Need Supermicro replacement parts, where to get for reasonable price

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Markess

Well-Known Member
May 19, 2018
1,146
761
113
Northern California
I recently needed a "Rear Window" for an SC836. I found only a few places on the Internet that listed it and they were all around $200 and all were drop ship from the Mfg. I said WTH and contacted Supermicro direct. They sold it to me for $15 bucks and change plus shipping. Lesson learned, if it's chassis parts or other odd parts, go directly to the source.
Definitely the best idea for a whole lot of Supermicro parts and small bits. I think @lpallard already tried that though and the Supermicro Store won't ship directly to them in Canada.

@lpallard, have you considered buying from the Supermicro Store using a forwarding service? I've seen a few people on the forums here that regularly use a forwarder for their tech purchases.
 
  • Like
Reactions: Fritz

lpallard

Member
Aug 17, 2013
276
11
18
Hello guys,

So I've got a replacement fan (FAN0126L4) to swap the "defective" middle wall fan. Didn't go as planned... Like they say "the plot thickens".

First I opened the chassis and tried to find the root of this high pitch noise... After 10 minutes or so using a paper cone in my ear canal, the noise seemed to come from "everywhere". For the record, I dont have tinnitus. :)

Then the only solution to me was to swap the "most likely" whining fan, and see if it helps. It didnt. Then I swapped another fan with the one I had removed from the chassis. Still didnt help. Finally, I swapped the last fan. Didnt' help.

On top of that, the replacement fan ordered from ebay is WHINY as *****. Cant use it. It's easily 3 or 4 times louder than the actual whine....

So I put back the stock fans, and replaced the CPU cooler fan thinking it must be it. After all, there are only 4 fans in this chassis, the two exhaust fans are unplugged.

Doesnt help.

To be honest, I'm a bit at wits ends. This interferes with my sleep as this noise is getting through EVERYTHING..... My girlfriend also can hear it so I know I'm not losing my mind ;)

I just ordered 3 NEW fans from Amazon. Somehow a reseller has listed FAN-0094L4 on amazon. Lets see after I swap all of them for new ones. If I still hear this noise, there will be a chassis for sale and I'll replace it with something "off the shelf". Not excited about this but this is enough.

But that's not all.... Another real strange thing happened when I was doing all of that. To replace the CPU cooler fan, I shutdown the chassis. Its running Freenas. So I used the WebUI to shut it down. Immediately, one of my UPS went self testing, and my pfsense router rebooted!

That's the first time I see that.

There are 3 machines in this rack. A freenas server (the whining machine), a proxmox VE server, and a pfsense router (the other whining machine that got fixed...).

There are 2 identical UPS's. Both are APC SMT750RM2U's. Freenas is connected to one of them, and the proxmox and pfsense machines are connected to the second. Both UPS's are connected to the same circuit breaker. Both power modules of the freenas chassis are connected to the same UPS.

Could it be some kind of surge? What would cause a UPS to enter self test exactly at the moment I press "Shutdown" in Freenas WebUI, then cause a reboot of a machine connected to a different UPS?

For those wondering, yes I'm already considering an exorcist...:eek:
 

Fritz

Well-Known Member
Apr 6, 2015
3,371
1,375
113
69
I have a 2U Quanta server that does something sililar. It goes wha wha wha wha, etc. etc. I replaced every fan in the chassis and it made no difference. Finally just gave up and let it whine.

An UPS will always give you trouble. Not a question of if but a question of when.
 

ramblinreck47

Active Member
Aug 3, 2019
142
56
28
Hello guys,

So I've got a replacement fan (FAN0126L4) to swap the "defective" middle wall fan. Didn't go as planned... Like they say "the plot thickens".

First I opened the chassis and tried to find the root of this high pitch noise... After 10 minutes or so using a paper cone in my ear canal, the noise seemed to come from "everywhere". For the record, I dont have tinnitus. :)

Then the only solution to me was to swap the "most likely" whining fan, and see if it helps. It didnt. Then I swapped another fan with the one I had removed from the chassis. Still didnt help. Finally, I swapped the last fan. Didnt' help.

