Yes, I've been looking more closely at the LMBA Armenian yDNA haplogroup E-M84, of which there are very many Ashkenazi samples. These Ashkenazis samples appear to fall into two main groups - Y14899 and PF6747.
I had already noticed that Y14899 is almost exclusively Ashkenazi, with the exception of two basal samples (a Sephardi and a Palestinian) - all deriving from a recent bottleneck (meaning it tells us little about more distant origins).
PF6747 looks similar, with an Ashkenazi-heavy branch (Z21018) and including some Sephardis and a Yemeni - but each with a substantially more distant (Bronze Age) TMRCA than Y14899 (per both yfull's estimates and my own).
I would suggest this swings the balance towards the proto-Ashkenazi E-M84 being of a Southern Levant/Red Sea/generic Jewish origin, rather than Anatolian, which might also explain the relatively heavy Egyptian-like aDNA in Ashkenazis. Interestingly, the closest related PF6747 I can find to Z21018 (in terms of STRs) is a modern Armenian.
I am tentatively leaning towards Armenian E-M84 making a more ancient (Bronze Age?) contribution to the Jewish gene pool, which thrived in individuals who were ancestral to proto-Ashkenazis, but was heavily diluted (or not heavily admixed in the first place) into individuals who were ancestral to other Jewish groups. However, I suppose this would be dependent on there being some kind of caste system within early Jewish populations that severely limited admixture between the ancestors of the different groups.