Find enclosed a solution for Excel. It is a little bit dirty
and small, but it works. In the enclosed sheet i restrict to
only a few (16) observations (since dynamic arrays are awfull
in Excel, but can be done), ignored weights (due to speed) and
used a logistic function as test.

Due to L-M one has to choose a rough but not completely absurd
initial guess for the parameters.

Excel can not compute symbolic or reasonable parse expressions
i use 'precomputed' functions and gradients (which is ok for
me). For a different parametric function one has to adapt f
and gradf within the VBA source and possibly the number of
parameters (and observations).

I also added a protocol from the original Maple solution (since
the easiest is to let an CAS do the work and link it to Excel
getting the desired flexibility) showing the statistics for it:
it is an example that a good statistics is not good enough.

Within VBA check through the menu bar tools / references whether you have
the following 4 'standard ones' activated: Visual Basic for Applications,
Microsoft Excel 9.0 Object Library (EXCEL9.OLB), OLE Automation (not need) 
and Microsoft Office 9.0 Object Library (MSO9.DLL).


David Holmgren once gave me the link 
http://www.cs.ubc.ca/spider/lowe/papers/pami91/pamilatex.html 
to look at the section "Solving for viewpoint and model parameters"