One of the reasons I ask relates to my knowledge of my vette programming. It's got (had) some weird stuff. And, the thing I didn't like the most was how the timing tables were set up.

In 89, GM used a primary timing table with columns for load and rows for RPMs. It also used secondary tables for PE (power enrichment). In the secondary tables, additional timing and fuel were added (summed) to the primary tables (creating a final total of fuel/timing for extra zippy-do-da-day.) Oddly, I found errors in the tables...or at least the ones claiming to be stock tables as stored on programmers websites (like Moates/Tunerpro).

To confirm the stock tables on tuner websites appeared accurate, I burned a chip with one...made the minimal changes for my 383 (FI size, min rpm, etc) and installed that chip. I observed very similar behavior to stock.

I could post pics of tables if you thought it might be relevant, but in the higher load columns, primary timing dropped -- fairly significantly. That's where the PE values were added back in...putting total timing back in the same ball-park as the lower load columns. (Keep in mind higher loads still have less total timing than lower rows but the table would not look linear w/o seeing the PE adders.)

There were also a stray cell here and there that was unexplainable. One were timing was 5-10 degrees lower than values in all adjacent cells. It made me wonder if timing tables were created using a mildly faulted algorith, if GM engineers were drunks, or if the tuner sites recreated tables while on drugs. In either case, I corrected the handful of timing table cell errors to observed driving performance. It was a bit smoother/better.

Finally, I removed all PE values (common practice in DIY tuner circles) and made my primary timing table show the summed timnig curves. This resulted in significantly smoother transitions since PE didn't pop in here'there. IOW, I no longer could feel the transitions while driving the car -- which was good.

If GM and Isuzu shared any timing technology, maybe the later ODB2 systems use similar techniques in their timing formulas. If so (and depending on how you're seeing your timing values), it could explain the sudden drop in timing. Since 25% TPS is where you see fuel PE, maybe that's also where timing PE kicks in. And, if the PE table has a starting value of something like 10-degrees, you may need to add it to the 16-deg in the base table -- for a summed value of 26-deg total timing.

Just a guess here, but it's the only thing I can think of after seeing (and recoding) my older GM ODB1 system.