PROBLEM: Dell Poweredge 2950 III Jet Engine Fan Noise

PROBLEM: Dell Poweredge 2950 III Jet Engine Fan Noise
FIX: Install a 47 ohms resistor on Fan1,2,3,4 and Power Supply fan
OBJECTIVE: To reduce the fan noise of Poweredge 2950 III without overheating the CPU
STATUS: Completed
QUESTIONS:
How noisy after the mod? I can stay and work in my office all day without being bothered by the noise.
How hot is the server? Tjmax is 100’C
What’s the Server Rack Ambient Temperature? 86’F
What is the Server Room temperature? 80’F
What is the house thermostat setting? 76’F
How much was the cost of the mod? $1.49 from Radio Shack
Why use 47 ohms? From my 2007 experiment to reduce the fan noise of a Poweredge 1550.
10v=12 ohms
9v=20 ohms
8v=30 ohms
7v=42 ohms
6.5v=50 ohms [6500 rpm]

BACKGROUND:
System Information:
Dell Poweredge 2950 III Small Form Factor
CPU-Intel Xeon 2.66GHz 1333 mhz
Memory-32GB
Drives-8-73GB, 2.5” SAS Drives
Purpose: Windows 2012 Hyper-V Server Host
Refurbished from Ebay: $600.00

Post Work Video
View Video1-Booting Fan Noise – Zero
View Video2-Floss Air Flow Test

Tools/Supplies/Skills Required:
Soldering iron
Heat Shrink
5 pcs. 47 ohms resistor 1/2 watt from Radio Shack p#-2711105
Soldering skills
Patience

Step by Step:
0. Proceed at your own risk.
1. Turn off power
2. Remove cover
3. Remove Fan 1,2,3,4, Figure 1
4. Cut “Red” wire, Figure 2
5. Solder a wire to extend the red wire and heat shrink, Figure 3
6. Repeat Step5 for fan2,3,4
7. Solder a 47 ohms resistor inline with the newly added wire extension and heat shrink it, Figure 5
8. Reinstall fan smartly back to the fan mount.  Smartly-meaning be careful not to let the wire get tangled with the rotating fan or short circuit it to ground, Figure 6
9. Remove 1 Power Supply [modify 1 power supply only, save the other one as a backup just in case this mod will burn it over time]
10. Remove the cover of the power supply [smartly], you can figured it out.
11. Remove fan from the power supply, Figure 10
12. Cut “Red” wire, Figure 11
13. Solder a wire to extend the red wire and heat shrink
13. Solder a 47 ohms resistor inline with the newly added wire extension and heat shrink it, Figure 12
14. Reinstall fan smartly back to the fan mount.  Smartly-meaning be careful not to let the wire get tangled with the rotating fan or short circuit it to ground.
15. Replace the power supply cover [smartly], it needs some thinking but you can use the other spare power supply as a visual aid of how the assembly should be.
16. Reinsert power supply back to the server
17. Turn on server
18. Monitor server temperature
19. Verify server temperature at the back
20. END

REFERENCES:
Figure 1
image

Figure 2
image

Figure 3
image

Figure 4
image

Figure 5
image

Figure 6
image

Figure 7
image

Figure 8
image

Figure 9
image

Figure 10
image

Figure 11
image

Figure 12
image

Figure 13
image

Figure 14
image

Figure 15-Core Temp
image

Figure 16-Carrier Touch-n-Go App [Home Thermostat Setting at 76’F, Server room temperature is 80’F]
image

Figure 17-Rack ambient temperature is 86”F
image

Figure 18-Room temperature is 80’F
image

Figure 19-Finish Product-Poweredge 2950 added to the 42U rack inside the home office.
image

Figure 20 Electric Bill
What’s next? “Off the Grid” Solar Power.  This Poweredge 2950  is rated 780 watts.
780 watts/1000 kw = .78kw
.78kw x 24 hours = 18.72 kWh/day
18.72 kWh/day x $0.10/kWh=$1.87 cost per day
$1.87 x 30 days=$56.1 cost per month
$56.1 x12 months=$673.2 cost per year

image

 

View All Poweredge Tags

Advertisements

142 Responses to PROBLEM: Dell Poweredge 2950 III Jet Engine Fan Noise

  1. Anonymous says:

    After adding the 47Ohm resistor to the 4 fans I am getting E1310 RPM Fan and E1313 error code on the front panel LCD display and the PSU fans are kicking into full gear on and off at say 30 seconds interval. My model is DELL PowerEdge 2950. The system video out put is null and the LCD Monitor is dark. I am not even sure the system is up and running, I am going to do the mod to one of the PSU and remove the other as per the document.

  2. Ashley says:

    This page is still very helpful.
    I have modded the 4 fans and 1 X psu fan with excellent results.
    I have also tried replacement fans which shaved another few db from the noise.

    The psu mod made a big difference after the chassis fan mods.

  3. Anonymous says:

    I have VMware ESXi 5.5 installed and running can I update bios and firmware with this OS?

    • Derek Zacharias says:

      I wouldn’t try it, if you brick you rig then your done. Just grab an extra disk and load Server2008 . Its very easy to update a BIOS from windows. Just download and execute. No fussing with the command line.

  4. Derek Zacharias says:

    Hi all,
    I have a PowerEdge 2900 Gen I I believe.
    I did the resistor mod with 33ohm 1/2 watt resistors on the red wire and it seems that my fans are cycling on and off. After doing so the power supply that is installed is on full blast. To fix this is tried to re flash the BIOS 2.7.0 and the BMC 2.50 with the original firmware. Still having the same problems it seems the fans are cycling anywhere form 0 – 1500 RPMS. With the fans still cycling I decided to do the BMC hack and I believe I successfully modded the firmware but I am still having issues with the power supply fan cranked up to 100 percent. Any help would be welcoming at this point. Thanks!

    • Anonymous says:

      I thought I would share my findings after attempting this mod… I originally went with the 33ohm 1/2 watt resistor. The fans did quite down, but would pulse on and off, the power supply fan remained on full blast. I thought that it might be the resistors so I tried the 47ohm 1/2 watt resistor. same problem. After thinking that the problem was the resistors that installed, i decided to take them out. I booted the machine. same problem… flashed the BIOS and BMC back to original… same problem. In short the server mobo is jack up… the fans are on full blast and I am worse off that I was in the first place when the server would at least settle down. Time for a new project.

      • caveman says:

        Just have to ask this question, did you put the chassis cover back? Also, install the Dell Open Manage software to get a temperature and fan rpm accurately. Don’t give up. 47 ohms works, mine still running for over 2 years now.

      • Derek Zacharias says:

        Ok, On the advice of blindcaveman I decided to try again. Long story short,The resistor mod does work. My advice, don’t try to jack with making the server fans run a bit faster with a 33ohm 1/2 watt, the Dell Poweredge Gen I fans will pulse for 0 – 1500RPMS. Just do the mod with the 47ohm 1/2 watt resistor. If you do the mod and the power supply fan kicks up, then put a resistor on the power supply fan as well. Done deal, don’t stress about the heat. blindcaveman has been running his for two years! nice an quite. Thank you again sir, nice work!

  5. Quella says:

    I wanted to update people here after receiving my resistor order on my modification experience of the four fans using a set of 2watt 30ohm resistors instead of the 47. I made the modification today and the system sits in a basement where it is often cooler, so we will see if things change by mid summer. A quick and painless modification and I have not had any issues with the fans spinning up and down as others.

