The fuel pump relay should only turn on for one second when you turn on the ignition switch, then not again until the engine is rotating, (cranking or running). At this point I think that is irrelevant because you already found a clue in the pump does not run for that one second, and you are missing a voltage. If you have a helper turn on the ignition switch, you should be able to feel the relay click on, then off.
Your relay numbers do not match what is on the wiring diagram. You listed numbers used with Chrysler relays, and I am very familiar with them. You found twelve volts on 85 and 86. Those are for the relay's coil. The engine computer grounds 85 to turn the relay on. These two voltages are normal for now.
At issue is the four volts on terminals 30 and 87. I suspect your meter is picking up stray voltage and you actually have
zero volts there. To prove my suspicion, measure those terminals with a test light while you have the voltmeter there too. If connecting the test light takes it down to zero volts, that zero volts is the accurate reading. My reason for thinking this is there cannot be any voltage on the line going to the fuel pump, at least if the relay is out of its socket or if the relay is turned off.
If you really do have four volts on terminal 30, there has to be a broken wire with a carbon track inside the insulation, or a corroded connection. Start by looking at the 15-amp fuel pump fuse. If it is not blown, measure the voltage on its two test points. If you have twelve volts there, it should be at the relay, terminal 30, too. If that is not correct, the wire between them has a break in it.
If you have zero volts on the fuse, I will have to study the diagram to see where to look next. Most likely there would be a break inside the fuse box. Also be aware that related to my corroded connection comment, GM was famous during this time period for experimenting with aluminum wiring. They were attached with brass rivets to the buss bars in the fuse boxes. Add salt from your feet, or when driving in salt, and moisture, and two different metals, and you have a battery. Specifically that causes "galvanic action", which is a fancy name for corrosion. Also, it is never acceptable to pierce a wire's insulation to take a voltage reading. When that is done to a copper wire, it can be expected to corrode through in a few years. Do that to an aluminum wire, and it will corrode through in a few weeks. The powder left behind can conduct enough current that a digital voltmeter can pick up some voltage further down the line. You will not get enough current through that powder to run a test light. That is why I suggested measuring terminal 30 with the voltmeter and the test light at the same time.
Image (Click to make bigger)
Thursday, May 18th, 2017 AT 10:10 PM