Question : Voltage Sensor [CPU VRM Voltage] detected a failure

I received an alert on one of my servers that had to do with CPU Voltage, and I'm not sure what to make of it:

Log: System
Type: Error
Event: 1154
Event Time: 4:02:36 am 16-Mar-09 UTC
Source: Server Administrator
Category: Instrumentation Service
Description: Voltage sensor detected a failure value
Sensor location: CPU VRM Voltage
Chassis location: Main System Chassis
Previous state was: OK (Normal)
Voltage sensor value (in Volts): 7.793

After investigation, I discovered this server has done this a few times in the past as well.  I've been Googling this, but haven't found anything definitive on if I should be worried about this or not?

Thank You,

Brian

Answer : Voltage Sensor [CPU VRM Voltage] detected a failure

VRM - Voltage Regulator Module. [That's the old term but it's the one still most commonly used.]
Intel likes to call them VRD now [Voltage Regulator Down] to indicate it's no longer a removable part and that it is soldered/integrated onto the motherboard.
- But Intel seems to be the only one that likes using "VRD".

VRM Input capacitors replaceable. - Yes.
I do 5 to 10 boards a month to keep my level of proficiency up. (I'm semi-retired.)
There are places like www.badcaps.net that do 100's a month.
-
Anyone with a decent skill level for basic soldering work can do it.
The hardest part is identifying and sourcing the proper type and grade of capacitors.
They are generically referred to as Low ESR but there are a number of grades of low ESR.
It gets confusing because cap manufacturers use terms like 'Ultra' Low ESR or 'Very' Low ESR and NONE of those terms are defined by an actual Standard. Brand A's 'Ultra' may equal brand B's 'Very'.
-
~IF~ the caps are bad the thing to do is look up the specs of the originals then find replacements that meet or beat those specs.

I brought it up because even premium brands of electrolytic caps will 'dry out' and fail eventually.
- Heat acclerates that.
- The VRM is the hot spot on a motherboard.
- And servers tend to run hotter than office PC's.

Power supplies are also a problem spot for drying out electrolytic caps.
There are techs that repair those too but harder to find than techs that repair motherboards.
Fixing power supplies harder to do because of the disassembly involved and the wide variety of caps they use/need. They don't use -many- but types/grades and needed physical sizes can cause problems/delays when replacements unexpected sizes have to be ordered -after- the work starts.]
It's not usually cost effective to do PSUs unless the unit is very expensive or proprietary and difficult to source.

A VRM is basically a small SMPS [Switch Mode Power Supply] type DC to DC Converter used to step +12v [+5v in old equipment] DC from the PSU down to the voltage the processor(s) need.
There is a controller chip that senses what the CPU is and via logic and the BIOS code it knows what voltage to give the CPU.
That controller chip turns MOSFETs on and off very fast so that the +12v is only applied part of the time which steps down the -average- voltage.
VRM output caps smooth out the inherent pulses so the CPU gets a steady voltage.
VRM input caps smooth the power going to the MOSFETs from the PSU.

Why I suggested VRM input...
VRM output should be under about 2 volts for modern CPU's
Your Voltage sensor said "in Volts: 7.793".
-> That's telling me you have +12v VRM-in and the voltage is dropping low.

It is probably dropping due to PSU output filter caps or VRM-in caps getting near end of life [one or more are starting to shorting out] -or- some other PSU problem.
.
Random Solutions  
 
programming4us programming4us