[Beowulf] Re: Emergency Power Off
David Mathog
mathog at caltech.edu
Mon Mar 19 16:20:24 PDT 2007
"Robert G. Brown" <rgb at phy.duke.edu> wrote
> APC has a button -- it is spec'd somewhat in the pdf Jim and I passed
> back and forth and you can look it up on their site. There are some
> other companies that sell them as well.
Found this one:
http://www.apc.com/resource/include/techspec_index.cfm?base_sku=EPW9&type=1
Pricegrabber shows that it's around $550.
>
> They don't look terribly easy to wire up, though, and say that they are
> only to be installed by "professionals".
I think that's for the ones that both shut off the main breakers and
the UPS boxes at the same time. The panel part of that is definitely
professionals only. I don't think that's necessarily so for the
UPS parts. The Tripp-Lite UPS units we have either have a separate EPO
port (whose specs I'm still waiting for, there's nothing in the
manual), or for the ones that can be controlled by a serial port, there
is a sort of hybrid serial port/EPO cable.
http://www.tripplite.com/products/product.cfm?productID=2448
Most of the Tripp-Lites that have serial line control have an inverter
kill circuit that only requires a ground wire and 12V (or less)
positive voltage applied through a 10K resistor. This is how
they can make the Y cable: just hook another 10K resistor to
the pin and bring that line out separately. Logical OR via separate
pull up resistors.
In any case, the current and dc voltage on these Tripp-Lite EPO
lines isn't very high. I'm thinking for the few that we have
in our computer room this circuit would probably due the trick
(sorry about the ASCII art):
+-[9V+]-[9V+]---SWITCH -+ (<--switch is normally open)
| +-10K-- UPS1 -------+
| +-10K-- UPS2 -------+
| +-10K-- UPS3 -------+
| +-10K-- UPS4 -------+
| +-10K-- UPS5 -------+
| | |
+-----------------------+ |
| |
+-common ground ----------------------------+
The pair of 9V batteries should last for the shelf life of the
batteries (5 years?) since there is no load except for the one time
the switch is used. The 10K resistors might not be quite
the right value, I'll have to get the actual specs and see what
the needed voltage and current are.
Possibly the switch needs to be closed for a couple of seconds, but
that could be handled mechanically in the kill switch (separate
mechanism to release it back to the open state, so one slap will
turn off the inverters.)
Clearly this wouldn't work for the Compaq EPO I cited earlier.
Caveat emptor, your mileage may vary, etc.
Regards,
David Mathog
mathog at caltech.edu
Manager, Sequence Analysis Facility, Biology Division, Caltech
More information about the Beowulf
mailing list