The Schatz lab just posted some pacbio bam files. A quick look shows a few interesting things, such as the read length looks like it clusters below 5k, but has consistent longer reads all the way out to 30k

Check out the rest in bam.iobio