I had a partial case head separation using a 1.8oz buffer

[LRRPF52]

[ATTACH=CONFIG]19151[/ATTACH]

Ok so I was able to enlarge the photo of the first failed case until it was 37.5" across my monitor. then, per my blue arrows, I measured the unsupported head area (or feed cone depth) at approximately 2.25". Scaling this down to the actual casing size I get an estimate of 0.0915" for the unsupported head area. Here is my math:

(2.25"/37.5") x 1.5245" = 0.0915"

Based on the failure score line location, this is my estimate of both the feed cone depth and the unsupported case wall depth. This is fairly close, bigger actually, than what I estimated from Bill Alexander's statements about the M16 style feed cone design.

So now taking 30% of this, I get .3 x 0.0915 = 0.027". <- This is the estimated additional feed cone depth for a standard M16 feed cone from an Alexander Arms feed cone, assuming that at least 30% improvement was achieved when Bill Alexander redesigned it.

This still puts this failure model as more significant than the bolt face depth out of spec model which was estimated at 0.005". Instead of 4x difference from my earlier estimation, it is now 5x different.

As such, the feed cone depth issue is still a viable failure model. This shouldn't be a surprise since Bill Alexander acknowledged it was an issue to begin with.

None of that approach is applicable though, since you don’t know how far the case was extracted already when the partial head separation occurred.

The side wall growth with that line is indicative of residual pressure on the case walls with a case that is already being extracted, not one that’s in battery.

This is again a textbook sign of early extraction, bad timing of the system.