On top of that, the replacement fan ordered from ebay is WHINY as *****. Cant use it. It's easily 3 or 4 times louder than the actual whine....

So I put back the stock fans, and replaced the CPU cooler fan thinking it must be it. After all, there are only 4 fans in this chassis, the two exhaust fans are unplugged.

Doesnt help.

To be honest, I'm a bit at wits ends. This interferes with my sleep as this noise is getting through EVERYTHING..... My girlfriend also can hear it so I know I'm not losing my mind ;)

I just ordered 3 NEW fans from Amazon. Somehow a reseller has listed FAN-0094L4 on amazon. Lets see after I swap all of them for new ones. If I still hear this noise, there will be a chassis for sale and I'll replace it with something "off the shelf". Not excited about this but this is enough.

But that's not all.... Another real strange thing happened when I was doing all of that. To replace the CPU cooler fan, I shutdown the chassis. Its running Freenas. So I used the WebUI to shut it down. Immediately, one of my UPS went self testing, and my pfsense router rebooted!

That's the first time I see that.

There are 3 machines in this rack. A freenas server (the whining machine), a proxmox VE server, and a pfsense router (the other whining machine that got fixed...).

There are 2 identical UPS's. Both are APC SMT750RM2U's. Freenas is connected to one of them, and the proxmox and pfsense machines are connected to the second. Both UPS's are connected to the same circuit breaker. Both power modules of the freenas chassis are connected to the same UPS.

Could it be some kind of surge? What would cause a UPS to enter self test exactly at the moment I press "Shutdown" in Freenas WebUI, then cause a reboot of a machine connected to a different UPS?

For those wondering, yes I'm already considering an exorcist...:eek:
Bad PSU? You mentioned all kinds of fans but the one(s) in the PSU.
 
  • Like
Reactions: Markess

Markess

Well-Known Member
May 19, 2018
1,146
761
113
Northern California
So I've got a replacement fan (FAN0126L4) to swap the "defective" middle wall fan. Didn't go as planned... Like they say "the plot thickens".
Could be that @ramblinreck47 is right. Maybe try to unplug all the fanwall fans, turn it on, and let it boot up to idle. Its not going to explode if you run it without fans for a couple minutes. See if you still hear the tone/sound/subsonic tingle/whatever. If you do, its obviously the PSU...or the poltergeist that's haunting you... and not the fanwall.
 

lpallard

Member
Aug 17, 2013
276
11
18
Bad PSU? You mentioned all kinds of fans but the one(s) in the PSU.
I dont think so. This "whine" has been present way before I swapped the power modules a month ago to PWS-920P-SQ's. Moreover, when the machine is powered off (but still connected to AC to keep IPMI alive), the whine disappears. AFAIK the power modules fans are still spinning, although maybe not at the same speed.... The power draw is also very different.

I'll try pulling all wall fans and see if this is still there. I've been thinking lately if this whine could be electrical related? Like a capacitor or something else on the motherboard, backplane or PDB? These are all from original 2008... AFAIK electrical whine is usually constant and not alternating but I may be wrong...

For the UPS issue (unrelated), I will contact APC directly. The only thing I see is that the Freenas machine has 920W power modules which are way higher than the UPS's capacity of 500W. So perhaps doing the shutdown caused a sudden surge to PSU's capacity (920W) which tricked the UPS into self test? I dont see why that would happen but these things are so complicated..... Finally, I also think about a grounding issue with the rack...

Do you guys have anything special for grounding? All of my components (servers, UPS, switches, etc) are simply installed in their rails. No grounding cables anywhere but I know the rack came with grounding cables but I didnt install them...
 

