Hoffman's plasticity UMAT

Hi everybody, First of all I would like to introduce myself, as a new user here. My name is Aleksander Marek and I'm student (3rd year of 1st degree) at Poznan University of Technology in Poland. I'm working on UMAT that accounts for Hoffman Plasticity and all of its simpler forms (Hill,
Mises) in field of paper mechanics. I am currently testing my UMAT against Abaqus implemented model on single plain strain element, fixed on one face and loaded with prescribed displacement on one node. So far my UMAT correctly predicts elastic behavior on both isotropic and orthotropic materials. I've encountered problem with plasticity (as early as on Mises phase testing). Routine seems to correctly predict plastic behavior until some threshold level of strain (displacement), beyond that point Abaqus/UMAT results diverge. I figured out that this means that base algorithm is working well (I compared results of elastic predictor-plastic corrector scheme on debug level with other available routines) but there seem to be some kind of bug. I suspect that it may be cause of bad performance on finite-strains but I don't know how to include them. I've done some testing on Anisotropic Elasticity+Hill Criterion but results seem to diverge from the very beginning (well that could be result of change in orientation material matrix while in UMAT it is constantly the
same). I attach some plots of Mises Stress against prescribed displacement (though I am aware that Mises Stress is not good measure for Hill criterion) and UMAT file itself. Any advice will be useful. As I understand Abaqus posts incremental strain already in local material direction, and I predict trial stress by having 

sigma*=STRESS+D*DSTRAIN

Is this correct?

 PS.I've made topic on forum (I'm new here) but it seems that I have some
problems with it (can't load some topics, some answers for FAQ etc.) . Routine file is attached as .txt