There was some interest expressed in this exercise, so here goes...

Being exposed to gas, some coating and corrosion is inevitable in the sender which was the culprit.

There are three connections: one for the empty signal (yellow), one for level (purple) and one for ground (brown). The ground runs to behind the right interior quarter panel. The empty signal should have 12 volts. The level signal has about 3.25 volts.

All combinations of connections were "open" so...

I removed and disassembled the sender. I found that the feed-throughs for the level and empty signals were corroded and no signals were passing from the cap to the actual sensors. This was fixed by soldering short pieces of wire from the protruding rivet to the "tangs." The wire on the empty feed-through fell off, but the heat and solder made it work. (For a time.)

The operation of the sender is fairly simple. There is a float running on a shaft. At the bottom, it shorts the empty terminal to the ground terminal.

The level is measured using a piece of resistant wire that runs in a "U." The ends of the wire are connected to the ground and level terminals. The float has contacts to both sides of the "U" so the resistance varies as the float moves. At the top, the resistance will be low, around 3 ohms. At the bottom, the resistance will be around 76 ohms. As the wire connects the terminals, if the float is not contacting the wires, the resistance will be near 80 ohms.

After the R&R, I reinstalled the sender, but the gauge didn't work well.
Occasionally, it would flick up especially when hitting bumps, so that was progress.

This suggested the gauge itself and the wiring. I picked up an extra gauge and compared it to the one in the pod. Both should similar resistances, which I forget, so I put the old one back in.

Looking at the wiring diagrams, the signal runs from gauge to the fuse panel on H3 and on to the sender from T3. "H3" means the 8th connector from the left (A,B,C,D...) along the bottom of the panel and the third signal from the top on that connector, starting in the upper left corner. The lower part of the cover needs to be removed to see these.
Note that the odd-ball feeds in the middle are not part of the left-to-right counting, but have letters molded into the plastic. You can also see some letters at the very bottom of the panel.

Both H3 and T3 showed 3.25 volts and the purple wire at the sender showed 3.22. A bit of loss, I figure.

Removing the connector and probing the terminals showed that the sender was not working right, again. The ground was good, and the voltages were right (12 and 3.22) on the connectors.

With the engine off and the connector on the sender, the resistance between the ground terminal and the level terminal was around 32 ohms.
Sounded good. Except that, with the connector removed, the resistance between ground and the level terminal was 53 ohms. Checking the terminals, connector removed, showed around 80 ohms from level sense to ground. The sender was hosed, again. (53 in parallel with 80 gives 32, abouts)

Removed the sender and checked it out. On the bench, the resistances once again looked right. I got some lacquer thinner and cleaned the wires and sliders on the float. I also used some 400 grit sand paper on the wires. (Carefully) Looking at the float, I noticed that on one side the wire didn't run through the contact fingers right. I used the sandpaper to brighten the fingers (carefully) and slipped the wires back into position.

I also soldered in a couple of stands of wire across the rivet and tang of the empty signal. Stuck this time.

Put it back in, and tried it...Yes. The gauge worked! (For the first time in the two years I've had it!)

'80 Euro S - Getting down the list