Another X9SCM bites the dust? No, PSU incompatible (Corsair TX750M)

[dikkiedirk]

I don't know what is wrong with QA/QC of Supermicro at the moment but I just got a call from my nephew, who is "infected" with the unraid-virus by me.

He decided to buy the same components as me: X9SCM-F, E3-1240 and 16 GB Kingston ECC RAM. He is using a different PSU though, an older Turbo-Cool 1200 from PC Power&Cooling. It has started one time without RAM and got 5 or 6 beeps indicating memory error. It started immediately after switching on the PSU without pressing the front-panelswitch. He switched off power and put the RAM on the mobo and switched on again, this time he had to switch on using the front panel switch. He got 3 beeps indicating normal POST. Mine acted just the same.

The LAN-cable for IPMI was plugged in.

 

But.....After some time the CPU-fan stopped like the system was switched off.

The solid and flashing LED are on but the mobo won't switch on anymore, either by front-panel switch or through IPMI.

 

Has another X9SCM-F bitten the dust? Or did the PSU die?

[Rajahal]

Take a look at the fan settings in the BIOS, maybe it is set to the 'power saving' mode.  I didn't think this mode would disable the CPU fan entirely, but maybe it does. It is also a very good idea to test the power supply if you have a power supply tester or multi-meter handy.

[dikkiedirk]

Take a look at the fan settings in the BIOS, maybe it is set to the 'power saving' mode.  I didn't think this mode would disable the CPU fan entirely, but maybe it does. It is also a very good idea to test the power supply if you have a power supply tester or multi-meter handy.

 

Unfortunately the mobo or PSU won't switch on anymore, so I can't enter the BIOS.

[dikkiedirk]

In what way must a server motherboard like the X9SCM and other components be treated differently than more mainstream components. By that I mean from assembling the individual components to first boot.

 

I have been working in electronics assembly so I know about ESD and precautions against it, but are there other "specialties" on a X9SCM that must be taken into considerations?

 

How likely is it that a motherboard fries a CPU or a CPU fries a motherboard?

 

 

[Johnm]

In what way must a server motherboard like the X9SCM and other components be treated differently than more mainstream components. By that I mean from assembling the individual components to first boot.

 

No different. Infact, other then a smaller set of "compatible hardware", they tend to be more durable IMO. We have had monkeys at work building up our servers on some days.

 

I have been working in electronics assembly so I know about ESD and precautions against it, but are there other "specialties" on a X9SCM that must be taken into considerations?

None.. ESD is a killer of PC's. Other then that,They are picky on RAM and need certain power requirements. Unclean power might fry a board. you did say the PSU is used. you never know what it has endured.

 

 

How likely is it that a motherboard fries a CPU or a CPU fries a motherboard?

Almost impossible. a PSU can fry both. Doing something stupid can fry one or both also. like hotpluging a PSU power 24, 8 or 4 pin in while powered on.

 

 

 

now that your rig is running. you could try his board or CPU or RAM in your system for testing.

[dikkiedirk]

In what way must a server motherboard like the X9SCM and other components be treated differently than more mainstream components. By that I mean from assembling the individual components to first boot.

 

No different. Infact, other then a smaller set of "compatible hardware", they tend to be more durable IMO. We have had monkeys at work building up our servers on some days.

 

I have been working in electronics assembly so I know about ESD and precautions against it, but are there other "specialties" on a X9SCM that must be taken into considerations?

None.. ESD is a killer of PC's. Other then that,They are picky on RAM and need certain power requirements. Unclean power might fry a board. you did say the PSU is used. you never know what it has endured.

 

 

How likely is it that a motherboard fries a CPU or a CPU fries a motherboard?

Almost impossible. a PSU can fry both. Doing something stupid can fry one or both also. like hotpluging a PSU power 24, 8 or 4 pin in while powered on.

 

 

 

now that your rig is running. you could try his board or CPU or RAM in your system for testing.

 

Yes, I could. But I don't need to. He picked up a Corsair TX750M. The system seems to be running now, sort off...

 

 

The system POSTs now and boots of USB stick. He seems to have an IPMI issue though. He has access to the motherboard through IPMI, can read te sensors, gets screen output on ikvm but somehow the keyboard in ikvm doesn't seem to work. Maybe we are overseeing something.

 

When powering the board down through IPMI (Power Cycle or power down) the IPMI LAN connection sometimes is lost. It takes a longer time than mine before he can power up again. Because of this powercyling doesn't work at all. Shouldn't the IPMI heartbeat be on always when the power supply is switched on?

[dikkiedirk]

This is the reply I got from Supermicro Europe:

 

First:

 

Hello Dirk,

 

 

This looks like a power supply incompatibility. Are you using a validated power supply, what is the model?

 

Second:

Hello,

 

 

We validated this board for 1U solution only. Therefore we only validated our 1U power supplies. Do you have a other power supply you can test with first, to make sure, it is indeed a power supply issue?

 

Third:

Hello Dirk,

 

 

I am not a 100% sure it is a power supply issue, but it looks like one. And because you do not have a tested power supply, I do not know if this is a power supply issue. A incompatible power supply can cause this kind of issue with ipmi and with the heartbeat led. It would be great if you have a other power supply that you can test with.

 

They are just guessing a bit. Better have a warehouse of PSUs handy when you buy a Supermicro board?

