Power Spectral Density Calculation in Matlab

clear all; close all;
L = 8000000;
PN_nf = -160;   % dBc per Hz
PN_20dB = -140; % dBc per Hz at 1MHz
PN_30dB = -50; % dBc per Hz at 1MHz
f_offset = 1e6;
f_offset2 = 1e3;
F0 = 2.5e9;

sigma2 = (f_offset/F0)*sqrt(10^(PN_20dB/10)/F0);
acc = 0;
actual_timestamps_2 = zeros(1,L);
accumulative = zeros(1,L);
for i = 1:L,
   acc = acc + sigma2*randn(1);
   actual_timestamps_2(i) = i/F0 + acc;
   accumulative(i) = acc;
end

fft1 = fft(2*pi*F0*accumulative);
fft1 = fft1(1:L/2+1);
psd1 = (1/(F0*L))*abs(fft1).^2;
psd1(2:end-1) = 2*psd1(2:end-1);
freq = 0:F0/length(accumulative):F0/2;
figure(1);
semilogx(freq,10*log10(psd1));

[psd1,w] = pwelch(2*pi*F0*accumulative,hamming(512),[],[],F0,'onesided');
figure(2);
semilogx(w,10*log10(psd1));
grid on

In this code, I am trying to find the power spectral density (PSD) of the signal "2*pi*F0*accumulative". In the first figure, I calculate the PSD using its definition i.e., taking fft squaring it and plotting he output. While in the second case I am using pwelch function.<br> While we see a constant 20dB/dec slope in the first case, the second case has two distinct slopes. Since the accumulative signal is just an accumulation of a random signal, it is expected to have a 20db/dec slope throughout (as in the first case). How can I plot the PSD using the pwelch function correctly, so that it does not show two distinct slopes in it. Figure 1 and 2 are shown below:

[Henry Giddens]

Maybe try to remove the max and min values from 'psd1' for the second plot (including the corresponding 'w' values