TPS43000: SEPIC & loop compensation design

Part Number: TPS43000
Other Parts Discussed in Thread: CSDPlease help me. I need to use SEPIC classic configuration L1 and L2 mainly because I cannot use a coupled inductor and I need insulation (and I did not know about Zeta converter). Current values are Vin = 6 to 8,4 V Vout 3,6 @ 3,3 A. Fsw 1 MHz. L1=L2 = 3,3 uH. Cout = 3 x 47uF ceramic. Cfly = 2,2 uF ceramic R1 = 37500, R2=R3=4870 C1=10 nF C2=270 pF C3=680 pF I developed the circuit but the man who's made layout did not make a good one, due other constraints in the board. In any case, the system has 2 main issues, I ask someone if can help me to found the root causes. first one, it doesn't start at design voltage (6V) but starts at lower ones , (2,5 to 3,5 V) second one, when I changed the N-mosFET from IRF5304 to CSD 17302 the system continue to start at lower voltage than designed but it cannot reach the steady state condition when apply loads, also light. Any suggestion? Some notes. I designed the compensation loop following the TPS43000 data sheet and other literatures from TI: how to calculate the crossover frequency? it should by from the formula fc=(R2 x C3/(2 x PI x Lp x Co)) x (Vin/(Vramp x (1-D)^2) that should be good for boost, flyback and also SEPIC topology. but I do not know the value of the ramp voltage, than I guessed it from error amplifier min and max output, 2V. Is it correct? I tried to apply the inverse formula and calculate the crossover, zeroes and poles frequencies in several TI literatures but in all that, the values found do not fit the premises: i.e. paragraph 2.4 of both SLUA268 and SLUA267. There are something not written that changes the values from formula? I know it is complicated, I do not need a solution, but a method to find the solution..... :) Thanks to everyone who will try to help