I could fix the Acrobat’s “3d data parsing error” using pdf-parser.py and qpdf:

1) Uncompress the original 3dpdf:
qpdf –stream-data=uncompress unc.pdf original3dpdf.pdf

2) Obtain the 3D object ID in the uncompressed 3dpdf (6 in my case):
./pdf-parser.py -t /3D unc.pdf

3) Extract u3d file with pdf-parser.py:
pdf-parser.py -o 6 -d unc.u3d unc.pdf

After that I compiled the .tex (beamer) with pdflatex and it works perfect in Acrobat.