

--

https://www.flickr.com/photos/27532210@N04/24781725611/in/dateposted/  

That's the power supply, plus two other pictures. It's supposed to be  
used but I see zero signs of use.  

http://www.ebay.com/itm/371543511693?  

That's where I bought it. If you need something like that, don't  
hesitate. But be advised that I don't sell anything.
Reply to
John Doe
Loading thread data ...

The lowest stress solution is probably just to get a whole new power supply that's quieter and drop it in, unless you're pathologically averse to spending money. ~500 watt server power supplies aren't particularly expensive.

Reply to
bitrex

For a Dell poweredge 2950, I added three parallel 100 ohms 1/4W resistors t o the PS fan and two of the four box fans, while removing the other two fan s and one CPU. The noise level is substantially lower, as well as the powe r bill. That was six months ago and the server is still up since then.

Reply to
edward.ming.lee

Suggestions?

Reply to
John Doe

Being interested in power consumption, I ordered a "kill-o-watt" power meter to go with the server (and other things). Some gamers spend $100+ on a 1000 W power supply while having no real idea what their system actually consumes. I had one before, but I think someone else is using it now.

Thanks for the real suggestions.

Reply to
John Doe

5 watt 3 volt zener diode for 12 volt fans, Higher current, zener - transistor base to emitter with 1k resistor. Did that for my Pentium someone dropped a server fan in, also too much noise.
Reply to
Wayne Chirnside

LM1117 and a trimpot?

Or something with a thermistor!

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics
Reply to
John Larkin

Thing is, the power supply fan controller will shut down the power supply if it detects something wrong with the fan. I don't know exactly what it looks for, but it won't operate if the fan is unplugged. I suspect it requires the fan RPM to be at least a certain value. Would be great to know. The need to research it harder. Just putting something on those inputs could work. Maybe (just maybe) even if I have to make a timer circuit for it.

Reply to
John Doe

I'm guessing that you could throttle down the fan voltage and the rotation sensor signal would still work. Try it!

You could fake the sensor signal, but that's more work.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics
Reply to
John Larkin

Seems like this would be the easiest way to try...

First, check an unused system board fan output to see if the voltage is reduced like it is on the used fan outputs, given a BIOS setting. If so, connect the noisy power supply fan supply leads to a system board fan output, but connect its signal wire back to the power supply signal input. That assumes the power supply doesn't also sense whether current is being drawn by the fan. But sensing RPMs is probably enough.

Seems like a simple oscillator circuit could run off of the fan's V+ and ground pins.

Reply to
John Doe

That is, if I knew what the frequency is. I might put this off until finishing the rest of my experiment with the legacy dual CPU six channel memory server. That is modding into it a GTX 960 video card, to see if it outperforms similarly priced end-user PCs. I'll put up with the noise for a few days.

Reply to
John Doe

the other option is to add damping & baffles

NT

Reply to
tabbypurr

snip

Sheesh... Just go buy a couple of fans with speed controls at the computer store.

The cost savings in personal time alone should be Yuge! Unless your personal time is worth the same as that of a Russian street sweeper.

Such lack of logic these days.

Reply to
DecadentLinuxUserNumeroUno

On Sunday, February 7, 2016 at 7:58:59 AM UTC-8, John Doe wrote: ...

...

Have you checked the BIOS settings?

When I replaced the motherboard on one of my desktops the fans always ran at 100% - that was the default BIOS setting although it did have the ability to manage the fans depending upon CPU temperature.

With the revised settings the fans only run at ~20% out of the time.

kevin

Reply to
kevin93

Alas, that's not a good idea. Server CPU heatsink capacity is well engineered, you probably need that fan making turbulent airflow (i.e. running noisy) to cool the parts.

Your only quiet option might be to fit liquid cooling, and use a very large radiator outside the box. That is a 1U (skinny) box, you won't fit a quiet-fan cooler inside it.

Reply to
whit3rd

DecadentLinuxUserNumeroUno wrote in news:n988ut$tmo$1 gioia.aioe.org:

Reply to
John Doe

Yes, but they don't affect the PSU fans.

--

>  
> When I replaced the motherboard on one of my desktops the fans always  
ran at 100% - that was the default BIOS setting although it did have the  
ability to manage the fans depending upon CPU temperature. 
>  
> With the revised settings the fans only run at ~20% out of the time. 
>  
> kevin 
>
Reply to
John Doe

Okay.

That doesn't explain why it runs at 100% all of the time.

If stuff is going to be outside of the 1u server box, it may as well be an ordinary PC power supply with a large and quiet fan. Many of them do over 560 W.

Modern video card fans using heat pipes don't even spin up until they are under stress, like when gaming.

--

Thanks to the replies/suggestions. I'll hopefully get back to it later.
Reply to
John Doe

Servers tend to be made slim so they can be stacked & racked, and that means they use small diameter fans, with noisy ball-bearings for longevity, that run at high speed to move enough air.In short, they scream.

If you have the option of fitting larger diameter fans, you'll find them much quieter in general.

Clifford Heath.

Reply to
Clifford Heath

Learn how to properly quote you retarded Usenet dumbfuck.

Variable speed fans have rpm output leads, idiot.

You are not even qualified to sweep streets. Goddamned crossposting retarded f*ck too.

Reply to
DecadentLinuxUserNumeroUno

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.