    Here are my number (in Decibels) before and after the modification:

    Home with no noise: 54 db (Quiet Home Volume)

    Before Start Up: 84db (Smoke Alarm Volume)
    Before Running: 74db (Vacuum Cleaner Volume)

    After Start Up: 72db (Busy Street Volume)
    After Running: 62db (Normal Conversation Volume)

    Much quieter and the family is very happy. I do not have the fan speed details because I do not have the monitoring tool installed, maybe next check-in I will post them. I wanted to say thanks for the recommendation and easy conversion.

    • Just to bring up a fair point of notice: if you do use all four fans the memory core temp will be at alarming high temperature. over 70 degrees celsius. to hot to have a long life. So take my advice, use a default fan in the 4th port on the right side.

      • Quella says:

        Good advice and I did not see your comment until I did all four fans already. I can modify one back to its normal state as I watch the system and see how the temp gets. It is not on 24/7 nor running heavy all the time. It is more a lab system which is on only during home testing.

  6. Roy says:

    Hi I Thought I should mention my best setup for now.
    I have 3 coolers with 20 ohm 0.5 Watt resistors and the 4th cooler is standard.
    Why did I do this, I have seen that with all 4 coolers with 20 ohm resistors and full load the memory core temperature gets over 73 degrees celsius. At this moment it sits steady on a 57 degree celsius level and holding. Firmware & BIOS to latest and no BMC mod.
    RPM’s of all from left to right: 4100 – 4100 – 4100 – 7350. The normal cooler is in fron of the intake for the memory so that its better cools. There is some noise but the normal cooler will slow down from 12000 RPM tot 7350 RPM. I can live with it, sins the server rack is in my garage.
    Is much quieter then with all 4 standard coolers. has anyone else made a fan mod for their memory unit?

    I also thought of using a broadboard to test different resistor types but I changed my mind now I tested the 20 ohm resistors.
    Will let you guys know more how it goes in a few days.

  7. Quella says:

    Thanks for all the great info here on the site. I have gone though all the comments and there have been some good recommendations from so many with the same issue. I did have a question I was hoping you could assist with.

    I’m looking at using 2watt 30ohm resistors (or somewhere in that neighborhood) to keep from having to patch the BMC; which you did not seem to have to do. Radio Shack no longer carries the recommended parts. Any recommendations on good online sellers of resistors and heat shrink materials in smaller quantiles? The wife and child are not happy with the Ebay F-16 I installed in the basement rack.

    • caveman says:

      Try Amazon, they have resistors there. Just be aware that some are Chinese vendor and can take time to deliver. Just mimic what I had, it’s still running inside my small office 24/7 with no issues. The noise is very tolerable I can sleep there if I want to. Enjoy.

      • Quella says:

        Your recommendation to try Amazon and see what I can find was a success. I tried one company that did not carry higher wattage resistors (I was looking for 2w just to be safe even through your has not had any issues), they recommended Digikey as a supply company. I also went with 30ohm just to try a bit faster fan speeds than what 47 would give me. Along with all of that I bought some 8mm (resistors state they are 5mm) heat shrink tubing for sealing everything up like you had. It will take a few weeks as you stated, however for 30 pieces at $4.00 US (smallest qty they had) it was a great deal if it can quite down the 86db fans when running in normal mode.

  8. Mike Waldron says:

    A client of mine did this hack on his own at my suggestion, since he sits a foot away from the server rack. It works but I’m getting a system board temp of 35c which seems a tad high. Should he/I be worried? Unmodded 2950 IIIs are about 22c by my sampling of a few clients that are unmodded.

    • Mike Waldron says:

      Here are the fan speeds under regular load
      System Board FAN 1 RPM 3450 RPM
      System Board FAN 2 RPM 3150 RPM
      System Board FAN 3 RPM 3525 RPM
      System Board FAN 4 RPM 3075 RPM

      • caveman says:

        Here’s the reading from my system with fully loaded Hyper-V system 32 GB ram. I still don’t have any issues. Been running 24/7 since I started the mod.
        System Board FAN 1 RPM 3450 RPM [N/A] [N/A] 2025 RPM [N/A]
        System Board FAN 2 RPM 3525 RPM [N/A] [N/A] 2025 RPM [N/A]
        System Board FAN 3 RPM 3450 RPM [N/A] [N/A] 2025 RPM [N/A]
        System Board FAN 4 RPM 3150 RPM [N/A] [N/A] 2025 RPM [N/A]

        Threshold Settings Status OK
        Reading 28.0 C
        Minimum Failure Threshold 3.0 C
        Maximum Failure Threshold 47.0 C

  9. Glen Bishop says:

    This thread is a bit old but I thought I would share my experience with the mod as well.

    Just completed this mod on my PE2950 Gen III over the weekend using 22ohm, 1 watt resistors and everything works perfectly without the BMC hack. The intake fans spin around 4200rpm idle and peak at around 4800rpm under load. Like some others have posted, the resistors do get a bit hot and if I had my time back I would have picked up 2 watt instead.

    To compensate for this, I positioned the resistors directly behind the fans but not touching the CPU heat syncs to give them a bit more cooling. I used some electrical tape to pin down the red wire on the fan shroud and positioned the resistor to be right in the path of the air flow being pushed to the back of the case.

    I stress tested the machine using prime95 (max heat/stress) for 12 hours and the 2xE5345 CPUs peaked around 80C while the motherboard temp was around 28C, room ambient temp of 23C. The fans still give off some noise (around 55dB normal usage), but it’s much better then the jet engines they were before. Since I plan on putting a good bit of CPU load on the server (ESXi host with 6-8 VMs) I went with the lower ohm resistors to keep the fans moving faster.

    Overall an excellent mod, will keep this is mind for future servers if they need to be used in a home/small office setup.

    Thanks for the excellent write up!

    • caveman says:

      Glen: Thanks for the response. I totally forgotten when I did the mod and by reviewing my Radio Shack receipt, I did it on Sept of 2013 and that makes it 2 years of uptime. My original mod was unchanged and the server is still running inside my office 24/7. I did replace 2 2.5 ” sas drives due to failure but cannot narrow down to temperature problem. Other than that, server is still up and running.

  10. Olaf says:

    Hi there, I made the correction using the resistor with good results. I was wondering if each fan is controlled separately with separate power line. If not why not use an adjustable resistor for all of the fans? So one could externalize all wires for the fans and use the adjustable resistor outside the box.
    This will give much better control!

  11. Rob says:

    The reason the fans go ballistic is due to not having the right drivers / software installed.

    I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.

  12. silolsilol says:

    Hmm anyone still working on this topic? My fans are always at:

    FAN 1 RPM | 30h | ok | 7.1 | 7350 RPM
    FAN 2 RPM | 31h | ok | 7.1 | 7425 RPM
    FAN 3 RPM | 32h | ok | 7.1 | 7650 RPM
    FAN 4 RPM | 33h | ok | 7.1 | 7575 RPM
    FAN 5 RPM | 34h | ok | 7.1 | 4200 RPM
    FAN 6 RPM | 35h | ok | 7.1 | 4200 RPM
    Fan Redundancy | 75h | ok | 7.1 | Fully Redundant

    Sometimes RPM is higher, and this server is loud as f*ck :/ I tried BCM mod, but I never get that low RPM. System Board Ambient Temp is always 25.0 C

    what now? 😦

    • Alex says:

      Then, What’s your problem silolsilol?

      Have you done the resistor modification yet?

      In my case, system board ambient temp is 20-21º in winter, that is still now. And it will be 28-31º in summer, I hope no more.

      I don’t now power supply fan rpm, but the cpu fans are about 4800-5000rpm, with 22 ohm resistors. The noise is acceptable.