Markess

Well-Known Member
May 19, 2018
1,146
761
113
Northern California
I dont think so. This "whine" has been present way before I swapped the power modules a month ago to PWS-920P-SQ's. Moreover, when the machine is powered off (but still connected to AC to keep IPMI alive), the whine disappears. AFAIK the power modules fans are still spinning, although maybe not at the same speed.... The power draw is also very different.
I'd forgotten that the noise was present with the old PSUs. As an FYI, I don't think the fans on the SQs spin when system is off. Mine don't seem to spin at least. You'll still have a power draw when off to power the BMC of course.

I'll try pulling all wall fans and see if this is still there. I've been thinking lately if this whine could be electrical related? Like a capacitor or something else on the motherboard, backplane or PDB? These are all from original 2008... AFAIK electrical whine is usually constant and not alternating but I may be wrong...
If its electrical, then I'd think that you physically pushing on the fan wall shouldn't make it stop.

Although if its resonance from something in the case, even electrical, I suppose the fan wall could be the culprit without it being the fans themselves?

Two other things to consider:

1. You mention you have a Noctua CPU cooler. Noctua fans can create problems, both because they can run slower than the system is expecting, and because they can intermittently register as Zero RPM for brief periods (think fractions of a second). The first condition causes the system to (usually gently) ramp up the fan till it reaches the minimum threshold set in the BMC. If the minimum is only a few RPM higher than the fan is at, its just a blip, but it can be a continuous cycle of up/down of just a few RPM and the doplering can seem to be constant. The second condition causes the fans to go to full briefly till the RPM registers again. Don't think the second instance applies here, but the first could maybe. Long shot to be sure, but something to check if you're out of ideas.

2. Have you tried manipulating the fan duty cycle via ipmitool? I think from an earlier post you're running FreeNAS? ipmitool is included and there's a few posts here on the forum on how to use it to manipulate fan speeds. If may be possible to find a duty cycle that spins the fans at an RPM that doesn't create the resonance...either a little higher or lower than current.

For example, with fan profile set to "Optimal", my boards set the duty cycle at 30%. On one of my systems with a super low power Xeon (E3-1220l with an active cooler), idle draw for the whole system was less than 20w and even with it straining its little Xeon heart to the maximum, it didn't break 40w. So even 30% was overkill and I was able to put it down to 12% with all the temps still staying down near ambient. Set that way, the system was dead silent from anything more than a foot or so. In your case, you may be able to find a fan speed that doesn't cause the vibration you're hearing.

If you try ipmi tool and find a duty cycle that eliminates the resonance, you can set up a boot script to set the desired fan profile at boot (any changes you make while testing won't persist when you reboot).

Cheers!
 

lpallard

Member
Aug 17, 2013
276
11
18
Trying to make progress. I replaced all fans of the fan wall one of them got real noisy lately and I could put my hands on 3 brand new genuine Supermicro fans from a local IT supply store for 10 bucks each. Fortunately some noise is gone, but a constant high pitched noise remains. I'm strongly thinking about the PSU fans.

However I read that the PWS-SQ's fans do not spin at low load. My power draw is < 120W (UPS says 115W, ipmi tool shows 65W for PSU1 + 68W for PSU2).

I need to play around with the fan speeds to see if this tone disappears or not.

Some questions before I play around with fan profiles via ipmitool:
  1. See "PSU" screenshots from the IPMIView tool (SUPERB tool from Supermicro). I assume Fan 1 and Fan 2 (under PMBus tab) are because these PSU's have a double fan arrangement (contra-rotating?) Can someone confirm these readings are indeed for each of these fans?
  2. Under "FRU" tab, I see additional values for fan speed, and they do not match those under "PMBus". Why's that?
  3. I'm confused about system fans. There are 3 fans in the mid fan wall, and a CPU fan. The exhaust fans are unplugged. I see 4 fan readings in the IPMIView tool (FAN1, 2, 3 and FANA). Fan 2-3 & A are all reaidng 2400RPM. I assume they are the 3 fans of the fan wall and FAN1 is the CPU fan? Am I correct to assume the fans in the mid wall would have the same speed settings?
  4. Unless I am not interpreting the system fans properly, I do not see the PSU fan speeds in the system fans. Can the PSU fan speeds be changed or are they controlled purely by temperature/load circuitry inside of the PSU's?
Have a great day!

