Nut upsdev - Nov 2008 - How do I approach this problem?

I'm having a problem that I hope those with greater experience might 
have some ideas on.
I live up in the mountains where the power is ill-conditioned. 
Transients which would cause a computer to reboot occur on a daily 
basis. Power failures are frequent and can last anywhere from five 
minutes to five days. I have all my computers on UPSs so the first two 
problems are solved. I have a generator for the third problem. The 
difficulty I am having is with the power failures that last long enough 
to drain the battery on the UPSs.

As this would damage the batteries, I programmed the UPSs to shut down 
their respective computers after roughly half the life of each UPS 
system's capability. The problem that remains is that I would like to be 
able to have these automatically restart once the power comes back on. 
For one thing, I have one computer hooked up to sensors taking data that 
I would like to be off-line as little as possible, and I have another 
which serves this data to the outside world when possible.

This is a low-budget operation, so buying bigger UPSs or auto-generator 
systems is not an option. Is there any way to have the UPS restart my 
computers once it knows its back on live power?

I am using nutups to communicate with the UPS on the computer hooked to 
the sensors. I have a tripplite on the other main computer and am using 
their package. Both are Linux computers.

Your help would be greatly appreciated.
--Jeff Cunningham

On Mon, Nov 10, 2008 at 07:23:58AM -0800, Jeff Cunningham
wrote:
> Is there any way to have the UPS restart my computers once it knows its
> back on live power?
Most BIOSs have a setting for this - it's usually set to "Last
State", which
unfortunately doesn't really work on some motherboards - "Always
On" will
work.

There's (at least) two other hardware routes you can take:

1. On the ATX motherboard power header, short the green wire to a black
wire. This forces the power supply on.

2. Rig up a relay to momentarily short the motherboard's power switch when
power returns, toggling the power on. I have not actually tested the
following circuit in a PC, but it's simple enough, and it should act a
little nicer than the first method:

                                          ____________
                      100ohm             |  12V relay |
12VDC+ ------->|------/\/\/-------+------|+           |
(yellow)     diode   resistor     |      |C       N/C |---------- To mobo
                                  |      |o           |           soft
           very large capacitor =====    |i         C |---------- power
                                  |      |l           |           switch
GND    ---------------------------+------|+       N/O |
(black)                                  |____________|


Basically, when everything is off, the N/C and C pins will be shorted
together (like you're pressing the power button). When power returns, this
will turn on the computer. Once the computer's 12VDC+ line puts out power
long enough to charge the capacitor and energize the relay, the relay
switches and the short is removed (like you had released the power button).

If you attempt to turn off a computer with such a circuit attached, the
capacitor will temporarily power the relay (the diode prevents it from
attempting to power the rest of the computer) and give the disks a chance to
spin down. Then the relay contacts will close and the computer will turn
back on.

I did go so far as to test how long a capacitor would power a relay - with
the relay I had (which was 380 ohms across the coil), a 15000uF cap would
power it for about 6 seconds, and a 25000uF cap would power it for about 10
seconds.

The resistor will see about 1.5 watts peak, but since that's only for a
second or two, a 1/4 watt resistor won't even get warm.

If you actually use this, please let me know how it works for you.
-- 
Robert Woodcock - rcw at debian.org
"If an elderly but distinguished scientist says that something is possible
he
is almost certainly right, but if he says that it is impossible he is very
probably wrong."
	-- Arthur C. Clarke