Have you tried using the predict() function? You should be able to store factor scores using predict(fit). See the attached screen shot for how this would work using the example data from the lavaan package in R. 

Thank you very much. I will try in the morning, seems it would work. 

Yes. It works. Thanks

It works, even if we have 2nd order Latent Variable:

data(HolzingerSwineford1939)

## fit model

HS.model2

Can someone help with interpreting the output of the predict function, especially negative scores?

CFA uses a Z transformation. So the mean value becomes 0. Values less than the mean become negative. Values above the mean positive.

http://www.statistics4u.info/fundstat_eng/ee_ztransform.html

Example:

+1 = 1 standard deviation above the mean

+0.35 = 0.35 standard deviation above the mean

0.00 = score equal to the mean

-0.25 = 0.25 standard deviation below the mean

- 2.30 = 2.30 standard deviation below the mean

values greater than |3.0|

or

values greater than |4.0| --> see if they are outliers (could be just asymmetric data)

Thank you very much Matthew Barstead , I have been looking for this for a while.

greetings.