Hi Le Roux,

I tried to at least emulate what Tracy did with his systems. All supply/return to one tank (in my case, right tank). This is apparently how many certified turbine installations work; all supply to the engine is from one tank and others transfer to it. I had the other 'main', plus two aux. I plumbed Van's 3 position switch as an aux selector to feed a transfer pump(s). The output of the transfer pump T-d into the regulator return line that went into the actual 'main' tank. I also emulated Tracy's concept of putting the regulator in the fuselage, and the line going to the engine is 'dead end'; all fuel that goes past the firewall is consumed by the engine. That way, no fuel is heated by the engine itself. Crude diagram is attached, showing flow paths. Note that I show a separate filter after each injection pump; probably overkill but I was being a bit paranoid during design phase. I forgot to draw the pre-filter between the main tank and the injection pumps; don't forget that doing the install. IIRC, pre-filters need to be something like 80-100 microns; actual injection filters something finer than 40 microns. 

I also plumbed two transfer pumps for redundancy. In the drawing I show them paralleled (don't forget check valves, if the pump needs one), but in my physical installation I put them in series because it was simpler to plumb in the space I had available. Note that the xfer pumps are low pressure pumps. I originally intended to use injection pumps, but Steve Boese did some testing and discovered that if you overfill a tank while transferring, the restriction of the little -4 vent tubing can cause pressure to rise to around 10 psi, which would destroy an aluminum  tank.

Obviously not the only way, but it's what I came up with after spending a lot of research time and 'head scratching' on the subject.

Charlie