EDIT: I made some recordings and analyzed the soundtrack a bit differently after talking with some folks. I also want to mention that I am posting as a follow-up to my original post and for those who may be interested as this applies to anything vibrating and/or making noise...

I filtered band by band in Audacity until the "high pitched noise" stands out clearly from the recording, or until the sound is totally identical to what I can hear in real life. The band that produces the same sound (so frequency is isolated) is in the 1100-1300Hz band.

Then spectrum plotting the unfiltered recording also shows a "peak" at 1148Hz. Awfully close to my previous plot which had a peak at 1150Hz...

Theory 1: coil whine. 1200Hz is 20x grid frequency. Since this noise has been there for a while now, maybe motherboard or backplane? However, it doesnt "sound" like coil whine to me but more like a bad bearing (sound is not constant and seems to go up & down in dB and oscillate randomly).....

Theory 2: after all the noises I eliminated so far, perhaps this new pitched noise is "new" and therefore comes from a defective PSU. Maybe a bad fan bearing or something else

Theory 3: this one is far fetched but what the **** So the FAN 2, 3 and A are all reading 2400RPM in IPMIView. 2400RPM is 40Hz. If the "frequencies" are shifted pretty much equally, and have similar periods, the resulting frequency will be 3x40Hz = 120Hz. Something could be vibrating (like the fan wall bracket or something else in the chassis) and could be responding in a non-harmonic way (due mainly to stifness and/or mass) and respond with a frequency multiplier of around 10.

I need to shutdown this server, take it apart and run one fu*** fan and thing after the other.....
 

Attachments

Last edited:

uldise

Active Member
Jul 2, 2020
209
71
28
Fan 2-3 & A are all reaidng 2400RPM. I assume they are the 3 fans of the fan wall and FAN1 is the CPU fan? Am I correct to assume the fans in the mid wall would have the same speed settings?
depends on you fan mode used - if you choose "Heavy IO" mode, FanA, B will spins faster.
 

uldise

Active Member
Jul 2, 2020
209
71
28
you mean Fan A. Supermicro board have two fan sets: Fan 1, 2, etc, and Fan A, B - these ones are for IO.
 

lpallard

Member
Aug 17, 2013
276
11
18
See Post 48, I added some more fun things :cool: but I'm still curious about the IPMI and fan speed questions...
 

lpallard

Member
Aug 17, 2013
276
11
18
you mean Fan A. Supermicro board have two fan sets: Fan 1, 2, etc, and Fan A, B - these ones are for IO.
Okay, I read a bit about this yesterday and its clearer to me now. However, I'm a bit worried about the CPU cooler. @Markess mentioned that Noctua fans can misbehave with IPMI. I looked around to get a Supermicro 3U cooler for X9SCL motherboard (LGA1155) and couldnt find.

Any recommendations??

I also tried manipulating the fan duties with ipmitool using @PigLover's tutorial (very well written btw!) but got

Code:
Unable to send RAW command (channel=0x0 netfn=0x30 lun=0x0 cmd=0x70 rsp=0xcc): Invalid data field in request
 

uldise

Active Member
Jul 2, 2020
209
71
28
Noctua fans can misbehave with IPMI
You have an older X9SCL-F, right? if so, then yes, this old generation boards need faster fans to operate correctly they fan control. i have one X9 board and i have problem with Noctua's as described above - mobo thinks that fan is not running, and rumps up all fans to full speed. you can set fan thresholds to almost zero, but problem persists.

also tried manipulating the fan duties with ipmitool using @PigLover's tutorial (very well written btw!) but got
that's fine, not all X9 boards support these raw commands, mine too.
 

lpallard

Member
Aug 17, 2013
276
11
18
@uldise Yes I do have a X9SCL-F board. So is there no way to manually adjust the fan duties with ipmitool on these boards?

