Compaq ProLiant ML350 G1 Fan Noise

[wumpus_byte]

Having worked at HP from 2001-2013 and with the ML350 servers in the middle of that stint, last year I picked one up for nostalgia. It's a G1 Pentium III 1 GHz tower.

But the fans are outrageously loud! :D I do recall the cacophony of noise in the server room in those years, but even running the one at home in my Cave is too loud! I think they both just run at full throttle all the time... Actually the 1U servers had similar fans with higher pitch that were harder on the ears.

Has anyone overcome this sea of noise with replacement fans - those that wouldn't otherwise negatively impact the system? Or is there a functional justification to keep these OEM fans creating a wind tunnel in the back room? Thanks!

 

[GiGaBiTe]

Fans in servers like that are designed for high static pressure to push through the drives, cards and CPU heatsinks. Using a lower speed fan with less noise will result in the parts inside the server running hotter. It probably wouldn't be as much of an issue on a PIII based server since they didn't generate that much heat, but the later Netburst Xeons most definitely required the high static pressure fans with individual CPU TDPs in the 100-170W range.

It may also cause the server to throw fan errors. I remember old servers from that era would throw fan errors if they detected the fans dropping below some ridiculous RPM like in the 5000-9000 range. The 1U servers had fans that spun even faster, sometimes in the 12,000 RPM range to make up for the much smaller blade area.

 

[wumpus_byte]

Fans in servers like that are designed for high static pressure to push through the drives, cards and CPU heatsinks. Using a lower speed fan with less noise will result in the parts inside the server running hotter. It probably wouldn't be as much of an issue on a PIII based server since they didn't generate that much heat, but the later Netburst Xeons most definitely required the high static pressure fans with individual CPU TDPs in the 100-170W range.

It may also cause the server to throw fan errors. I remember old servers from that era would throw fan errors if they detected the fans dropping below some ridiculous RPM like in the 5000-9000 range. The 1U servers had fans that spun even faster, sometimes in the 12,000 RPM range to make up for the much smaller blade area.

Thanks for your reply. I understand that the system expects the high rpm now. Didnt know there may be a monitor on that, though it totally fits the scenario. And a replacement fan could cause the errors as you mentioned... Talking myself out of buying replacements here... Could be for the better if the system expects otherwise! Appreciate it!

 

[hunterjwizzard]

Its not that the airflow is necessary so much as nobody cares about noise in a data center. The machines need to be designed to potentially run for years at a time with no maintenance, so its better if the cooling system is more efficient than it needs to be, hurray.

I, too, am in the boat of having too noisy of a vintage machine. I so far have no solution.

 

[GiGaBiTe]

Its not that the airflow is necessary so much as nobody cares about noise in a data center.

Not really correct. Try cooling two Paxville MP Xeons in a server with normal fans. They'll quickly turn into smoldering craters with their 165W TDP.

High speed fans are definitely required in many types of servers.

 

[hunterjwizzard]

I've dealt with a ton of DC equipment that has high noise low CFM fans because the mfg simply didn't care. On the support end of the scale its always really entertaining to have to explain to customers who install this stuff in offices "well no its really not designed for that"

 

[littlejackal]

Bumping this from beyond the grave a little, but I'm trying the same experiments with a P3-500 Proliant 1850R and the system does indeed have a fan fault sensor which induces a mandatory shutdown. The fan connector on mine is five pin, with two pins being common ground, one being 12v, one being a key so you can't insert incorrectly, and the last seems to be a sense pin.

I tried to replicate this with a 3 pin fan by bridging the ground across the same first two pins, 12v to the third, and the third pin in the last position, but no luck there. The more I think about this the more I think I ought to try a four pin fan and use the sense pin there because the 3rd on a 3 pin fan I understand to be PWM which wouldn't make sense in a sense pin, no pun intended.

I'm just guessing at all these pinouts though, as I haven't been able to find confirmation of a G1 proliant (or G0 1850R) fan pinout anywhere.

 

[GiGaBiTe]

Instead of bridging voltages and potentially blowing something up, probe the sense pin of a running fan with an oscilloscope and see what it does.

 

[littlejackal]

Yes, absolutely. I'm abandoning any further attempts until I understand the behaviour a little more.