RF and Grad amplitude depth in .seq

[schmidt-jo]

Hi all,
long time no see :) hope to catch up at some point and thanks for your great work in the meantime!
I recently ran across an artefact which i assume might be due to how meticulus things have to be in sequence programming (also asked around in the pulseq gang pulseq-admin/pulseq#189).

After almost going crazy i wen over all timings and did all simulations i wondered if it might have something to do with how shapes are stored in the .seq

Due to the various timing rasters, the degree of freedom with timing is limited, hence i was relying on just calculating the correct gradient amplitudes for balancing my sequence. I still got some artefacts in my images when using certain spoiling gradient moments, although my own simulations had sub 1e-17 error margins. I wondered if it was since i did my simulations using my own sequence obejcts and shapes, set strictly on the timing raster but using float64 depth.
When set to the .seq file the depth appears to be reduced which might lead to amplitude offsets and thus could create the errors i see when using certain amplitudes.

Did anyone encounter something similar and do you have best practises or any guidance?
I was intentionally trying to code my sequence such that i can change around many things (such as rf durations etc.) but this might be suboptimal when it comes to such differently set gradient amplitudes?!