problem in TS searching



Hello everyone:
     I met a strange thing in transition state searching in MOPAC. There
 are three reactants,  HOCH2CH2NH2, H2O and HCONH2 to complete the
 reaction like the following:
 NH2
 |
  HOCH2CH2NH2+H2O+HCONH2--->(+)NH3CH2CH2O(-)+ H2O +HCONH2 --->
 (+)NH3CH2CH2O-C-O(-) +H2O
 |
 H
 I use pseudo reaction coordinate technique to find the transtion state
 of the first reaction, but when I use PM3 MMOK TS PRECISE to refine it,
 I failed. Then I cut off the HCONH2 from the system, I can refine the
 first TS I got, because I got one negative frequency.
     Does anyone have some experients about the problem I met, hope you
 can give me some suggestion. Thank in advance.
 Regards,
 mao
   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 |  Mao Xiang                                                      |
 |  Lab of Molecular Regulation for Microbial Secondary Metabolism |
 |  Shanghai institute of Plant Physiology, Academia Sinica        |
 |  300 Fenglin Road, Shanghai, China, 200032                      |
 |  Tel: +86-21-64042090-4791                                      |
 |  Fax: +86-21-64042385                                           |
   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~