Perhaps you could retry to post the vi, since it seems to not be here. The 3 most common reasons (IMO) for a failed fit are 1) poor initial guesses which cause an under/over flow or 2) using the proper scaled values for the fit (see: http://digital.ni.com/public.nsf/3efedde4322fef19862567740067f3cc/6f23b120b7a2c585862566f700130a0a?OpenDocument) or 3) using an invalid format for the model equation.
If you wire an indicator to the Error Output node of the L-M vi does anything come up? This can be pretty useful in diagnosing the issue.
