i'm not convinced that a resistor can do anything useful with a narrow band sensor output. all a resistor will ever do is drop a bit of voltage across it and with a NB sensor, the output is not proportional to lambda. the ecu would just see slightly less voltage both when lean and rich.

an ecu (generally, particularly old types like i am using) in closed loop adjusts the fuel to see a big difference between the 2 voltage and cycles around that point using a simple step/cycle control method - they can't have a tuned loop like a PID controller to settle somewhere as the feedback signal is not proportional to the controller output (injector duty cycle)

it's at the light throttle openings and cruising where it's worst, which is how i know that it's during closed loop conditions.

obviously ideally i would do a full datalogging exercise with a couple of wideband sensors first to know what's fully going on, but there is enough evidence for me to know that my initial idea should be an improvement, although costly. a piggyback system of some kind might be an answer, but it would have to be one that modifies the lambda sensor outputs or i lose them and go open loop all the time which is not really something that i want to get into as it will require a lot of mapping effort. as would a stand alone ecu - unthinkable on a v4 engine with vtec revving to 12000 rpm really with the effort that i would want to put in.

some potentially interesting products here: http://www.perfectpower.com/ but they require more research