Flip angle does not change contrast of EPI sequence as expected

[Jan-Meyer01]

What happened?

Hi, I created a sequence in Pulseq that is basically the Siemens ep2d_fid sequence with a short TE (around 11ms) and made two versions with different flip angles - 10° and 21°.
The first should give me a PD weighted image and the second a T1 weighted image, however the two images, as simulated with KomaMRI, look pretty much the same as if the choice of flip angle didn't matter...

The sequence is 2x accelerated and has 6/8 phase partial fourier to achieve the low TE, so the recon directly in KomaMRI looks aliased, but even there you can see the problem that the images :
FA 10

FA 21

In order to properly reconstruct the simulated image using GRAPPA (24 reference lines) I tried the multi-coil simulation from the PR #548 which worked fine. The custom script for this, as well as an example phantom (2D), is attached below - you probably need to adjust the paths, as I defined them for my project repo (can also invite if that's easier). Then I used a python script for the recon and compared the results:

Again the images look identical (also plotted the difference here).
Do you know if this could be an issue with the sequence, phantom or the simulation itself? Any help would be appreciated!

Sequences, scripts, coil sensitivities, simulation results and the phantom are attached as .txt to get them through GitHub:
ep2d_slice1_FOV232x232_Res64x64_Pat2_FA10.seq.txt
ep2d_slice1_FOV232x232_Res64x64_Pat2_FA21.seq.txt
run_simulation_KomaMRI_multiCoil.jl.txt
sense_2D_8coils.mat.txt
ep2d_slice1_FOV232x232_Res64x64_Pat2_FA21_signal_raw_THS_2D_8coils.mrd.txt
ep2d_slice1_FOV232x232_Res64x64_Pat2_FA10_signal_raw_THS_2D_8coils.mrd.txt
recon_KomaMRI_THS.py.txt
sub-phy001_ses-001_phantom_KomaMRI_slice150.phantom.txt

Environment

OS x86_64-linux-gnu
Julia 1.11.6
KomaMRIPlots 0.9.2
KomaMRIFiles 0.9.1
KomaMRI 0.9.1
KomaMRICore 0.9.1
KomaMRIBase 0.9.1