Overall, the clipping issue is still present, but definitely better. It looks like your using some sort of VU meter and adjusting to that, but the nature of these sounds makes it highly inaccurate. You're trying to record some complex transients that have peaks far shorter than what most meters will display. So when the meter says you're not clipping, it's just an average of hundreds of milliseconds, and your peaks are accruing within a handful of milliseconds. The easy solution is to just aim for a lower peak, ~-6db/-9db. It's not gonna hurt much and the level can be brought back up with software.
The VM-68 is a great example of something that looks alright on screen (as far as clipping), but ends up sounding like a pile of manure. Can't say what happened with this one, I'm guessing that the mic was too close to some gas release. Stepping back from the gun should give it a much more musical note.
I think my last post is still incredibly relevant, and for now, the best way I can help you with putting out good sounds. Maybe this spring I'll go down a pball field and try the agony of recording some of myself. For now, shoot me the raw clips of whats in b25 or any new stuff you got and I'll run it through some processing.
-Last post was referring to b24 sounds. Also, the pgp sounds out of place being the only one recorded indoors.