Note from Chad, the original builder: "My engine will be a Mazda rotary, which uses a high-pressure fuel injection system with a return. After much debate, I'm following in John's footsteps. Each tank will have a pump underneath each rear seat, which will be tied together at the fuel rail. A return will feed a solenoid that will pass the return fuel to the selected tank. Only one pump will be on at a time. This is a useful setup with some interesting side effects. First, its controls can be simplified to the point that feed and return are switched together, so there is no need for the pilot to manage both devices, and accidental cross-feed is prevented. However, deliberate cross-feed can be engaged when desired. This is a nice feature if you're on the ground and about to refuel. You know the fuel you have remaining is good, so you can pump it all into one tank (which is now a "known good" tank) and refuel ONLY the other tank. You can then take off and perform critical maneuvers on known-good fuel, and only use the new fuel when it is safe to make the switch. If there is contamination, this gives you a response option. You can also use this technique to empty a tank for inspection." "In the first picture below you can see several of the components in my fuel system, including the pump, tank fittings, and sump level sensor." PK note 6/8/2020. The project arrived without fuel pump, fuel valve or gascolator. The prior builder even stripped off the AN fittings. I plan on using the fuel flows standard to aircraft fuel injection, with return back to the source tank, and that requires a dual switched fuel valve. I have one of the 6-port Andair valves from a Cirrus, and need to install hard lines bringing the fuel through filters and pumps forward to the dash-mounted valve, then return lines to the source tank and the engine. This is going to be a challenge and learning experience. (There are wires going into the tank, located at about where the sump would be. This is the fuel level sensor, an array of 5 gas float switches to each side, with a resistor network designed to give output varying with fuel level. It was a clever, if overly complex solution to fuel monitoring, described in C21 "Fuel Level Sensors". I've tied off the wires and will put them under a protective cover for future use by someone else.)