Help with Fan noise / controller

[narz]

In my quest to find a budget upgrade, I've picked up a HP DL120 G6 rackmount server.

The fans on this server are expectedly noisy as hell, and I can't find any other 6 pin fans to replace them. I was hoping to put a fan controller to lower the speed/noise as the server won't boot without a fan connected. (has 4 dual fans)

The dual fans look like:

The pin out:

Black = ground
Orange = open-collector tachometer
Red = +12 VDC
Yellow = open-collector, low-pass/high-fail, trip-point (speed) and locked rotor alarm
Blue = ground
Green = PWM - pulse-width modulated speed control

They questions I need help with are:

1) If I splice into the wires 12v (red) Gnd (black/grey) Tach (Orange) and put a fan controller between the fan and the board this will work?

2) This fan controller says it has 10W available per channel, the Server fans are 12v 1.86A (22Watts?) will this blow up something? or will it just work at a lower RPM.

I tried connecting a normal fan to the board pins(12v,GND,Tach), but during the boot process the fan was spinning and the board was complaining a fan was missing. I read somewhere that the culprit might be the yellow pin Locked rotor alarm, but not sure how to by pass that..

Any advice appreciated! Any... I'll have to sell it if I can't sort out the fans.

[peterb]

At your own risk....

I suspect that if you take the pin that the yellow wire is connected to down to ground, you may be able to use a conventional fan. If the pin is low, it indicates the fan passes (according to your description).

If you haven't already found it, this might be useful

http://www.avc-europe.eu/cms/upload/..._Catalogue.pdf

As for the fan controller, overloading it is not a good idea, although the specs don't make sense, either the current rating is too high, or the power rating is too low. The power could be per fan, but that doesn't add up either. I think I'd want more details before I bought it.

[narz]

Thanks for the link Peter, it indeed confirms the yellow is a go/no go signal incase of fan failure.

At your own risk....

I suspect that if you take the pin that the yellow wire is connected to down to ground, you may be able to use a conventional fan. If the pin is low, it indicates the fan passes (according to your description).

I also read the this was successful for someone else, but how would I actually drive the pin to ground or 0v? i.e connect the pin to the chassis?

As for the fan controller, overloading it is not a good idea, although the specs don't make sense, either the current rating is too high, or the power rating is too low. The power could be per fan, but that doesn't add up either. I think I'd want more details before I bought it.

I noticed that too, the 2.8A didn't go with the 10w rating, so its probably 8.3 watts in reality and will blow to smithereens. To that end I am considering this instead now, 20W per channel.

So do you agree this should work theoretically if I connect the 12V, GND and TACH from the fan controller to the existing fans? as this will take away the need for new fans and tinkering with that yellow alarm wire.

The fan controller lowering the voltage, as the fans are PWM controlled will that be an issue?

appreciate the advice mate.

[peterb]

D Just connect the pin the yellow wire goes to to the 0 volt line.

I can't tell if that controller is specifically for PWM fans . You can control the speed of any DC motor by pulse width modulation the power line, but a specific PWM fan has the PWM controller built in, and the PWM input is just a control signal. If that is the case, you might be able to control the fan speed with a simple potentiometer connected to the control line.. you need to experiment!

Other wise, the control signal needs to be connected to give maximum speed, the the fan controller will control speed directly.

But if the fans are noisy, they will still be noisy with the controller. Presumably they are being controlled anyway, so if you reduce the speed manually, you risk overheating the system. And those are pretty serious fans designed to shift a lot of air!

You said it was a server, so that may be engineered to give a better airflow than you might expect to get in domestic situation, but you may need to monitor system temperatures carefully.

So you need to carry out some experiments with the existing fans, or just decide to replace them with something a bit more inherently quiet, and then consider an external controller if required. Bear in mind that reducing noise involves reducing airflow, meaning less effective cooling.

