On 2009-07-10, Tristram Scott <> wrote:
>
>
> Hi All,
>
> I have a SunFire 280R which has been working without much trouble for a
> number of years. I recently bought a new 8 socket surge protector with an
> electricity monitor from Maplin:
>
> http://www.maplin.co.uk/Module.aspx?ModuleNo=227272
>
> My intention was to keep track of power consumption from this server, and
> also to have the machine isolated in case of an electrical fault. The
> power meter sits on about 2.0 A under normal conditions, in the order of
> 440 W. That seems fine for such a machine. What I am not sure about,
> though, is the reported leakage current.
O.K.
> With both power supplies connected, the leakage current is reported to be 1
> mA. I get the impression that this is not out of the ordinary for such a
> server. However, yesterday the surge protector tripped, blaming the
> leakage current, which apparently had exceeded the threshhold level of 5
> mA.
The leakage current is actually an imbalance between the current
into one side of the line and out of the other (which is the normal path
for each power cord). If there is an imbalance, it is presumed to be
going to the system's ground and possibly through a person.
The power sockets in the back of the system contain capacitors
between the power leads and between each and ground, and inductors to
block RFI (Radio Frequency Interference).
The ground pin is also firmly connected to the chassis of the
computer, and through that to the rack in which it is normally mounted.
Usually ground cables are run from a threaded hole on the back for the
purpose to the rack so ground current does not go through the rack
slides, which may not be reliable conductors. This also connects to the
grounds on anything else in the rack -- drive trays, tape libraries,
monitors, and whatever else is powered in the system.
Normally, a rack full of computer equipment is not expected to
be connected to a GFI (Ground Fault Interrupter), and mine certainly is
not. Instead, it is connected to a UPS (an old BEST Products one which
just keeps going with a new set of batteries every four years or so)
which is connected to a higher current wall outlet (30 A IIRC) which
does not have a GFI, and which depends on the ground connection through
that outlet. The UPS also monitors the power consumption of the total
load to which it is connected.
> I can see that there might be some leakage in these things by design,
> presumably in the filtering at the inlet, but I don't see why there would
> be a sudden surge in the leakage current, as was reported by the surge
> protector.
How about a static discharge, if the room is air conditioned and
the humidity is low. Shuffle your feet over the carpet, walk up and
touch the rack and you are likely to trip the GFI.
Or -- a burst of RFI from somewhere outside the computer rack
could have caused a spike in current through the capacitors and that was
sensed to shut down the system.
Or -- do you have a modem, or something else connected to phone
lines? An incoming ring signal could be the cause of the trip.
I, personally, would not use a GFI circuit (leakage sensor) on a
computer server. Just separately wire the rack to a good ground and
don't worry. The GFI is designed to sense leakage current in a two-wire
appliance (no safety ground) and to shut down the power if there are
signs that leakage at a dangerous level could be going through a person.
> To me this looks like an intermittent fault in one or other of the power
> supplies, rather than normal variation within tolerance limits.
Can you get a second one of these power filters/meters, and
connect each power supply cable through a separate one? If so, you are
likely to trip only one at a time (the one which is connected to the
faulty power supply -- if such exists). If it trips both at once, the
cause is external.
> Does anyone else with a similar server have the facility to measure the
> leakage current and report back as to what is considered normal?
I have two SF-280Rs, and do not have a way to check this,
because I would never expect to use it that way. One power supply of
each is connected to the UPS, and the other directly to a wall outlet,
so if there is a failure in the UPS, the computer will continue to run.
> Alternatively, does anyone have any insight as to what might be the cause
> of the increase in leakage?
Several suggestions above. But it is not a problem as long as
you can keep the power monitor strip from shutting down your computers.
> I'm not at all keen to have to scrap this machine, but neither am I happy
> about having faulty electrical equipment in active service.
Don't scrap it. It is not the machine's fault, but rather
*your* fault for connecting it to a GFI equipped power monitor. The
machine is behaving normally. Spikes of 5mA leakage is negligible for a
server or anything else with dual power supplies.
Hmm ... looking at the web site, I see that it is using UK power
connectors and the price is in pounds, not dollars, so the "GFI" term
which I have been using throughout this article needs to be changed to
another UK-specific term which I don't remember.
Is there a way to switch off the leakage protection? If not,
totally remove the monitor from the system. It *will* keep generating
problems.
If you have a good UPS, it will protect the computer from surges
at least as well as the device which you have.
Good Luck,
DoN.
--
Email: <> | Voice (all times): (703) 938-4564
(too) near Washington D.C. |
http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
Similar Threads
Linear Mode