      Best regards.

  13. Ryan says:

    Hi there, I proceeded exactly as you posted but now my server fans are switching on and off sporadically. Is this a problem? Did I do something wrong or do I need to modify the firmware?

    • caveman says:

      Monitor the fan rpm using Dell Open Manage and find out which fan is not rotating at a desired rpm. If you installed the resistor value, then your fan rpm should be the same. Let us know the result so that we can learn also.

  14. Fames says:

    2950 is a 2U form factor. What about the 1U form factors like 1950? Is it possible to such a mod on them, theoretically speaking? Is there anything fundamentally different about the 1Us’ heat dissipation system?

    • caveman says:

      Originally, it was a Dell Poweredge 1550 that I experimented in 2007. I later learned that the following resistors bring down the fan rpm to a more tolerable noise level. I placed the Poweredge 1550 in my very hot garage [100’F in Round Rock, Texas] and did the torture test for 2 months. The result: the Poweredge survived with a 47ohm resistor in all fans.

      10v=12 ohms
      9v=20 ohms
      8v=30 ohms
      7v=42 ohms

      The modification theory should be the same, lowering the voltage to the fan will lower the noise level. I don’t have a 1950 so I cannot really say if the fan mod will damage your system. Hey, it’s your turn to experiment and let us know if it works. I love to get my hands on the 1950’s since it is a lot cheaper now but I have 3 servers on power off because it costs me $20.00 per month to run all. I instead ran with my 2950 with 32GB memory and used Hyper-v to consolidate my server needs which is only Sharepoint 2012 and SQL.

      Yes, you should be able to use the same mod on 1950 but proceed with caution. Do not use on production server and make sure you have a bare metal backup and a fire extinguisher handy just in case… i’m just kidding.

      Please let us know your findings and take pictures and notes for the rest of us.

      • Fames says:

        Sure, I will post more info if I try the mod on another model. But first I will search for a 2950 (I haven’t made the purchase yet).

        I am testing a Proliant DL160 G6 right now, that is considered a new generation, “rather silent” model, and the noise around the case is at 70db. If your mod results in 53db, I think my choices are clear. 2950 III it is…

        I haven’t seen anyone who has tried the mod with the Proliants, and sincerely, I don’t think it will be possible. The design is very deficient, given the size of the fans and their number (6 pairs with intake and outtake fans). Visibly the pairs are not well-synced and they increase the noise horribly (which means they are aerodynamically and energetically deficient as well).

        I intend to run a Debian-based Linux with different database servers and an R server on it. I may run heavy computational tasks on the server some times, and I wonder if the server will behave under heavy load after the mod.

  15. Macgyver says:

    Hello,

    i am trying to mod my PowerEdge 2900 but i ‘ve hit a strange problem.
    I started by modding one fan only as a test, but while i ‘ve followed the instructions above to the letter, the modded fan starts and stops all the time, making all the other fans work on max.
    In DRAC, under Fan Probe Information, modded fan’s rpm reading fluctuates between 0-1050 and the rest unmodded ones around 4800.
    I tried modding another fan in case the first one was problematic but i got the same results.
    For the mod i used 47Ohm 1/2W resistors.

    Thank you in advance

    • aster says:

      Read the comments above, you need to mod all 4 fans at once. Also 1/2W is a little bit low. Try to get 1W resistors.

    • caveman says:

      Your system is a Dell Poweredge 2900 which has larger fans. The 47ohms resistor might not be the right resistor to lower the rpm. I’ve found the somewhat correct value for 2950 Gen3 to lower the voltage of the fan using a 47ohm resistor by trial and error. With your system, you have to do a few trials to get the right fan rpm without violating the lower spec limit of the fan rpm set in the BIOS. Your objective is to lower the voltage to the fan by testing several combination of resistors and then listen to the noise generated by the fan. If satisfied, you found the correct combination.
      Do the following:
      1. Have a few 47ohms resistors handy.
      2. Cut the red wire and extend with an alligator clip use a few inches of extra hookup wires. [extending the red wire makes the testing much easier to swap different resistor values]
      3. Connect the 47ohms resistor to each fan.
      4. Boot and go to Dell Open Manage app and observe the fan rpm. Record reading on paper.
      5. If the rpm is too low, connect 2 47 ohms resistor in parallel to all the fans
      6. Test and observe but do not solder yet.
      7. Repeat Step #3-#6 with other resistor combination until you find the correct resistance that satisfy the fan rpm requirement and your noise level satisfaction

      Once you find the right combination of resistors that makes the fan rpm stay within the lower spec limit of the fan rpm spec, then you found the mod resistor for the 2900. Remember, the fan rpm monitor in the BIOS is a close loop system, meaning – that the BIOS will try to compensate the fan rpm if the threshold is violated as specified in the BIOS. So when the fan is rotating at lower rpm in violation of the threshold, the BIOS will try to compensate and send a command to the fan to rotate more … therefore louder.

      Be patient, you will get the noise level tolerable. This is how I accidentally found the 47ohm resistor mod. It was not a scientific process, it was all trial and error.

      My system still running 24/7 since I posted this mod more than a year ago.

    • Erik says:

      Hi:
      I have modded my PE 2900 and the best way is replace six 92mm pwm fans with quiet ones. I found some from Ebay for about $9.00 a piece from Hong Kong. Next you have to mod the BMC firmware in your Poweredge 2900. Can find info from the web, the threshold for PE 2900 is around 1250rpm. You need to change to a higher value to about 1600 or so. Think about what we are doing to the fans. First, installing lower rpm fans and your server is still trying to spin at preset values of 1250. Physically the new fans are spinning at a lower rpm. The Firmware will compensate by spinning at max. That’s why you need to change the thresholds to a higher value around 1600. I am still playing with the threshold a little but I am happy with the setup. Still can hear some fan noise but it does not sound like a jet engine in the room. My CPU’s are around 20 degree “C”, 31 degree “C” when it’s working, fbdimms are about 38 degree “C”.
      Poweredge 2900 with two 3Ghz cpu’s, 32g byte of rams, eight 2TB hard drives.
      Hope this helps…

  16. Nu Era says:

    I believe there’s a way to do this with just software – must be done under linux however.

  17. Wishmaster says:

    Hi, I have a small question, I am looking to do this mod on my 2950 Type III as well and just cant find the same type of resistor in my country. is the following okay to use as well? are there changes between the one you use and the one i found in my country? except for more pieces in one purchase.

    100 PCS. 47 Ohm 1/4W 5% Carbon Resistor Bargain pack:

    http://www.ebay.nl/itm/100-PCS-47-Ohm-1-4W-5-Carbon-Resistor-Bargain-pack-/220485802618?pt=UK_BOI_Electrical_Components_Supplies_ET&hash=item3355fa5a7a

    • caveman says:

      You need to find a resistor that is 1/2 w that is close to 47 ohms, it does not need to be exactly 47 but somewhat close. Larger resistance = lower voltage= slower fan rpm = less noise. I can’t say if 1/4w is safe. Heat is your enemy if you go for a lower wattage. I have 1/2 watt on mine and it’s been running for >1 year with no issues. Keep looking.

  18. crazydrve says:

    This is for anyone who doesn’t know with some of the dells like the optiplex 755 there is a 4pin to 5pin cable you can buy to use a normal pwm fan

    http://www.frozencpu.com/products/8469/cab-152/FrozenCPUcom_4-Pin_PWM_to_5-Pin_Dell_Adapter_Cable_-_8_Length_CB-PWM-D5.html?tl=g47c251s1325

  19. crazydrve says:

    First I would love to say thanks for finding this stuff out. Just did my pwe2950 gen 3 6×3.5 hd and temps seem good except for ram. Does anyone know what ddr2 server ram should run at?

