I've read about a dozen threads on here about diagnosing fuel sending units and fuel gauges. But I can't get it to line up with my situation, so asking here. Here's the situation.

82 Z28 with mechanical fuel pump on engine block. Will be switching to EFI soon. Sending unit was wildly inaccurate. I put Holley retrofit in-tank EFI pump in (made for thirdgens), removed mechanical fuel pump, and installed regulator to feed carb. That all works well. Fuel gauge reads zero when ignition power on and when turned off, rises slowly to 1/4 over 20 seconds and stays there. If ignition power is applied, gauge falls back to zero. Tank is full for sure. I've got a 2 pin connector at the rear of the vehicle on the drivers side beside the tank that runs through the tail light area into the car and up to the front. Since I have electric pump now, ran a 12 gauge power wire from pump up to front of car to a relay, as is standard practice.

Here's what I've done to troubleshoot.
1) Unplugged harness at tank. With multimeter, took resistance reading (harness to sender) with one probe on pink sender wire and one probe on black ground wire. Got .8 ohms. Tank is full so I was hoping for closer to 90 ohms.
2) With harness unplugged, took resistance reading on other side of harness (towards gauge). Gauge is at 1/4 full at this point. Got 68 ohms, one probe in each side of connector.
3) With harness unplugged, applied ignition power. Gauge immediately went to past full and stayed there when power turned off. From what I've read, this is a good indication. Took resistance reading with power off at this point and gauge pegged past full - 44 ohms. Here's the scary part - with the harness unplugged and ignition on, the pump ran. That harness has the ground wire for both the sending unit and the electric pump. How could the pump be running without ground (harness unplugged)?
4) With harness unplugged, connected pink and ground wire (harness to gauge) together with paperclip. Then applied power, gauge dropped to zero. From what I've read, this is a good indication.

From what I've read, I believe:
The gauge cluster is good.
I should not get .8 ohms resistance with a full tank if I'm measuring correctly (one probe in each pin of connector leading to sending unit/pump). Should get 90
Maybe the float is stuck on this brand new pump/sending unit?
Maybe the sending unit wire is touching metal somewhere causing the 0 ohm reading? Hard to believe as it's wrapped in corrugated plastic conduit.
Maybe the wiring got jammed between the tank and the body of the car when installing the tank and caused a short?
I have no idea why the pump runs with the harness unplugged (with just power wire running up to relay). Is there another source of ground for pump?
I wish I'd checked varying resistance on sending unit before installing pump into tank so I could eliminate bad sending unit. Lesson learned.

Probably gonna have to drop tank again to check all this wiring and test sending unit out of tank. I hate dealing with wiring and electricity - hoping someone can make something of this mess and give me some ideas before I go and drop exhaust and rear end and suspension. Thanks!
I assume the varying resistance readings in the wiring from harness to gauge is because of the varying position of the gauge (full vs 1/4)?

 

Well, I guess I'll start dropping the tank again tonight. Was hoping to get some ideas about what might be wrong

 

Doing the job twice sucks! The fuel pump running without the harness connected could be explained by the tank grounding via the tank straps to the body. I think, there is a plastic rubber coating on the straps to prevent the tank for grounding to the body, that way the only ground would be the ground wire from the tank harness that connects to the back of the car (can't remember).

 

Yes, it does suck the second time around, although it was much quicker.

It appears the float was stuck at the empty level. I took it out of the tank and tested it with multimeter. Ohms ranged from 0 to 90 as designed. I think it got stuck on the cross brace/baffle inside the tank. Once I got it all back together, I put the tank on end and moved it back and forth so the float would move. Ohms moved throughout the whole range, so I confirmed it wasn't stuck. Got everything installed and gas gauge works fine now. Lesson learned - function check the sender right after tank install and before everything else gets put back together!