Alchemy 2000 MM3 energy

 Dear CCL netters:
 I am doing a conformational search using the alchemy 2000 program and I
 do MM3 geometry optimization of each predicted conformation. I
 found that the MM3 energy value of the optimized conformations is not
 consistent. I mean after doing MM3 geometry optimzation energy of say 10
 conformations and recalculating them again it is not ncessary to get the
 same MM3 energy value. Since my molecule has a ring of about 20 atoms I
 did the calculations for a similar smaller ring system and I found that
 the energy values are consistent which means that the problem is with my
 molecule only. I printed the output file with different MM3 energy values
 for the same conformation and I noticed that there is a difference of
 0.00001 A in the coordinates of some atoms and a difference of 0.0002 in
 the initial steric energy values and after the geometry optimization each
 output has a different route with quite different final energy and
 geometry. I would greatly appreciate it if some one noticed this before
 and know a solution to this problem to inform me. Thank you very much in
 advance.
 Best regards,
 Adel El-azhary