    I used the 47 ohms resistor 1/2 watt. fan rpm is 3900.

    windows 2008 r2
    2 x CPU E5440 @ 2.83GHz
    system board temp is 31c
    cpu’s are around 50c -60c
    ramp temps say anywhere from 65c – 75c – (4gb ddr2 5300) x8 for 32gigs.
    cpu usage is about 25% and physical memory about 60%

    Also the resistors feel very hot to me. I cant touch them for more than a second. used 4″ wire length

    any help is appreciated and thanks again
    crazydrve

  20. 20648527rp says:

    Maaaan, your tutorial is amazing.I own a 2950 II which I believe is the same configuration than yours.I have 1 questions.Those Dell Optiplex 755 you talk about the have a a slitghly different cable color configuration than those shown here , there’s a magenta cable instead of a blue one… does it matter ? Do I still have to cut the red one on the Dell Optiplex 755 ?

  21. Don’t try this with 1/4W resistors. I did my 3rd gen 2950 using the un-labaled resistors I had and they melted on the first boot. I looked them up on the site I had purchased them from and indeed, they were only 1/4W. All of them melted the shrinkwrap during first boot fan ramp up. They appeared to still be working – but I shut things down until I can get some higher wattage resistors.

  22. JDK says:

    I also fired mine up with the new modded fans and she was quiet on the heat sink fans but the PS fans were still a bit much. Unplugged one and modded the inside one with a 33-ohm 1/2 watt resistor and it knocked it down to a much nicer level. I left the other PS unplugged. I dont think it liked that. Now Im getting ps redundancy errors across my screen. 3 errors and they are all redundancy. Ill wire the other one up tonight. I ran out of resistors.
    I do notice that the fans pulsate up and then down constantly so I imagine Ill have to do the modified BMC flash.

  23. JDK says:

    Any place in particular you can get say 33-ohm 1 watt or 2 watt resistors? Id like to up the wattage a little on mine but I cant find them anywhere.

  24. Alex says:

    Finally, we’ve started classes in high school and we’ve been able to do the resistor modification.

    As I mentioned earlier (a previous message), I was worried about amount of resistor watts (1/2w). I tested 1watt resistors in June, but to complete the project really I have decided 25watt resistors, YES!, I will sleep like a baby.

    I have used a series of two resistors per each fan, one 22ohm and 25w resistor and one 10ohm and 25w resistor.

    Besides, each series ends in a toggle switch. I’ve used two toggle switchs, 2 pole and 3 positions each. Each toggle switch control 2 fans.
    In this way, with the toggle switches, I can select 0ohm, 22ohm, or 32ohm (the sum of the two resistors).

    More or less, if fans run about 7900rpm without resistors (0ohm) then runs about 4900rpm with 22ohm and about 3600rpm with 32ohm. But I am happy with the 22ohm option (almost without noise).

    Unfortunately I have been a stupid and I haven’t taken photos (or good photos), because of that I’ve had to open the server, again, and take really bad photos.

    Here you can see the photos, with explanations, I took:

    https://plus.google.com/photos/102170275494726101529/albums/6064980385456579921

    I hope someone can use these ideas that I’m writing in this text right now.

    Thank you so much to everybody for the help you have given me.

    Best regards.

  25. JDK says:

    Just soldered mine up while I was at work. Ill throw them into the f-16 when I get home. Hopefully I wont have to modify the BMC. Im using 33-ohm 1/2 watt

  26. Anonymous says:

    Hello everyone, I moded my 4 Dell PowerEdge 2950 CPU fans and one fan had a problem, after placing the 47 Ohm resistor the fan went to 600-750 RPM so the easy solution is to wire another 47 Ohm resistor in paralel with the other. Fan runs at 3400 Rpm now and the resistors are sharing the load and they are cooler.
    Hope it helps.

  27. Hector says:

    Hello everyone, i decided to try this on my PE 2950 last night. However when i tried starting up the server there was no response, didnt even turn on. I tried an unmodded psu aswell with no response. Both PSU’s have the green led on, so thats working. Could the mobo or cpu have died??

  28. Scott says:

    Hot damn! Followed your instructions (except for the PSU, left it alone) and I’m 2.1K-2.5K on my fans. You rock! I am not worthy!

  29. aster says:

    Hello, I have tried your mod today but in openmanager, with 47 ohm resistor, it says that the modded fan has 600 rpms (and ofc all other fans are going 100%). I have measured resistance with resistor and its 65ohms (resistor is 47ohms for sure). So maybe its because cable/soldering points?

    I will try 33ohms tomorrow. I have newest bios and bmc. Test system is Win 2008 R2.

    Anyway thank you very much for the article, I hope I will get it working little bit quiet, its very nice server (and cheap) and I’d like to use it as home server for developing (EXSI) and freebsd as home router.

    • Rafal says:

      Hi.
      I have also fixed rpm on one fan which was modded, after modded 4fans everything was ok.
      Measure resistance in circuit always isnt good idea.

      • aster says:

        What do you mean? You mean that I should mod all 4 fans and then check the speed? 600 rpm reading is just false reading when one fan is modded? Will it jump up after all 4 will be modded? I have checked all the cables and cables are looking ok. No broken isolation or circuit.

      • Anonymous says:

        Yes, i have strange rpm when only 1 fan was modded, after mod 4 at time,and put it back all was good.But i used lower resistors – 22ohm/1Watt.

    • aster says:

      Thank you guys, already did the mod. I have user 33ohms 1W resistors. Its a lot quiet now, RPMS are from 3000-3600 and everything is working smooth. Will test it on Windows for 1-2 days and then i will install EXSI. If there will be any problems I will write it here.

  30. Alex says:

    First of all, thanks a lot to everybody for sharing all posible solutions.

    However, my thought is:

    The max current of the system fan is 1.40A, and the voltage is 12V. Therefore the heat dissipation is (P=V*I) 16.8W, that is the value shown in the fan datasheet.

    That means when the server starts up, or when the cover is opened, the 1/2w resistors are going to burn.

    I am thinking 20w resistors should be the solution.

    What do you think people?

    Again, thank you very much.

    • Rafal says:

      Hi Alex.
      You are thinking little wrong.

      Max current is 1.4A and 12V so its =16.8W but this is total power of fan not for resistor.

      If you want calc total power ,at first we need measure volatage fan during normal work.
      Please open cover, lock switch on the left (then motherboard will be think as cover is closed),
      then please measure fan voltage during normal operating work.
      I cant do it because i’m after mod.

      Btw you might have some right about overheating resistors after full run of resistors.

      • Alex says:

        Hi Rafal,

        Thank you for your answer.

        Maybe you are right, but I am going worried putting a resistor that only works during normal work. I want to feel better thinking the resistors are not going to burn.

        Take a note that my server ambient temperature, now, is about 29ºC, and rpm are about 8450.

        Therefore, I have decided to buy 25w resistors. I know 25 watts is more than I need, but I will sleep peacefully.

        Best regards.

    • caveman says:

      There’s theory and there’s practice. This mod came out from trial and error. Mine is running now for more than 10 months with no issues. We can deep dive on the science of the mod but bottom line, the goal was attained which is – to utilize an old powerful x64 box at home without incident. Thank you guys for trying out the mod. Please modify the mod so that we can continue to evolve this project and make it better.

      • Alex says:

        I have simulated the modification, but there are strange things.

        I only modified the 4º fan.