You seem to confirm that the noctua fan may be not best for these older boards. I however noticed when I opened up the chassis (didn't remember since its been many years I built this server) that the CPU fan (Noctua) has only 3 pins connector. So unless I am wrong, its running full speed all the time. Yesterday I removed all chassis fans, the thing is dead quiet. So the CPU cooler is not making any noise and neither are the PSU's or other electronic component (coil whine, etc).

I however discovered something interesting, and it may be the root cause of the noise and whining I've been experiencing in the last few months. When I removed the chassis cover, I could hear the three BRAND new chassis fans hum at a low frequency. They run at 2400RPM. I believe this "hum" is normal as it sounds like a small fan spinning at medium speed. Air makes noise also when its pushed by a fan....

The issue is when the cover is put back on. The hum seems to amplify a lot, and there's a high pitched noise coming off from the cover. The fan wall also vibrates a lot. I believe my "theory 3" of post 48 is close enough.... I will try to add some padding on the inside surface of the cover to see if it would dampen a bit the noise. @Fritz you may want to do the same if you've faced hard to pinpoint noises.

What I take from all of this: These things are not designed to be in your living space :D If I had a basement, they would have quickly been sent there....
 

uldise

Active Member
Jul 2, 2020
209
71
28
So is there no way to manually adjust the fan duties with ipmitool on these boards?
as far as i know, no.

that the CPU fan (Noctua) has only 3 pins connector
i always use 4Pin PWM Noctua fans. with 3pin fans you can reduce the fan speed with low noise adapter cable that comes with most Noctua's, until you use industrial series. last time i purchased 3Pin Industrial Noctua fans by mistake - they can work at one speed only. if you connect low noise adapter cable, fan won't spin at all.
but i agree that most Noctua's are silent enough even at full speed, unless speed are less than 2000RPM for 12cm fan for example.
 

Fritz

Well-Known Member
Apr 6, 2015
3,371
1,375
113
69
The worst fan noises I've ever tackled and still have is my Quanta 2U. The fan wall is composed of 4 double fans of an odd size, 60mm maybe. They go wah wah wah wah wah......... I ordered some off ebay and they made no difference what so ever. I think it may be a physics problem rather than a mechanical one. It's annoying as hell and can be heard from anywhere in the house. After it I decided only SM gear from now on. They're much easier to deal with / modify.
 

lpallard

Member
Aug 17, 2013
276
11
18
Anybody using this Supermicro CPU cooler? Specs on SM sites states fan dimensions of 92L x 92H x 25W (mm) while not providing overall dimensions. Amazon states overall dimensions of 14 x 19 x 12 cm.

I believe this will not fit in my 3U chassis. Current cooler is Noctua NH-D9L with HxWxD of 110x95x95 mm and barely fits in the chassis (about 2mm clearance with the chassis cover.

I couldn't find a Supermicro PWM 3U cooler for LGA-1155... Thats strange when considering the popularity of these sockets...
 

Fritz

Well-Known Member
Apr 6, 2015
3,371
1,375
113
69
Anybody using this Supermicro CPU cooler? Specs on SM sites states fan dimensions of 92L x 92H x 25W (mm) while not providing overall dimensions. Amazon states overall dimensions of 14 x 19 x 12 cm.

I believe this will not fit in my 3U chassis. Current cooler is Noctua NH-D9L with HxWxD of 110x95x95 mm and barely fits in the chassis (about 2mm clearance with the chassis cover.

I couldn't find a Supermicro PWM 3U cooler for LGA-1155... Thats strange when considering the popularity of these sockets...
That's a crazy price. It's cheaper on eBay.


Yea, I avoid 11xx for that reason and requiring ECC UDIMM's.