The QC probes on the random primer Affy arrays are completely different from what they put on the 3'-biased arrays, and how you get them is different as well, as you have to use the oligo/pdInfo pipeline rather than the affy/makecdfenv pipeline. Normally you would be able to query the pd.hta.2.0 database directly, but we are unfortunately in a period where the pd.hta.2.0 database doesn't have the right data for the probeset types, and since Benilton is refactoring it for the next release, it won't get fixed until April.
Note that this only extends to the part of the database that says what 'type' a probeset is. The mapping of probes to probesets is unaffected by this issue.
However, there is still a way to get these data using the netaffxTranscript.rda that comes with the pd.hta.2.0 package:
> load(paste0(path.package("pd.hta.2.0"), "/extdata/netaffxTranscript.rda"))
> annot <- pData(netaffxTranscript)
The 'Main' probes are those that are intended to measure transcripts, and all these other probesets are intended to be controls of one type or another.
Note that the ordering of this object (annot) will NOT be the same as the ordering of your data, so you have to make sure you get things ordered correctly!