Well, as I said, these couplings can be quite finicky. Unfortunately, it's also difficult to monitor them by TLC when using DMF as the solvent, so that adds a degree of difficulty. At this point, my first suggestion is to allow the reactions to go a bit longer (24 hrs or more). Something else you can try is maybe refluxing the SOCl2 reaction to form the acid chloride, then add the diamine/Et3N later.
I can't guarantee anything, peptide couplings often cause me headaches as well, so you're not alone.