[DaveLEL]

All the 1U servers I have worked on used high pressure fans like the one you show, they are needed to force a high volume of air through the heat sinks, making changes to the fan will result in overheating. In order to make a server only 1U high, a high air flow and noisy fans is a compromise that is needed- this is normally no problem when used in a server room.
The servers I have at home are all Dell, the fans normally operate at full speed for 30-50 seconds at power up and then drop to may be 50% for normal operation, are the fans on the HP staying on full all the time ?
Best solution is to put the server in a separate room, I have a spare bedroom, but know others that have servers in the loft.
Just don't forget you have one in the living room set for Wake-On-Lan, and wake it by mistake when working on the laptop at 2:30am!

Dave

[narz]

D Just connect the pin the yellow wire goes to to the 0 volt line.

The fan that I will be adding only has 4 wires 12V, Ground, Tach, PWM. So there is no "yellow/alarm" wire to connect to the pin on the board. So i'm not sure where to find a 0v line or such? (sorry i'm not good with these electrical terms)

So you need to carry out some experiments with the existing fans, or just decide to replace them with something a bit more inherently quiet

I agree they will probably be still very loud even with a controller and I will be attempting to replace them first with some beQuiet or Artic 120mm fans and skipping the idea of a controller for now.

The servers I have at home are all Dell, the fans normally operate at full speed for 30-50 seconds at power up and then drop to may be 50% for normal operation, are the fans on the HP staying on full all the time ?

These fans work the same, I should have of mentioned I'm removing the motherboard from its rack mount case and putting it in a bigger standard ATX case which will have better airflow and additional case fans if required to keep air flowing.

I will be monitoring the temps closely to ensure it doesn't over heat (it won't be running 24/7)

Just don't forget you have one in the living room set for Wake-On-Lan, and wake it by mistake when working on the laptop at 2:30am!

hahaha seeing how loud these things can get that would wake up neighbours!

Thanks for the help guys, I just need to understand what I'm supposed to do with that 0v / alarm pin that's on the board before I experiment with the fan swap next week!

[peterb]

Ground and 0V are usually the same, but technically ground usually means the chassis, while 0V is the line coming out of the PSU, which in 99.9% of the time will be connected to the chassis - but if you can connect it to the mobo ground, that would be preferable.

[Pob255]

Ground and 0V are usually the same, but technically ground usually means the chassis, while 0V is the line coming out of the PSU, which in 99.9% of the time will be connected to the chassis - but if you can connect it to the mobo ground, that would be preferable.

computers are grounded through mains via the psu, the chassis is grounded through the psu.
If the chassis is not grounded through the psu, then it's not grounded (it's not live ether) it will still act as a Faraday cage and EM shield. (but a case really doesn't need to be metal most of the time)

The easy way is to connect the alarm to one of the two ground pins on the socket, probably be able to just use a jumper (ie like a motherboard clear CMOS jumper) but that would depend on the exact pin layout on the socket.
That's probably why the socket has 2 grounds in the first place.
You can test if it'll work by carefully bridging the pins with this other fan in, the big issue will be if it's doing something more than a simple open circuit on fail.

[narz]

Thought id update with the outcome

I ordered a Arctic F12 /120mm fan and some jumper cables to get it connected the server board. After wiring it up, the motherboard was still complaining about a missing fan despite the fan spinning and everything looking good with the "alarm" pin connected to the GND point. Still no good.

I decided to use the board with the automatic fan control disabled (By setting a BMC jumper on the board) this allowed me boot and use any fan albeit at full blast (1800rpm) still much much quieter than the servers.

Everything looked good with idle temps at roughly 35/40c and passmark scores better than before, but under load it was shooting up to 70-80c and throttling it self! I had enough of it at this point and just went back to my trusted ol' Q6600.

Mission failed, but thanks for the help!

[hesham1516]

Have not tried a DL120 personally, but DL360s and above generally have a high fan speed on first boot which quiets down after like 5-10 seconds.