How important is pressure?

[silverdemon]

Hi all,

I recently got a new computer, a Dell PowerEdge 2450.
2x PIII 1000MHz
1.25GB RAM
4x 36GB SCSI
2x PSU (330W)
oh yeah... It's actually a rack-mounted server, 2U high, here are some pictures:

The front

Top view, I put a keyboard on it to show the size of it.

With the doors opened

mounted in my home-made rack (I will tidy it up later)

Top view of the 'rack'

The console, everything is temporary, the thing will run headless and be controlled through the network

Sorry about the quality of the pictures, I still don't have a good camera, so this'll have to do for now...

As most of you probably know rack-mounted servers produce a massive amount of noise. I checked for noise levels on the blower which cools the processors and the mobo and it does 52dB!
So I would love to change the fans in this apparatus but I have one little question: How important is the pressure generated by the fans in the system?

The blower for example only pushes 28.25 CFM of air, but it does that at a ridiculously high pressure (33.0mmH2O). I could easily change this blower for a fan with twice the airflow, much lower pressure but also at 27dB or even less!
You should know however that I only have this server in my home-made rack. I will therefore be able to cut some holes in the top of the casing through which air can enter and/or exit.

So once again, can anyone tell me how important the pressure produced by the fans (in my case) is?

Thanks in advance!

[Luke122]

In a typical rack mount situation, the fans need to move air from the front of the case all the way through the back, and still keep up enough pressure to efficiently remove heat. Most rackmount enclosures are built to be able to withstand having several servers or devices stacked on top of each other, with minimal airflow around them, which is why they need lots of airflow through them.

I wouldnt cut any holes in the casing, as it'll disrupt the air path, and possibly lead to hotspots. Maybe just try a different (quieter) fan first, and see fi the temps stay stable. I switched out all the fans on an Antec 3u case running dual Xeons with MUCH quieter fans, with no noticeable effect on the temperatures.

[silverdemon]

Ok, I'll try to switch off or replace some fans before cutting.

Just for the record: I was planning on cutting a hole just in front of the front-most fans. This should guarantee enough cold air entering the case. Or at least, I thought it would.

If anyone else knows any tips/tricks, please let me know

[Luke122]

Holes in the front should be ok, I was thinking you meant onthe top or the sides.

[silverdemon]

well, maybe my word were a bit misleading. I was actually planning on cutting in the top... let me explain with some pictures:

This is where I planned the slit/cut

schematic top view

As you can see the slit will be in front of the front-most fans, so they will have a nice supply of cool air. With this slit and some wire management I think I can create a tunnel that is more 'free of obstacles' than before. Remember that I don't have a real rack, so the top side is free, there's no other server/obstacle there.

I haven't tried replacing fans (or the blower) because I don't have any lying around atm. But I have tried to get temps from the thing. However, no program I tried could find anything like a sensor. I tried different programs, like speedfan and everest, but they all couldn't find ANY sensor..

So if anyone could help me out with the temps or has anything to say about the location of the slit, please let me know

[jdbnsn]

Another option you may consider is compartmental cooling. With this approach you would be able to cut vents in the case to vent each compartment with smaller, quieter fans to cool each compartment individually. But I agree with Luke that is you use a unified cooling plan, vents may be a bad idea. There is always watercooling!