[BetaQuasi]

I can honestly say you're having some bad luck here!.  My X9SCM just plugged and played, no issues at all.  The only odd issue I've had is that if I set my IPMI adapter to a static IP, I seem to lose the ability to connect to it after a day or two.  So I just set up DHCP with a reserved IP instead and all is well.

 

There's also plenty of folks here in Aus that had no real issues with them that couldn't be resolved with different levels of BIOS on them (overclockers.com.au is full of X9SCM owners.)  In saying that though, almost all of us are using Seasonic 80 Plus Gold PSU's, or a Corsair 80 Plus Gold model.

 

 

[dikkiedirk]

My own system is running fine too, it has a Seasonic X-760.

 

My nephew, who has this IPMI issue, has a Corsair TX750M.

 

Maybe I try my Seasonic on his motherboard.

 

Maybe the TX750M is a different build (OEM) than the original TX750 V1.

 

I call back later and let you know.

 

Never know that there are so many compatibilty issues with one motherboard.

[Zen]

I believe the OEMs for the different Corsair TX750 models are as follows:

 

TX750 - CWT

TX750v2 - Seasonic

TX750M - CWT

 

I personally prefer power supplies that are based on Seasonic designs.

[dikkiedirk]

The TX750M certainly doesn't work to well with the X9SCM. If you aint gonna use IPMI it will work, but then you miss half the fun of the board.

 

I hooked up my Seasonic X-760 to his board and he has no issues anymore with powering down and up and powercycling the board.

 

The list of compatible PSUs get shorter and shorter  :(. And it is left to us end-users to find out which is compatible.

[dikkiedirk]

Replying my own thread. Just a tiny bit of information from a Supermicro RMA techie in the Netherlands:

 

Our X9SCm is a single CPU server board. You cannot use just any power supply for server boards (does not matter which brand board you use) For instance with IPMI the power supply must have capabilities for remote switching This is done with pson and pwok. Also ripple/noise on voltage is more critical then by desktop boards. Then there is the max load and minimum load for the PWS to function okay. Not all PWS can handle peak load from serverboards.

 

5.1  PSON#

 

The PSON# signal is required to remotely turn on/off the power supply. PSON# is an active low signal that turns on the +3.3V, +5V,+12V and –12V power rails. When this signal is not pulled low by the system, or left open, the outputs (except the +5VSB) turn off. This signal is pulled to a standby voltage by a pull- up resistor internal to the power supply.

 

5.2 PWOK (Power OK )

 

PWOK is a power OK signal and will be pulled HIGH by the power supply to indicate that all the outputs are within the regulation limits of the power supply. When any output voltage falls below regulation limits or when AC power has been removed for a time sufficiently long so that power supply operation is no longer guaranteed, PWOK will be dearest to a LOW state. See Figure 2 On/ Off Timing for a representation of the timing characteristics of PWOK. The start of the PWOK delay time shall inhibited as long as any power supply output is in current limit.

 

 

 

 

 

[dikkiedirk]

I believe the OEMs for the different Corsair TX750 models are as follows:

 

TX750 - CWT

TX750v2 - Seasonic

TX750M - CWT

 

I personally prefer power supplies that are based on Seasonic designs.

 

How can you tell which OEM a specific model is?

[Zen]

If you're familiar with the OEM designs, you can open up the power supply to determine who made it.  But most people usually just Google for the information.

[dikkiedirk]

If you're familiar with the OEM designs, you can open up the power supply to determine who made it.  But most people usually just Google for the information.

 

Very helpfull. Just glad that I didn't open the TX750M and voiding warranty in the process. I found the UL-code (E307858) which told me that the TX750M is a Seasonic OEM. Still not fully compatible with the X9SCM, where a real Seasonic X-760 is. I already set-up a RMA for the TX750M.

[Zen]

Are you sure the TX750M is a Seasonic design?  Every review I've seen that mentions the TX750M's OEM has said that it's a CWT design.  I don't have one on hand so I can't check for myself.

[c3]

http://www.hardwaresecrets.com/article/How-to-Discover-Your-Power-Supplys-Real-Manufacturer/370/2

 

Several excellent ways to solve the question.

[dikkiedirk]

Are you sure the TX750M is a Seasonic design?  Every review I've seen that mentions the TX750M's OEM has said that it's a CWT design.  I don't have one on hand so I can't check for myself.

 

I'm not sure of it. I just found the number E307858 linked to Seasonic in a list on tomshardware.com. When I look in the UL database and on the 80plus link I find that it is in fact a own Corsair design (maybe manufactured by CWT). Certainly would explain the fact that it had the issues i explained before and the Seasonic PSU works terrific.

I don't trust Corsair anymore that's a SURE thing.

 

Rather annoying that the end-user is completely left in the dark on this matter. Only Supermicro 1U PSUs are validated. It might be Seasonic OEM's though.

[Zen]

As you found out by looking it up in the UL database, E307858 is registered to Corsair which doesn't tell you who the real OEM is.

 

With Corsair, sometimes different models within the same model line use different OEMs.  For example, in the original TX series, the TX650 was a Seasonic design but the TX750 was a CWT design.  Confusing, huh?

 

But I buy a lot of Corsair PSUs, mostly ones using Seasonic designs but occasionally I'll buy one with a CWT design.  I just do some research ahead of time to know what I'm really getting.

 

Supermicro boards can be picky about PSUs, but you can always ask what others are using.  I recently installed a SuperMicro board (not a X9SCM) with a Corsair AX750 PSU (Seasonic design) and had no problems with it.