Consider

$ cc_vib_analyze.py -m 300 -M 7000 -f50000 -p -G -d 20080919/20080919121041_zp_time_50000Hz

The guess looks reasonable.  It is a bit off due to some popcorn
noise, but it looks like it should fit.  The fit, however, looks bad.
It is definately too flat.  Looking at the git.log gnuplot output, we see

  After 10 iterations the fit converged.
  final sum of squares of residuals : 2.58573e+09
  rel. change during last iteration : -6.04806e-06
  ...
  Final set of parameters            Asymptotic Standard Error
  =======================            ==========================
  
  A               = 5115             +/- 184.1        (3.599%)
  B               = 5476.03          +/- 433.1        (7.909%)
  C               = 2.73854e+18      +/- 4.604e+17    (16.81%)
  ...
  correlation matrix of the fit parameters:
  
                 A      B      C
  A               1.000
  B               0.905  1.000
  C               0.970  0.963  1.000

The high correlation matrix can't be helping...