        How is it possible that rpms are the same for 33ohms resistor and 39 ohms resistor?

        With 33 ohms:

        Fan1: 8175 rpm
        Fan2: 8250 rpm
        Fan3: 8250 rpm
        Fan4: 3675 rpm

        With 39 ohms:

        Fan1: 8100 rpm
        Fan2: 8250 rpm
        Fan3: 8175 rpm
        Fan4: 3675 rpm

        The resistors have 2 watts and 1 watt respectively

        Regards.

      • Rafal says:

        Alex,
        Are you wating about 10minutes after mod?
        Because fans rpm are stabilized few minutes, then check rpm.

      • Alex says:

        Hi Rafal,

        No, I didn’t wait 10 minutes after mod. I didn’t see the clock, but since the first test until the last test with the same resistor maybe 10 minutes.

        I am going to show you (everybody) the tests that I did.

        *Original, without any resistor:

        – Ambient temp: 28ºC

        FAN1: 8250 rpm
        FAN2: 8175 rpm
        FAN3: 8250 rpm
        FAN4: 8250 rpm

        Core1 Temp: 44ºC
        Core2 Temp: 47ºC
        Core3 Temp: 42ºC
        Core4 Temp: 47ºC
        Core5 Temp: 40ºC
        Core6 Temp: 43ºC
        Core7 Temp: 41ºC
        Core8 Temp: 47ºC

        Core temperaturas didn’t get modifications because the FAN4 is not in front of any processor, it is situated at right of the heat sink.

        All tests were done with the cover opened and the sensor intrusion locked.

        * With 22 ohms resistor:

        FAN1: 8250 rpm
        FAN2: 8325 rpm
        FAN3: 8175 rpm
        FAN4: 4950 rpm

        Any minute later, the ambient temperature higher (because the cover was opened):

        FAN1: 8775 rpm
        FAN2: 8925 rpm
        FAN3: 9000 rpm
        FAN4: 5025 rpm

        Later, temperature higher:

        FAN1: 9150 rpm
        FAN2: 9300 rpm
        FAN3: 9225 rpm
        FAN4: 4950 rpm

        Later, temperature higher, about 32ºC:

        FAN1: 9675 rpm
        FAN2: 9600 rpm
        FAN3: 9675 rpm
        FAN4: 4800 rpm

        * With 33 ohms resistor (unfortunately I didn’t have enough time to do more tests than one):

        FAN1: 8175 rpm
        FAN2: 8250 rpm
        FAN3: 8250 rpm
        FAN4: 3675 rpm

        * With 39 ohms resistor (I heard any beep, but I don’t know whether it was the server or the small UPS I have):

        FAN1: 8100 rpm
        FAN2: 8250 rpm
        FAN3: 8175 rpm
        FAN4: 3675 rpm

        One minute later, temperature higher:

        FAN1: 8550 rpm
        FAN2: 8550 rpm
        FAN3: 8475 rpm
        FAN4: 3675 rpm

        Another minute later, temperature higher:

        FAN1: 8775 rpm
        FAN2: 8925 rpm
        FAN3: 9000 rpm
        FAN4: 2250 rpm

        Some minutes later, temperature higher, about 31-32ºC

        FAN1: 9225 rpm
        FAN2: 9300 rpm
        FAN3: 9225 rpm
        FAN4: 3600 rpm

        And, that’s it.

        I am indecisive between 22 ohms or 33 ohms resistor. For me 39 ohms resistor offer a unstable behavior and close to minimum rpm threshold in occasions.

        I can’t do more tests until September. I’m sorry.

        Thanks and regards.

  31. Rafal says:

    Hi all.
    I did this mod.
    Thanks for author for sharing this tutorial.

    Dell 2950 , 1 x e5320 , 16gb ram , 3 x sas.

    Status Probe Name Reading Minimum Maximum Minimum Maximum
    system board 1 3525 RPM [N/A] [N/A] 2025 RPM [N/A]
    system board 2 3675 RPM [N/A] [N/A] 2025 RPM [N/A]
    system board 3 3525 RPM [N/A] [N/A] 2025 RPM [N/A]
    system board 4 3825 RPM [N/A] [N/A] 2025 RPM [N/A]

    But with 22ohm/1Watt resistors on red wire.
    After that fans slow down from 7k rpm to 4k rpm.
    Idle temp is as before = room temp.
    But still i little hear that fans try little spin up and slowndown, every 5second.

    Do you guys have same noise?

    • caveman says:

      Mine is running around 3500 rpm and the noise is tolerable. Server is in the 24u rack inside my home office and I can work/read/function as normal without being bothered by the noise.

      Thanks for the comments. Keep it coming.

  32. I made the changes outlined by this post and I have been very pleased with the noise reduction. Like other people who have commented, I too was concerned about how my modifications would affect system temperature. As a result of this concern I developed a couple of ways to monitor the temperature to alert me if certain thresholds are exceeded. In order to share my findings, I’ve created two blog posts for different ways to monitor your server’s temperature. I hope this helps others who are considering this easy custom mod.

    Monitoring Host System Temperature with PowerShell
    http://idledeveloper.com/?p=24

    Monitoring Temperature With Raspberry Pi
    http://idledeveloper.com/?p=347

  33. Pieh0 says:

    Just installed 22Ohm resistors on the 4 fans, and it’s working great. Fans are going at about 4.5k rpm, and while you can hear it, i can hear the fans in my pc over the server.

    I tried to used 47ohm, and while it was silent, it also filled up Drac’s error log in 30 mins, and in response the server kicked on the psu fan to 100%.

    So thanks for the info, if only Dell had this as a default setting, and then turn the fans onto jet mode when it needed it.

  34. It’s over a week and the resistor was not too heat (I’ve use 47 ohm 1/4 W ressistor) .
    I’ve receved my 32 GB of RAM this morning, now let virtualise all ! (proxmox install and creation of VM and Container tutorial in preparation for my website)

  35. Dman says:

    What website is that under Figure 20 Electric Bill?

  36. Anonymous says:

    What website is in Figure 20 Electric Bill?

  37. Anonymous says:

    What website is that on Figure 20 Electric Bill?

    Thanks

  38. Julienth37 says:

    Hi,

    I’ve buy a Dell PowerEdge 2950 II. I want to slow down my fan like you do.
    I will try the resistor mod on the 4 system fan and on the 2 PSU fan with 47 ohms resistor on the black wire.

    I will get you in touch of result !

    Tanks, for your post caveman.

    • caveman says:

      Skip the power supply mod. Let it run with the 4 fan mod first and see if u are satisfied with the noise level.

      • Julienth37 says:

        I will try the mod this afternoon and post myfeedback there.
        The server was in a shop with 2 person working everyday and client, if the noise must don’t annoy them, I think I don’t touch PSU).

      • Julienth37 says:

        Mod done ! (with 4 47 ohms resistor)
        The silence was veryyyyy cool !
        Next step buy 32 GB of ram (actually 4 GB was a bit small for the vitualisation by KVM and OpenVZ container).
        I will write a article on it (on my website there -> http://www.julienth37.tk , it’s in french sorry for you english guy’s).

  39. t.l.h_1234 says:

    after the resistor modification I get E1310 error. how do I determine if the BMC needs to be flashed or if the mod is not good, i.e. the fan is not working?

    • caveman says:

      Step By Step:
      1. Check if all the fans are working
      2. Update to the lastest BIOS and Firmware
      3. Install Dell Open Manage software, it will show the fan RPM threshold and current reading.

