That Plextor is not a real one, it's just a rebranded BenQ drive, but yesterday I ordered a real one (PX-716A). As soon as the drive arrives I will create a .sub file with it, too and will use px_d8 on those suspicious sectors. If I need futher help on this I'll ask.I still bid on 02.74 as I still claim that CloneCD is to dumb to handle the gap between data and first audio track properly. The problem is that both for Plextor.sub and LG1.sub drives were used which are simply unable to read the sectors of the pregap. What CloneCD does is nothing other than performing normal READ CD command - the very same command that CDTool executes when using its sector viewer, but how CloneCD interprets the returned data is another thing.
LG1: With CDTool I am unable to read LBA 41734 to 41807, so the contents of the .sub file for those sectors are a best guess from CloneCD I assume, but nothing that actually came from the drive. That would explain the fallback to Q-Control=0x06 as CloneCD might (wrongly) assume that in this range of LBAs is still within the data track and still before the common pregap of 02.00.
Plextor: With CDTool I am unable to read LBA 41734 to 41957, i.e. the entire pregap area (asummed it really is of length 02.74). When Plextor drive reads data sectors it has a subchannel data offset of +1, i.e. returns the subchannel data from one sector ahead. That way CloneCD still receives the subchannel data of the first pregap sector with relative position of (-)02.73 which is saved in .sub file for LBA 41734.
I claim that CloneCD assumes that the pregap between data track and first audio track being of length 02.00, starting with relative positions (-)02.00 and ending with (-)00.01. Everything falls into place. It seems that once a read error occurs while reading first audio track's pregap sectors CloneCD skips reading further sectors. But regardless of sectors being able to be read or not - CloneCD seems to align the received subchannel data according its assumption. Even in the LG2.sub file, created using a drive which is able to read -all- pregap sectors without any problem, the subchannel data of the first sector in pregap is doubled and the subchannel data of the last sector in pregap is dropped. On drives which fall into a read error during reading pregap sectors CloneCD skips the rest of the pregap and subchannel data for that sector range is regenerated according to CloneCD's assumption, generating Q-Control 0x06 subchannel data until reaching the 150th sector before Track 2 according to the CD's TOC.
Me too hopes for Fireball taking a look at the subs just to be able to take a third opinion about the whole problem into account. 
EDIT: Here is the TOC of the CD...
Track 01 (Mode 1, LBA: 0)
Track 02 (Audio, LBA: 41958)
Track 03 (Audio, LBA: 70239)
Track 04 (Audio, LBA: 95789)
Track 05 (Audio, LBA: 121439)
Track 06 (Audio, LBA: 156831)
Track 07 (Audio, LBA: 173868)
Track 08 (Audio, LBA: 195981)
Track 09 (Audio, LBA: 217572)
Track 10 (Audio, LBA: 242641)
Lead-Out (LBA: 265118)
PS: Just for interest... Are you using the program Subcode analyzer by nue for inspecting the .sub files?
]]>