CCL:H-bond accuracies within MM3

 Dear netters,
 I'm using MM3(92) to calculate conformations of compounds with
 intramolecular hydrogen bonds. From the resulting (energy-minimized)
 structures I obtain parameters like e.g. d(H...X) and <(X-H...X), i.e. the
 distance between donor- and acceptor-atom and the often used "hydrogen bond
 angle", respectively.
 My question is:
 What are the accuracies in the above mentioned parameters ?
 (Are we talking about 0.01 or 0.1 Angstrom ? and 1 or 10 degrees ?)
 Please respond if somebody has any ideas about this topic or perhaps
 literature references on it.
 If there is enough interest I will send a summary to the net.
 Thanking you in advance.
 Marcel H. Langoor
 Utrecht University
 Dept. of Analytical Molecular Spectrometry
 Section IR/Raman
 P.O. Box 80083
 3508 TB Utrecht
 phone +31 30 2536811
 fax +31 30 2518219
 e-mail: m.langoor { *at * }