      BIOS Version 2.7.0
      Release Date 10/30/2010
      Baseboard Management Controller 2.50.00
      Remote Access Controller firmware 1.33.00 (Build 17)

      My failure threshold is 2025 RPM
      Current Fan Reading is ~3100 RPM

      NOTE: You don’t have to modify the power supply.

  40. t.l.h_1234 says:

    which is better: replace the fans update the bios or add the series resistor?

    • caveman says:

      It depends on your skill level and wallet size. I did not want to spend anymore more $$$ on servers and I know how to solder and besides I wanted to do some experiment. It turns out, adding a resistor was a success and my system is still running with no issues. Fan replacement is easy since it is just a simple swap but the glory is not comparable when you modify something on your own and watch it collect uptime. Do the resistor mod if you are comfortable soldering.

      • t.l.h_1234 says:

        the following is my setup: do you see any issue with overheating if I opt for the resistor modiifcation?

        Model 2950 Dell PowerEdge 2950 Server III
        Processor 2 2x Intel Quad Core 3.0GHz 12M L2 Cache X5450
        Memory Installed 32 GB 32GB Fully Buffered Memory
        Memory Size 4 GB Eight Slots Available: 8 x 4 GB Memory Sticks Installed
        Hard Drives Included 2x 146GB 15K SAS Hard Drive
        Riser Board PCI(e) Riser Board with two PCIe port
        External Media 1 DVD-ROM
        Power Supply 2 Redundant 750W Hot Plug Power
        Backplane 6 6x 3.5″ SAS/SATA Drive Bays (2 drives, 4 empty bays for you to use for future drives or blank fillers)

      • caveman says:

        My setup is 8 x 2.5″ drives and 32gb memory with no issues. It’s been running 24/7 since I did the upgrade.

  41. Joseph says:

    Will this mod work for the 2950 as well is is the 2950 iii different?

    • caveman says:

      Open the your 2950 server and compare the model and size of the fan. I have an old 2850 and it has the same fan as my 2950 III.

      Theoritically, to reduce the fan noise, one has to reduce the fan RPM and to reduce the fan RPM, one has to reduce the voltage to the fan motor, hence a resistor is installed to reduce the voltage from 12V to somewhere around 7-8V.

  42. moto211 says:

    Are you confident that 1/2w resistors will hold up over time? The closest consumer fans I can find to the stock Dells are these: http://www.futureelectronics.com/en/technologies/electromechanical/thermal-management/fans/Pages/9036179-AFC0612DE-AF00.aspx?IM=0 and they’re rated for 18w.

    • Costin says:

      I used two resistors of 100 ohm / 0.5 Watt in parallel and they work just fine.
      There is some warmer area around the memory banks, but it’s not excessive. I don’t think there is any issue.

      The fans are about 5$ a piece on e-Bay if you want to have a backup plan. Just make sure they look exactly like yours.
      You don’t need to change the power supply fans, those are not noisy at all. I also bought some power supplies but ended up not using them.

      Be aware that even with a resistor in series at power on those fans are making noise, although much lower than without a resistor. It is perfectly acceptable to me, although it’s noticeable louder than my PC.

    • caveman says:

      If the function of your server is mission critical, I do not recommend changing anything inside the server that is not fully researched and tested. Buy a lower RPM fans just to be safe.

      In my case, I needed to cut down my electrical energy cost and running 3 servers 24/7 was costing me around $60.00 a month and so I was desperate to consolidate my server needs into just one unit. At first I was very hesitant of the idea but then I remember that I’ve already tested this “resistor in series” mod on an old PE1550 for around 3 months a few years back and so I took a calculated risk.

      The server is inside my home office just about 6 feet away from where I sit and the noise doesn’t bother me at all. Sometimes by looking at it, I feel satisfied knowing that I was able to re-purpose a loud server for home use.

      Am I confident if the mod will hold up with the passage of time?

      I don’t know yet. The server was modified at around August of 2013 and has been running since. It is now February [2014] and if > than 5 months is a good number, then I will say I am confident.

      I’m running Hyper-V 2012 on it and even if I lose this server because of the resistor mod, getting another cheap one is not really an issue. I can just migrate the drives easily.

      It’s up to you to proceed with the mod, I think other people were successful also using a resistor but just not adding comments here and conveyed their experiences.

      I was worried at first but as the day goes by, my confidence was getting bigger. Now, I have an SQL Server 2012, Sharepoint 2013 and an Exchange Server 2010 running 24/7. Also, this server holds all my documents, pictures etc. I do ran a biweekly backup of my data and .vhdx just to make sure I am ready for a good disaster recovery training.

      Fan failure or not, I think it can be fun executing my disaster recovery.

      Please note that I have the Poweredge 2950 III with 8 – 2.5″ SAS drives and 32GB RAM, 2CPU and 2 Power supplies. I’ve modified 1 power supply only since the noise level was not really that bad. Also, I’ve learned from my month long Kill A Watt study, the cost to run a dual power supply compared to just one was the same.

      • moto211 says:

        My main concern is that when resistors fail, they do it in blaze of glory. Seeing that you’ve been successful so far with 47ohm 1/2w resistors, I decided to give this a try except with 3w resistors. I figure that will damn near eliminate the possibility of them going poof.

      • caveman says:

        I had the same feeling. I truly understand your point. Yes, during the first few days, I was thinking what if the resistor blows up [since it is hot] so in my first experiment, I placed the server in my garage thinking that if it catches fires, it will only burn the garage [which my thinking was wrong]. But then I did some deep thinking and simulating a blow up resistor, I’ve concluded that it will not catch fire, there is not enough fuel to ignite [something like that..]. My worries vanished after 1 week and the temperature of the resistor did not really do any harm. So, since you are a scientist [in this matter], why don’t you do your own trial and error experiment and find out what really will happen with a very warm resistor. I think overall, we are really on the safe side. Sometimes, we just have to jump in the water and find out if we will sink or not.

        Please don’t forget to blog about your experiences regarding this little project. Let us know your observations and learning. I have another server to be modified but I’m still debating on what to do.

      • moto211 says:

        I have done contained resistor failures before and a majority of the time it’s just a flash and some smoke. Looks spectacular but low likelihood of catastrophe. If the improbable is going to happen though, it will happen to me. So, I figure that up rating the resistor is a good preventative measure.

        I don’t blog, but I’ll definitely post back with my results.

      • paul says:

        Would using a lower RPM fan reduce the noise level to something comparable to the resistor mod?

      • caveman says:

        Yes, I believe so. I bought a replacement fan for my daughter’s Dell Optiplex 755 http://www.amazon.com/Dell-Computer-Assembly-Connector-Compatible/dp/B00GK00N24/ref=sr_1_4?ie=UTF8&qid=1392176218&sr=8-4&keywords=optiplex+755+fan and it was super quite. It will fit on PE2950 [size was the same but RPM was lower] but the cost was around $20.00 x4, so it was not an option for me. The resistor mod was only around $2.00. Maybe you should try it and let us know if you can find a comparable fan with a better and safer result.

  43. Pingback: Monitoring Host System Temperature with Powershell. | Idle Developer

  44. Dave says:

    Get a HP DL380 G6 (search ebay) – things run so quite you start to wonder if they are even on!

    • Anonymous says:

      Except they dont, Running said server always at 78% fan speed. Tried various ILO firmwares, shit man im lost.

  45. Costin says:

    Thanks for the solution, worked perfectly for me.
    I bought from craigslist a Dell 2950 third gen. at a price that I could not refuse and this fan trick helped me keep it.

    I noticed that the noise does not come up much from the power supply and actually I am running the server with one power supply that is not modified. The noise is perfectly acceptable, although it’s higher than the level of my desktop computer.

