Armen Kechechyan, Michael Tokarev, and Alexandre Alakhverdyants (JINR) reported an observed systematics in reconstructed  momenta with respect to simulated ones in region |cos(Phi)|<0.2.  

Relative difference is -1.2%. The observed difference has been associated with missing in Sti geometry SVT cone and Support rods. After adding these volumes the difference becomes +0.5%.

The conclusions are:

1. Adding SVT cone and Support rods reduces the problem.
2. It looks like in Sti rods weight/density is overestimated by 1/3. This could be due to that in Sti elliptic tube is replaced by tube.

The differences for pions only old => new 

.