Here's another potential solution:
There is such a thing as a relay with a delay. They are available
occasionally throughC&H Surplus at bargan prices. I believe that I
have seen them at McMaster-Carr also. Using a capacitor and a resistor
you could easily make one. Just do away with the lower switch, and use
the duration on the delay to pump bilge water down low enough that the
backwash wont re-trigger the remaining higher switch.
Good luck...

On Mon, 13 Jun 2005 23:35:39 GMT, "Roger Long"
wrote:

Larry, Brian, and Ian, you'll all get a kick out of this puzzler.

They finally got the water turned on in the marina so I could wash the
shipyard filth off and out of our boat. I also decided to watch the
bilge system in action.

I DOESN'T WORK!

I repeated the hand tests just to be sure and it functions exactly as
intended. Pick up a top float switch and it runs until the switch is
dropped. Fine for checking that the pump is running.

Pick up a bottom switch. Nothing happens. Hold the bottom switch up
and lift the top switch. Pump starts. Drop the top switch and it runs
till you drop the bottom switch.

When you fill it slowly with water however, it acts like the bottom
switches aren't even there. It starts when a top switch goes ON and
stops when the top switch goes down a bit.

Another strange thing: It doesn't go into the endless cycle that the
pumps did when I tested them with the same hose length and simulated
sump size. There is much less backflow.

Here's my theory on that: In my basement test, I ran the hose out the
window and up the hill. In the boat, it has a long gentle rise along
the middle portions ending in a high loop. I had the test set up
arranged so the pump didn't suck air. In the boat, the switches are
set to get the bilge as dry as possible. The water is moving pretty
fast through the hose. The inertia, the siphoning in the high loop,
and the pump still pushing some even though cavitating, all conspire
to clear the line. It looks as though there is about half as much back
flow in real life as in the test rig which matched the calculated
volume.

I'm stumped though about the current behavior. I confirmed that the
ball in the lower switches is rolled back when the pumping starts. You
can look at the circuit again he

http://home.maine.rr.com/rlma/BilgeCircuit.jpg

But I don't think it's the logic. I have a functioning bilge system
that doesn't feedback cycle but I'm left with four inches of water to
pump out by hand or by manipulation of the float switches.

All this goes to show why they test rockets before sending anyone up
in them.