    I also used 1 Watt resistors and they fit perfectly fine.
    I could not see an improvement in using the power supply with resistor and without so I decided to skip it for now just for safety, I noticed somebody mentioned a burnt power supply.

    The server has 16 Gb of RAM and all the memory banks are full.
    I noticed the box is a bit warm, not exaggerated but I think that’s expected. I am considering putting a small silent fan over the memory cover some time later.

    • caveman says:

      Good to hear your report.

    • dcg says:

      I skipped modifying the PS also.
      I do see periodic inlet temperature problems if I let the house get above 80 degrees, but that’s obviously a problem with where I’m putting the unit.
      When I leave on vacation, I’m going to pop 2 unmodified fans back in (out of the 4).

  46. Pingback: Harrier Jumpjets! | Hexland's Blog

  47. dcg says:

    I find that 32ohms on the chassis fans (left PS’s alone) can result in “warning” level temperatures when under heavy load in a relatively isolated (not great airflow) area… Just FYI..

  48. Unbalanced says:

    Yep, resistors were soldered on the red wires. If temps are higher on the PSUs, I’m sure it wouldn’t take much for a component, cap or whatever to fail, especially since the server was used.

  49. Unbalanced says:

    47 Ohms 1/2 Watt resistors may be a bit too much for the power supplies. I had one fault a day after the mod. Since the PSUs are really cheap on ebay, (about 10 bucks per PSU) I may attempt using a pot for resistance and dial in the rpm. The mod for the chassis fans work great. Temps stay cool, and no problems what so ever! Thanks for the tips with the mod! I may do this to my poweredge C2100!

  50. I just wanted to thank you for this mod. I was about ready to put this 2950 back on eBay because of the noise. I modded all the fans and not even the power supply, pretty simple actually and I can barely hear it running in the next room .

    Thumbs up….A+

    Kelvin

    • caveman says:

      Glad you were able to find this project useful.

      • Anonymous says:

        No Doubt. Just had a quick question. I installed that Core Temp App I see in your pics to check my temps. My numbers are slightly higher than what yours are in the pick. Do you think adding the 47ohm resistor to the PS would help bring those numbers down at all?

        Kelvin

      • caveman says:

        Its worth a try but the task for the psu resistor add on is a little bit complicated. On mine I only added a 47 ohm to one. I end up using dual psu since the electrical consumption was not a factor. Also, please note that I have the Pe2950 III with 8 2.5″ drives.

      • Anonymous says:

        Great. I think I may try that this evening. I have the exact same box as you with the 2.5 drives Gen 3. except I only have 4 drives. Thanks again. Lifesaver with this one.

  51. dcg says:

    Have any information on fan RPM after installing the resistors? I’d like to know how fast they are turning.. You should be able to see this info in openmanage or ipmi-sensors (linux).

    To the guys that are reporting fan throttle up and down:
    Apparently the BMC firmware kicking up the fans is a known issue. If the fans drop below a set point, the machine thinks a fan has failed and will throttle up to compensate. On my 2950, I believe the latest version of firmware is 2.5.0.
    To fix, you need to hack the BMC firmware – I did it in Centos 6.4 and it’s supported in Unbuntu (as a no-install-CD) to reduce the minimum threshold that triggers panic and fan throttle up.
    Do a web search – there is a python based tool to help…..

    • caveman says:

      Current Reading 20131206
      Fan1=3300 rpm
      Fan2=3450 rpm
      Fan3=3450 rpm
      Fan4=3075 rpm

      Min RPM is 2025 rpm BMC v2.50.00 [original from Dell, no hack]

      • dcg says:

        Thanks.. I used 2 x 68ohm resistors in parallel to deal with the watt load on the resistors and keep them from getting hot. Resulting resistance is 34 ohms.
        I get some level of fan cycling – RPMs are currently at:
        Fan1=3300 rpm
        Fan2=3450 rpm
        Fan3=3450 rpm
        Fan4=3075 rpm

        That is, they’re probably spinning through 600 rpm differences all the time. You can hear it, but it’s no where near the screaming jet that it used to be.
        No mod to power supply fan.
        53 dB outside the case…

        System board ambient temps are between 31-33 degrees C, stable for days now. However, my system doesn’t have much load on it.
        Running 6 x 3.5″ drives.

        I modified the firmware and have it set at 975 rpm minimums. If the fan cycling annoys me enough or if this thing ever gets hot (it’ll send me email through OpenMange) – then I may look at installing aftermarket lower-rpm fans.

      • caveman says:

        Good to hear that you were able to get what you wanted.

  52. Hi !!

    Congratulations: It Appears to be a great job !!!

    One question: I couldn’t read anything refered to the number of processors installed. So, I wonder, prior to make it with my PE:

    Would be suitable or convenient this solution in case of two processors installed?

    With Many thanks !!

  53. daviddob says:

    After updating all of the firmware to the newest versions and reflashing the bios and bmc everything seems to be working great. I decided to go with the 33ohm 1/2w resistor and my idle fan speeds are between 2900-3200rpm according to ESXI health and status. For those of you running esxi what do you use to monitor CPU/ram/hdd temps? My esxi only shows the ambient case temperature(22C) and i was wondering if there was anything that could get more specific as case temp can be unreliable.

  54. Rob says:

    I am putting the treatment on a 2950 this evening.

  55. David says:

    I tried using 47, 30, and 20 all resulting with that error and the fans kick up to 100% I am trying this with 2 servers both with the newest bios (2.7.0) is there anything I am missing? Is there a way for you to send me a modified BMC that I can flash to it? What do you recommend I do at this point? I do not understand why it isnt working as the 3 resistors seem to work well in slowing the fans to different degrees.

    • caveman says:

      Confirm the following:
      1. Go to support.dell.com and update all bios related updates, I did this on mine.
      2. Install dell open manage software so that you can view what fan is below the threshold and how much.
      3. You placed the resistor on the “red” wire and made sure that the other wires are intact. The other wires are for the rpm monitor. If its not giving any rpm, it will give an error
      4. I have Pe2950iii and i did not have to do any BMC software mod.
      5. Remove all the resistor and do 1 fan at a time and observe error and fan speed.

      BIOS Information:
      BIOS-2.7.0, 10/30/2010
      BMC-2.50.00
      Remote Access Controller-1.33

      System Board FAN 1 RPM Reading-3375 RPM Min-2025 RPM
      System Board FAN 2 RPM Reading-3450 RPM Min-2025 RPM
      System Board FAN 3 RPM Reading-3450 RPM Min-2025 RPM
      System Board FAN 4 RPM Reading-3075 RPM Min-2025 RPM

      Fan Status: https://dl.dropboxusercontent.com/u/80421698/PrintScreen/20131026001.jpg

  56. David says:

    Yes it is up to the latest bios and when booted the fans are nice and quiet, until the server notices the low RPM and the light flashes yellow, the psu fans ramp up to 100%, and the error E3010 is displayed E1310 – RPM Fan ## (Causes: RPM of specified cooling fan is out of acceptable operating range.)

    • David says:

      not E3010, its E1310 sorry

    • caveman says:

      What is value of the resistor you are using? You need to decrease the resistance (more voltage) to the fan. I am using a 47 ohms resistor, Try 40, 30 and see if the rpm is reading in the lower threshold of the bios settings.
      10v=12 ohms
      9v=20 ohms
      8v=30 ohms
      7v=42 ohms
      6.5v=50 ohms [6500 rpm]

  57. David says:

    I installed the resistors and they quiet the major fans down a ton, however the server notices the drop in rpm and the power supply fans kick up to 100%. Is it possible to put a resistor on those fans and have the system still boot, or will doing so cause the server to fail to boot? If that is the case does anyone have a hacked BMC for a dell poweredge 2950 II they could send to me as I am not a proficient Linux user and I work mainly with hardware.
    Thanks in advance.

