No systematic variations calculated, weights missing in hepmc file

Hi,

I've got the problem that I don't get systematic uncertainties and the corresponding weights in the hepmc file if I use [QCD] at the end of 'generate', i.e. loop corrections. I'm using Pythia8 to shower the events, and the Madgraph version is 5.2.6.5.

As an example: If I do 'p p > t t~', I get systematic uncertainties and all the corresponding weights in the .hepmc file. The 'sm'-model is used.

If I do 'p p > t t~ [QCD]', I don't get systematic variations, and the weights are present in the .lhe file, but not in the .hepmc file. The 'loop-sm'-model is used automatically.
A further difference is that the run_card.dat looks quite different in the two cases (longer / more options in the first case).
Some pertinent details from the run_card in the [QCD] case:
lhapdf = pdlabel
 91700 = lhaid
 True = fixed_ren_scale
 True = fixed_fac_scale
 91.118 = muR_ref_fixed
 91.118 = muF_ref_fixed
 -1 = dynamical_scale_choice
 1.0 = muR_over_ref
 1.0 = muF_over_ref
 1.0, 2.0, 0.5 = rw_rscale
 1.0, 2.0, 0.5 = rw_fscale
 True = reweight_scale
 True = reweight_PDF
 True = store_rwgt_info

The summary I get in the first case (no [QCD]) looks like this:
# original cross-section: 611.966
# scale variation: +31.6% -22.5%
# central scheme variation: + 0% -45.1%
# PDF variation: +1.88% -1.88%
#
# dynamical scheme # 1 : 437.959 +28.3% -20.7% # \sum ET
# dynamical scheme # 2 : 356.783 +26.1% -19.5% # \sum\sqrt{m^2+pt^2}
# dynamical scheme # 3 : 449.901 +28.3% -20.7% # 0.5 \sum\sqrt{m^2+pt^2}
# dynamical scheme # 4 : 335.91 +25.4% -19.1% # \sqrt{\hat s}

The summary I get in the second case looks like this:
   --------------------------------------------------------------
      Summary:
      Process p p > t t~ [QCD]
      Run at p-p collider (6500.0 + 6500.0 GeV)
      Number of events generated: 100
      Total cross section: 8.034e+02 +- 5.8e+00 pb
   --------------------------------------------------------------

Any help with getting the systematics variations right in the loop-corrections case would be much appreciated.

Cheers,
Früd