[Beowulf] Emergency Power Off

Jim Lux James.P.Lux at jpl.nasa.gov
Sun Mar 18 10:05:44 PDT 2007

At 06:52 AM 3/18/2007, Robert G. Brown wrote:
>To answer my own question (GIYF, after all:-) there is a white paper
>HOWEVER, if you love your local firemen and want them to live (or love
>yourselves and the other employees who sit near the data not-a-centers),
>the same white paper says that an EPO switch is still a very good idea
>for small server rooms and wiring closets that are not "data centers"
>but are just "data closets" or "data rooms that aren't quite centers".

Do not underestimate the power of lobby...  e.g. The front matter of 
the IEEE Emerald Book, (which used to be called "grounding for 
sensitive electronic equipment") explains why the "sensitive" was 
removed.  Mfrs didn't want their particular piece of equipment to be 
referred to as sensitive.

There's a huge "small office and retail" sort of market for middling 
sized UPSes, and APC and the like would like to sell into that market 
as a "plug and play" product.  Also, this is where the local AHJ can 
play a role.  THEY can say, we don't care if your installation isn't 
in a room that meets ALL of the requirements of 645, we want you to 
provide a disconnecting means anyway.

However, an explanatory note in NEC Article 645 (Information 
Technology Equipment) says, in part:

"...For those ITE rooms constructed in accordance with NFPA 75, 
Article 645 contains electrical installation requirements that are 
*less* stringent than the requirements of the first four chapters of 
the Code..." (my emphasis on "less" added)

For instance they let you connect things by flexible cords that 
normally wouldn't be allowed, have subfloors and run cables of some 
types of insulation that would normally not be allowed, etc.  I 
haven't checked other parts of the code, but there might well be a 
"disconnecting means" requirement in there somewhere.

Another way that folks avoid the code is by putting the UPS into a 
rack with the equipment being powered.  That puts it on the other 
side  the "line of demarcation" between that which is subject to code 
and that which is "internal to the equipment" and subject to 
different rules.  And, here, the regulatory environment gets very 
tricky, depending on who's doing the racking and stacking.  In Los 
Angeles county, for instance, an "assembly" like this would require 
UL (or other recognized testing lab:RTL) certification or, possibly, 
wet-stamped drawings by a PE in order to be legally 
sold/transferred/etc.  So that means you can't just order it up from 
your distributor already configured, unless they get the right certs. 
{There are exemptions for R&D equipment, and things destined for the 
movie industry, or things destined for a computer room (oops, now 
you're in Art. 645 territory again)}

If you do the rack and stack yourself, it's sort of in a gray 
area.  As always, it's the paper artifacts you maintain that will 
determine how it gets handled in the event that it becomes important 
(like if a fire happens, and people start to really take a look at 
the stuff...).

What I find interesting is that in California, I could not legally 
perform electrical installations that I would design as a PE.  That 
has to be done by a licensed electrician, and I understand some of 
the purported logic (there's a world of difference between practical 
and theoretical knowledge), but I view it as a fine bit of union 
protectionism promulgated by the IBEW. Yes, there is an exemption for 
work done by a homeowner on their own property.. at least for a while.

>This white paper STILL doesn't help a whole lot with the actual nuts and
>bolts of how to wire a room kill -- it makes is seem like every room is
>supposed to be a unique piece of engineering where you should expect to
>drop thousands of dollars on the process (more than the cost of the
>UPS's being controlled by a factor of two or so) rather than a simple
>"wire the following standard cat5 rj45 cables from here to here, plug
>them into the EPO switch suitably mounted on the wall here, test" type

Typically, these things are wired in two different ways:
1) The "shunt trip", where all the EPOs are normally open, in 
parallel, and go to a special connection on the main breaker feeding 
the room. (Which has the effect of shorting the output of the circuit 
breaker, tripping it.. although that's not actually how it's done). A 
separate set of contacts would be used for the trip on the battery 
side of the UPS

2) The "interlock" scheme, where the EPOs are normally closed, wired 
in series, and breaking any, removes power from a latching relay.

Most UPSes have (at least as an option) some sort of scheme where an 
external contact closure can be used to control the relay.

There are some subtleties in how all these relays get wired up, 
because you need to have power to actuate a relay in the first place, 
and having UPSes around makes the power switching a bit more complex 
(so that once you've turned the relay on, the battery voltage doesn't 
keep it on, even if you remove the original power).

Latching relay schemes are very common: Relay coil and NO contacts 
are in series with power. "Turn ON" button shorts contacts, applying 
power to coil, pulling in relay. NO contacts are now closed, in 
parallel with button, so button can be released.   The OFF button is 
a NC type in series with the whole thing, so that if it is pressed, 
the power is removed from the coil, the latching contacts open, and 
you need to press "ON" to get the thing repowered up.

Where there are numerous things that need to be controlled, you wind 
up having many sets of contacts on the back of the EPO operator. (The 
operator is the actual pushbutton part.. for industrial applications, 
they are made with layers of contacts that you just attach to the 
back.. need 30 sets of contacts, no problem)

For a complex system, there might actually be a separate "emergency 
shutdown power" bus to save the need for zillions of contacts.  This 
would be engineered to "come up first and go down last".  For 
instanace, your EPO might be a NC contact set that provides power to 
a whole raft of contactors (i.e. the UPSes would be configured so 
that their internal contactor is "closed on application of external 
coil power")

With half a dozen UPSes, several different phases of AC power, and a 
dozen racks, this can get pretty complex.

>But I will persevere.  Google is still there, and the information I seek
>MUST be somewhere in webland...;-)

James Lux, P.E.
Spacecraft Radio Frequency Subsystems Group
Flight Communications Systems Section
Jet Propulsion Laboratory, Mail Stop 161-213
4800 Oak Grove Drive
Pasadena CA 91109
tel: (818)354-2075
fax: (818)393-6875  

More information about the Beowulf mailing list