  58. I love you for this!!!!! I had the Ebay ad ready before I saw this post! Thanks again. easy to do no headaches with trying other fans. Thanks again

  59. Jordan says:

    I am planning on doing something like this to mine as well as the noise is aweful. I was wondering if the resistors get really hot though? And if so, did you take any special measures to insulate them from other parts?

    I don’t think the second power supply will cause you to use much more power. I have the same server as you with 5 2.5″ drives and both power supplies plugged in to the same UPS. It averages around 250-300W under a medium load running 6 VM’s. It appears to balance the power draw across both supplies. I experimented running each one from a different power source, one on a UPS and the other on an inverter that runs all of our switches and routers and is connected to a huge Surette battery. It seemed to work fine and draw from both supplies simulatniously.

    I would base your solar calculations off of your average consumption instead of the maximum draw of 780W. It would need to me running at 100% with all 8 drives to come any where near that figure.

    Thank you for the great write-up.
    Jordan

    • caveman says:

      Jordan,
      I had 6 objectives when I started this project
      1. Build 1 server to host Hyper-V 2012 [turn of the rest]
      2. Find a cheap server that has VT support.
      3. Reduce electric bill in the house
      4. Reduce noise in the home office
      5. Reduce heat production in the home office.
      6. All actions must have zero contradictions – meaning there should be no undesirable effect of all the actions that I execute on the project.

      I guess I forgot to mention/describe heat dissipation of the 1/2 w 47 ohm resistor [thanks for reminding me since this is the most important aspect of the project, a killer I should say]. The resistor is not super hot, just warm and comfortable to hold with your fingers for a long time. I did this mod on PE1550 long time ago that I placed in the garage and I was checking daily the temperature of the server and fan health but everything was ok after 3 months. I did came up with a brilliant solution –
      1. extend the red wire by 3″
      2. add heat shrink [yellow in picture 6, https://blindcaveman.files.wordpress.com/2013/08/image9.png, I did not use an electrical tape because the adhesive will melt overtime.
      3. place resistor on top right side of the fan
      4. make sure resistor is not touching the cover when you close it.

      When you do your own mod, study mine but try to improve it. Mine is already robust and tested but it was not pretty and took a long time to do. I did not have time to triple think my actions and I just execute the tasks without putting much effort of the what ifs.

      Things in mind:
      1. Why not use black instead of the red wire. This will be safer if the black wire touches metal instead of red – no short.
      2. Extend all the black wires and connect it to a black box with all the resistors in breadboard with override switch.

      Regarding power consumption, I’m experimenting with your idea of running two power supplies instead of one using a Kill A Watt. Currently, the server consumed 189kW for 640 minutes of uptime. In Texas, this cost me $19.89. According to my spreadsheet,
      Cost per hour=$0.03
      Cost per day=$0.74
      Cost per month=$22.33
      Cost per year=$267.00
      I am happy with these numbers but I’m just curious what good or bad thing this 2 psu does to me. I noticed an increase on the fan noise [but not annoying] since I did not modify the 2nd power supply. Having 2 PSU might decrease the heat dissipation, who knows. Ok, more experiment on my plate.

      By the way, this is not a production server. This is only a server for learning purposes and home use. My grand plan for this project is to operate in a remote mountain somewhere in Asia using Mesh Network technology. More on this later. I’m still researching more about solar power and off the grid setup.

      If you guys has a cheap solution/idea to run a poweredge 2950 server off the grid, please let me know.

      ra

    • caveman says:

      Jordan, you were closely right about the two power supply consumption: Kill A Watt showed a neglible increase of kW used. I’ve only tested/collected for 5 hours but I think this is conclusive. Not much electrical consumption difference when running 2 PSU instead of just 1.
      The benefit are obvious:
      1. PSU Redundancy
      2. Better internal cooling since the 2nd PSU draws airflow out from the server.
      3. 1 PSU will not be overloaded, which means less heat dissipation and cooler PSU.

      With 1 PSU
      Cost per hour=$0.03
      Cost per day=$0.74
      Cost per month=$22.33
      Cost per year=$267.00

      With 2 PSU
      Cost per hour=$0.03
      Cost per day=$0.77
      Cost per month=$23.18
      Cost per year=$278.21

      • Jordan says:

        Hey Caveman,

        Sounds good! Mine has been running great at about 36 degrees. If you need any help with your networking let me know, I administrate one of the largest private networks in Canada 🙂

        Take care,
        Jordan

      • caveman says:

        Mission accomplished!. Good to know. I’ll let you know if I have questions regarding admin stuff. Thanks.

  60. Richard says:

    @caveman – I have just bought a secondhand 2950 and woke the whole house when I booted it up the other night (with apologies to the baby in the next room). I could not believe this thing!

    So I’m off to replicate your fine work here, I just have a quick Q – do you plug in both power supplies? From reading this post I assume you only have one in use (with the fan on it running at lower speed).

    I’m becoming a little power obsessed due to electricity charges over here (I’m in Australia) – I’m also planning on removing some of the 6 drives until I need them (this will be a sandpit machine running linux)

    • caveman says:

      Richard,
      I am using 1 power supply since one power supply registered a 0.29kwH on the Kill A Watt meter. My cost per day in Texas, USA is $0.73 per day or $21.92 per month. I have not inserted both psu to see how much the two powersupply will cost me per month since there is no purpose for me. I figured,I don’t need a power redundancy here since power is very stable.

      I have the Small Form Factor 2950 III with 2.5″ SAS drives so the heat generation is less, You have the 3.5″ so its going to be warmer. Experiment on this and see if removing drives can affect your noise level and temperature. There are 2 SATA port in the motherboard that you can use to run the system if you need to. I’m running 1 SATA 3.5″ externally to add a 500GB space. SAS drives are costly compared to SATA.

      Take your time in your mod and maybe you can improve on what I did. The 47 ohms resistor works and my system is running just fine.

      Yes, this thing is very loud but I got it in control.
      1 PSU 47 ohms [leave the spare as is just in case there is an issue in the future.
      4 FAN 47 ohms

  61. in addition one of the heat-sink fans failed. a month ago. yet to receive a replacement from my supplier and am sure after replacement the temps will reduce.

  62. Blind CaveMan! thank you so much! i did this over the weekend and now my server is as quiet as the cemetery. 70% of the jet noise generated by the poweredge 2950 within the office is gone. currently monitoring the temps and they are looking good. Thanks.

    • caveman says:

      Do you have the 2.5″ drives on your server? Mine is ok because I have the smaller drives in it (8 total). Please monitor the temperature closely. Mine runs at average 50’c in both CPUs. It’s been running almost a month now in my home office and very quite.

      • four (4) 3.5″ drives in raid 5 mode. the server sits in a small room with a dedicated or always on AC. i have been monitoring the temps since and my average is 57’c on both CPUs. i opted for this solution because i have done everything( bios update, replacement of all fans, blow out dust etc.) to reduce the noise but to no avail.

  63. caveman says:

    Everything works fine. Here’s the update report at Day013. https://blindcaveman.wordpress.com/2013/09/05/update-poweredge-2950-iii-fan-mod-day013-uptime/ . It is running 24/7 as my only server at home.

  64. Anonymous says:

    Hi there
    Any issues with the server since adding the resistors?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: