Probability of Symbol Error Rate (SER) of M-FSK plot against Es/No

Hello,

I tried to plot theoretical probability of Symbol Error Rate (SER) of M-FSK plot against Es/No but I think the results are wrong. Normally, when we plot BER of M-FSK against Eb/No, the error rate decreases with M increasing. So, shouldn't that be the same case with SER plotted against Es/No? Because when I plot SER of M-FSK against Es/No, I get the error rate increasing by increasing M. Below is the code. Please help out. Thanks.

Es_No_dB = -3:20; % Eb/No values in dB

Es_No_li = 10.^(Es_No_dB/10); % Linear Eb/No values

Colors = {'k','b','r','g','y','c','m'}; % Cell array of colors

Markers = {'o','x','+','*','s','d','p'}; % Cell array of markers

for n=1:6

    figure(1)

    M = 2^n;

    SER_MFSK = (1/2)*(M-1)*erfc(sqrt(0.5*Es_No_li));

    semilogy(Es_No_dB, SER_MFSK, 'color', Colors{n}, 'marker',Markers{n}, 'LineWidth',2)

    axis([-3 15 10^-5 20]), hold on, grid on

end

legend('M = 2','M = 4','M = 8','M = 16','M = 32','M = 64')

xlabel('Es/No (dB)'), ylabel('Symbol Error Rate (SER)')

title('Analytical SER Curve for M-FSK over AWGN Channel')