Thanks Roger,
I'm close to solving the problem.
The first solution is little additional price at close loop (injection maps). i make the AFR at close loop 14,3-14,5:1 by no problem for adapt, because the lambda is correct by ECU when is 0,97-1,03.
Also the second solution is this one :
There is a map, the ECU make a check of the agreement lambda "1", (128 in hexadecimal code). This value when i set it to 132 (in map) or higher the checks by rich mixture in close loop is "ok", and then ignore the adapt, so no cutting fuel.
At latest ECU (twin spark) this value (128) is variable to rpm (from 1.100-4.900). That i mean "lambda map". Variable for 8 set of maps (value 128-132). This is for all settings of CCP.
Just add +5 to these values, so the ECU when read a richer AFR at close loop accept this as correct. Also i think that there are no open loop at twin sparks. The ECU works typical at close loop at all revs, the AFR can be "correct richer" by lambda maps only at high revs (after 5000+ (has value 132 at latest cells).
By richer fuel, the Lambda narroband send to ECU a voltage 0,55-0,8V, the ECU accept this signal in hex code Value=132+ compares by value of cells (132) so is OK for ECU.
The voltage by AFR Lambda =1 is variable near to 0,45 Volts, that means at ECU value 128, So accepts by ECU only at 700-4.900 revs of rpm (cells in map is 128).
I make the test to a R1150RS and to a R1150GS, i m waiting for results after few miles..