Two Plextors lose in DAE against AOpen

I ripped an Audio-CD that I burned some years ago. Although the CD is virtually scratchless. Both my Premium and my 716 (both latest FW) were slowing down in EAC and starting error correction. The same in Plextools. To keep it short: I bought my DVD-ROM AOpen 1648 AAP pro (1.04 FW) because of the great error correction it showed in a review (IIRC it was CDRinfo's review) and it came out better than the Plextors.

In more detail:
If not otherwise noted, EAC and "buffering drive" set to yes. (Premium and 716 do buffer and it's safer anyway albeit slower).

0) Premium without C2: saw error correction and aborted
1) Premium with C2, gap detect B (I know that gap detection should have no influence but I sometimes get EAC errors/rereads at the gaps)
2) Premium with C2, gap detect C
3) 716 with C2, gap detect C
4) 716 without C2, gap detect C
5a) Premium in Plextools mode 5 (recover best bits), aborted when slowing down due to errors
5b) 716 in Plextools mode 5 (recover best bits)
6) AOpen AAP pro, without C2

I then did an inverted mix. 1 and 2 were identical (so errors were corrected in the same manner). 1 and 3 as well as 1 and 4 as well as 3 and 4 were different. While 1 and 3 differed only in one click (mainly) around the error location, the version without C2 was significantly different for the rest of the song (inverted mix yielded nothing near zero). This may be due to phase displacement on account of the error/correction. 5 was also different from any of the above.

1 to 5 exhibited a click noise at the error location.

6 however read the track without using the EAC error correction at all (100% quality). There was also no click sound. After correcting the file offsets by hand and doing an inverted mix, the output was identical to 1-5 in the non-error regions.

My conclusion therefore is: The AAP pro does a superb job when it comes to DAE and error correction. I really did not expect that. I'm feeling now a bit weird, because I thought for years that the Plextors DAE capabilities were unmatched...

Any thoughts about my tests and findings?