I wired it several times as the datasheet shows but I can't get any regulated voltage. In order to understand what happens, I removed the mos and put a potentiometer instead of the divider. The result is (IMHO) what it should: The gate voltage is maximum when the divider is below the 1.235V reference voltage the goes to "nearly zero" (in fact around 1 v) when the divider is above 1.235V.
Then, I rebuild the circuit with the mos and did the following measurement:
Light blue: Vgs
Starting with a 1A load, I raised it until 10A.
The output voltage goes from "too high" to "too low". The voltage is correct only when on marker 2 in the below capture.
The pump voltage falls to 0v on marker 1.
I definitely don't understand what happen. Did I do a wrong choice with the mos? Should I work around capacitors of the charge pump? Is there a reason why the gate voltage in first test goes to "nearly 0v" and not 0v? Is it a regular behaviour that the charge pump voltage goes to 0? I supposed that It should stay always high, only the Vgs changes but may be I"m wrong.
A lot of interrogations...
I changed MOS for IRF610 (datasheet) and... It works! The specs does not fit my needs but the voltage is regulated with. Still wonder which characteristic makes the original mos unsuitable.