Probably one of three things
not in order of probability. Test first then replace or repair.
- The main fuel gauge sender wire from terminal B to the gauge sender terminals in the instrument panel has a resistive short to ground. It's not a dead short to ground which would drive the fuel gauge needle to E.
- Either power or ground connection, or both, through the instrument panel to the fuel gauge has failed.
- The fuel gauge itself is malfunctioning.
3a. Connections to the bias resistor are loose.
3b. The bias resistor has failed from vibration wearing through the power connections or a cracked ceramic substrate.
3c. The gauge itself has failed.
You can pull the main bezel, the cluster lens, and the fuel gauge without disturbing the instrument cluster bucket. This will give you access to the power, ground, and signal connections for the fuel gauge.
Power is hot with the ignition switched ON.
Ground is Ground all the time.
The B terminal of the valve plug gives you access to the other end of the common gauge wire.
Don't overlook the plastic printed circuit. If you have higher than one ohm resistance from the sender clips to the B terminal at the valve pull the cluster bucket, being careful not to break off the tabs, and unplug the cluster connector. Then you can test resistance from the fuel gauge wire on the cluster connector to valve terminal B and test the resistance of the printed circuit sender trace with a meter as well.
Removing the nuts from the bias resistor on the back of the gauge and re-installing with star washers can fix dodgy connections to the resistor. Don't be a ham handed gorilla when you tighten them back down... you can and will crack the ceramic substrate of the resistor. Snug with a nut driver is good enough. If it looks like the resistor is cocking to one side loosen it up and get it level with the back of the gauge can.