It was the fuel pump.
A voltmeter won't be useful for diagnosing the problem – because of the switching to ground configuration of the electrics a test light is needed. I attached safety pins to the test light leads and pushed them into the two (heavier) wires leading to the connector at the pump, under px rear seat. After several tries the light went on, and the pump didn't run, so I knew it had to be the pump, and I ordered and installed a new one (aftermarket $200 – the Ford part was $575).
The aftermarket installed fine and has a 1 yr warrantee. My gas tank was nearly full, so I used dishwashing rubber gloves, but to make things easier wait until the tank is near empty to do it. I was happy to inspect the tank bottom and find it clean, not a trace of phase-separated ethanol. But premium gas here has only recently started to contain ethanol. There was likely some very find sediment, but I didn't want to disturb that. If anyone wants more details on the diagnosis and install ask now, because I'll have forgotten in a week. I'll check back, but the website has been giving me issues, very slow page loads etc.
Runs like a charm now, full throttle manifold resonance is music to my ears!
So I took the old fuel pump assembly apart. I found, to my amazement, that the fuel is actually pumped through the motor, right over the carbon brushes and commutators, and around the rotor and bearings (presumably for cooling, and to quench the sparks), and then on its way to the engine. I was not previously aware that the electrics themselves were immersed in gas, I expected that these would be sealed away from the fuel, but this was clearly not the design.
The two carbon brushes still had lots of length in them, but there were 1/32” or so grooves worn into the copper commutator contacts (which are oriented axially, instead of the more traditional radial arrangement, giving the motor a slimmer profile). Note that the motor/pump is staked and swaged on assembly, so I had to sawcut the heavy steel case to find this out- you would not be able to make a repair to this.
My scientific wild [non-permissible content removed] guess is that the brushes might have gotten caught on an edge of the copper, where it “makes and breaks” contact depending on where it stopped when it was shut down. Repeated starting attempts jog it back and forth slightly until it frees itself, until the next shut down.
Popular New Cars
Popular Used Sedans
Popular Used SUVs
Popular Used Pickup Trucks
Popular Used Hatchbacks
Popular Used Minivans
Popular Used Coupes
Popular Used Wagons