ProLiant Servers (ML,DL,SL)
1748058 Members
5006 Online
108758 Solutions
New Discussion юеВ

Re: DL145 Acoustic Noise PC2700 v PC3200

 
Richard K Robinson
Occasional Contributor

DL145 Acoustic Noise PC2700 v PC3200

Hi
As a developer working from home I unfortunately have to sit in the same room with my server equipment. After recently purchasing a DL145 G2 it quickly became apparent that there was an unacceptable level of acoustic noise from the server. During my investigations into the problem I discovered that there is a huge difference in system temperature between PC2700 and PC3200, fan speed with PC2700 is 6000rpm (System Temp 27Deg) v 11200rpm (System Temp 36Deg) with PC3200. Obviously I realize that faster RAM , higher drive currents etc. will result in higher temperatures. I am just a bit puzzled as the RAM controllers are onboard the processors and there does not seem to be a great difference in Processor temperature between the two RAM speeds. So I would be very Grateful if some could explain the reason for huge difference in system temperature between these two RAM speeds.

TIA
Richard
8 REPLIES 8
raadek
Honored Contributor

Re: DL145 Acoustic Noise PC2700 v PC3200

It's not a rocket science - they system temp refers to what's going on your motherboard.

As memory modules (themselves) generate a lot of heat, they cause MB temperature to raise.

So faster memory => more heat => louder fans (unfortunately)

Regards
Don't panic! [THGTTG]
raadek
Honored Contributor

Re: DL145 Acoustic Noise PC2700 v PC3200

Bear in mind that you have only two fan speeds - slow & fast.

So although the temp delta might be small it causes fans to switch from one mode to another.
Don't panic! [THGTTG]
DLC_1
New Member

Re: DL145 Acoustic Noise PC2700 v PC3200

Our DL145 G2 is ridiculously loud, also. We are a small shop and our server room is also our kitchen.

Our fans stay nice and quiet until the first time that the system is put to heavy use, and then the fans kick in and will never change back to lower speeds. Nice cool (almost cold) air blows out the front of the box, but the fans are still on high generating an enormous amount of noise.

Based on this thread, I found where all of the cabling is bunched up next to the front corner of the mother board, but I had no idea what I was looking for in the way of a thermal sensor.

I tried to move the wires all back away from the board and we'll see if there is any success, but I'm wondering if I could get some guidance as to what the thermal sensor looks like.

Thanks, David
DLC_1
New Member

Re: DL145 Acoustic Noise PC2700 v PC3200

The box made it two days running nice and quiet, but it has now gone back into "jet engine" mode. :-(
Richard K Robinson
Occasional Contributor

Re: DL145 Acoustic Noise PC2700 v PC3200

Hi,
Ok, well I have investigated this problem and here are my conclusions! The reason for the high levels of acoustic noise are due to some very poor design on HP's part.
Here is the non techie short version:-
The Fan drivers on the motherboard drive two Fans the SYS fan and one other, the system temperature sensor is in close proximity to the fan drivers. The faster the fans are driven the more heat is produced by the fan drivers. As more heat is produced by the drivers the BMC detects this and tries to compensate for it by increasing the Fan speed. Basically if the fan speed increases above idling speed it becomes a totally self defeating system.
Here is the Techie long version.
Two of the cooling fans are driven by four MOSFETS on the main PCB, when these two fans are idling (About 5800rpm) they are driven from the 5V supply, this produces no supplemental heating within the server environment. However when running at higher speeds these drivers deliver voltages in the range of 5V to 12V and fan speeds of 6000 RPM to 12000 RPM. These fans are beasts and take quite a lot of current which causes the drivers to dissipate some not inconsiderable heat through the main PCB, the system sensors proximity to the driver IC's is such that they detect the heat produced by the drivers. The management controller compensates for the increase in system temperature produced by the fan drivers by driving the fan drivers even harder which as you have guessed produces more heat! Basically poor design on HP's part. I suspect that this problem would have been discovered during the design verification process and HP decided for whatever reason not to correct it.
Is there a solution?
Only one non techie solution keep your server in an air-conditioned room and make sure it is kept pretty cool!
Regards,
Richard.
DLC_1
New Member

Re: DL145 Acoustic Noise PC2700 v PC3200

I decided to contact HP support on this issue. Their answer was that the fans in this box only have one speed. After I explained that the box will run with very little audible noise until the first sustained high CPU usage event, and then the fans kick up to a very high audible noise, the suggested solution was to make sure the BIOS is up-to-date.

I tried previously to update the BIOS, but I don't have a diskette drive and the CD I created will boot on my desktop, but not on the server.

I guess I'll continue to try to get the BIOS updated, but I have very little faith that any change in behavior will occur.
Richard K Robinson
Occasional Contributor

Re: DL145 Acoustic Noise PC2700 v PC3200

Hi,
No, I am sorry that will not make any difference, I assume that the fans are handled by the lights-Out firmware. obviously HP tech support have no idea. If you access the Lights-Out remote management on the server you can monitor the fan speeds and they will vary in speed in about 1000RPM steps from 6000RPM(nice and quiet) to about 11000(RPM)(pretty dawn loud!).
Best Regards,
Richard.
Ernest Jones
New Member

Re: DL145 Acoustic Noise PC2700 v PC3200

Recently, we purchased 3 DL145 G2 systems. Everything was going great running Fedora Zod until last Friday when I returned from lunch to hear jet turbine sounds erupting from my office. Long story short, I removed the riser card assembly from each machine to allow more air to flow out of the back of the server. Problem solved! No more loud noise even in my warm office. The heat created by the power supply can now vent and the fans stay nice and quiet... This may not be the best solution but it works for me...