Date: Sun, 4 Jul 1999 18:39:02 -0400
Reply-To: David Beierl <synergx@IBM.NET>
Sender: Vanagon Mailing List <vanagon@gerry.vanagon.com>
From: David Beierl <synergx@IBM.NET>
Subject: H2O tank senders solved!
Content-Type: text/plain; charset="us-ascii"; format=flowed
Thanks to Steve Sittser who donated a bum sender to the cause, I can now
tell you all about how they break and how to prevent it. These senders are
prone to two problems, and fortunately Steve's had both of them, otherwise
I wouldn't have noticed the second one. Both are fixable within
limits. Here's the story:
These senders consist of a (styrene, I think) hollow plastic rod on which
slides a float containing a magnet. The float is constrained to the middle
8 inches of the rod, with about 4-5 inches above and 2 inches below. This
gives three effective positions, at the top, at the bottom, and anywhere
else. The rod is slender and flexible.
Inside the rod is a long slender piece of double-sided glass PC board
material, i.e. a copper-foil sandwich filled with GRP (glass-reinforced
plastic). At the points corresponding to the float endpoints, the foil is
interrupted on one side and a 560 Kohm resistor and a magnetic reed switch
are soldered in parallel across the gap. The top switch is normally open
and the bottom switch is normally closed. The two foils are shorted
together below the bottom resistor/switch pair, and the lead wires are
taken from the two foils at the top of the device.
With this arrangement, when the float is at the top it closes the upper
switch, resulting in approximately zero resistance between the two
leads. This corresponds to a green LED. When the float drops a
quarter-inch or so the top switch opens, giving a resistance of 560K and a
yellow LED. When the float drops to the bottom, it opens the bottom switch
and the resistance is now 1120K, a red LED.
Problem 1: There are three places where water can get into this
arrangement: the bottom cap, the entry for the wires at the top, and the
joint btw the rod and the top fitting. On Steve's rod the leak was at that
joint, which incidentally is the one that is stressed when you bend the
rod. If any water should get in, even a drop or two, it will soak into the
fiberglass and provide a small but unplanned current path btw the two
leads. This will disable the red, and sometimes the yellow LED, i.e. red
becomes yellow or green and yellow may become green.
Problem 2: The reed switches are tiny glass tubes with wires in each end
and the ends melted shut. If you bend the rod away from the side the
switches are on, it tries to pull the ends off the switches. Sometimes it
succeeds. Again, the upper switch is much more heavily stressed than the
lower one. That's good, b/c the lower switch is more exotic and probably
harder to get.
Observation: The typical failure happens after "vigorously cleaning the
tank." Now you know why.
Prevention: First, take the rod out and thoroughly seal the three
openings. Elmer's Household Cement bites into the plastic all right, and I
imagine model airplane cement would be as good or better. The opening at
the top where the leads come out should get something
flexible. Polysulfide rubber caulk (Lifecaulk, 3M 5200) should be great if
you happen to have a tube around. Silicone would be a lot better than
nothing. Second, pretend the rod is glass and DON'T BEND IT. If you're
scrubbing the tank (or doing anything else inside) unfasten it so it hangs
free.
Cure: If you can hear a tiny click as the float reaches each end, your
switches are ok. Use a plumber's tubing cutter to carefully cut the rod
open near the top, and very carefully draw out the internals. Dry the tube
with alcohol and/or air and set it aside. Take the innards and dry them
with *gentle* heat, air stream, vacuum, desiccant, whatever you can come up
with. Multiple soakings with alcohol and air-drying would probably work
just fine -- soak for fifteen minutes, dry for a day or so, repeat several
times. If you have an ohmmeter that reads that high, stick the circuit
into the rod far enough for the bottom switch to reach the float, and
measure for 1120 Kohms +/- five per cent. Then reassemble the rod and
cement it together where you cut it. Mechanical details I leave to your
imagination -- sleeve, whatever. The rod doesn't have to be any particular
diameter there. Just make sure it's really sealed, and at least as strong
as before. Check before sealing it up that the float will operate both
switches. Extra glue inside won't hurt the circuit, but may prevent you
from taking it apart again, God forbid.
If you have a busted switch, probably the upper, do as above (if water got
in) and also solder in a new switch. The top one is an SPST/NO (single
pole single throw, Normally Open) and the bottom is an SPST normally
closed, which is probably harder to find. This one actually uses a single
pole double throw switch with the normally-open lead cut off. Radio Shack
used to carry the SPST/NO reed switches, but no longer. Look for
electronics parts distributors.
If you want me to fix these, I'll do them for $20 a pop plus shipping, $30
if it needs a switch, no charge if unfixable except shipping if you want it
back. I can fix LED panels, too...
david
David Beierl - dbeierl@ibm.net
|