Why I get 'Warning: Infinite or Not-a-Number function value encountered.' using dblquad in this case?

n = 1.42;
A = 980e-9; % lambda
wox = sqrt(log(2)/2)*(A/(pi*tan(2*pi/180)));% wox
woy = sqrt(log(2)/2)*(A/(pi*tan(10*pi/180)));% woy
k = 2*pi/A;
z = 5.8e-6;  % La distancia
Rlx = 38.7e-6;
Rly = 3.7e-6;
Alpha = 78*pi/180;
RF = 0.2e-3;

syms x y x1 y1 real  % A = lambda       
exp_X1 = int(exp(- x1^2/wox^2 +(1i*pi*(x - x1)^2)/(A*z)), x1,-5*sqrt(2)*wox,+5*sqrt(2)*wox);
exp_Y1 = int(exp(- y1^2/woy^2 +(1i*pi*(y - y1)^2)/(A*z)), y1,-5*sqrt(2)*woy,+5*sqrt(2)*woy);
tlx_in = exp(-1i*(2*pi/A)*(n-1)*(Rlx-sqrt(Rlx^2-x^2))); 
tly_out = exp(-1i*(2*pi/A)*(n-1)*(y*cos(Alpha/2)));

F = exp_X1.*exp_Y1.*tlx_in*tly_out; 

F1 = matlabFunction((abs(F)).^2); % matlabFunction(Yn_xyz*conj(Yn_xyz)); 
text = func2str(F1);
new_text = strrep(text,'erf','erf_');
F1 = str2func(new_text);         

sal = dblquad(F1,-Rlx,+Rlx,+Rly,+RF)

This code attempts to calculate the integral of F1 on the domain 'x': [-Rlx, Rlx] and 'y': [Rly, RF].

Testing shows that the error occurs on the domain in 'y'. For example, if I limit the integral in the y-axis as: [Rly, .25*RF] a numerical answer is obtained. If I put the [Rly, .3*RF] interval warnings arise again and NaN is obtained.

In my code I need to make this integral several times and it is not always as simple as limited to ".25*RF" in 'y' domain for an answer, plus I've noticed that the 0.25 value varies according to other parameters.

Thank you very much for the help!