1.) The Premiums (especially the first one) have extraordinarily fast and accurate C2 error correction if you can bypass the drive's caching. When ripping software says it can 'defeat' the caching, they are usually saying that the cache does in fact fill up with information, but they are smart enough to empty the cache out before attempting to re-read any particular sectors of the disc (otherwise you will always get the same results twice and never know when the disc is truly inconsistent). The bigger the cache, the more this slows down the ripping process. PlexTools or an obscure subcommand in EAC, are the only ripping processes I know of that can get completely around the caching of the Premiums. There's no indication that XLD does any secure ripping beyond the usual clear-out-the-cache-then-reread.
For writing, I strongly doubt that XLD or any native OSX software will let you adjust the laser strength (a feature called VariRec), or let you adjust the pit size (GigaRec/AMQR). At minimum, you are supposed to keep PlexTools around as a utility for unlocking those on a per-disc basis, even when PlexTools itself isn't the software you are using to carry out the burning. Now from a ripping standpoint, PlexTools will only bypass the caching for its own ripping process - it doesn't just grant cache-bypassing to any other ripping software you use while PlexTools is running in the background. EAC is capable of the cache-bypassing independently of whether or not you also have PlexTools.
2.) The Premiums are more about the bells & whistles you have to unlock, rather than the base characteristics of the drives. Most drives that can burn CDs slowly, don't produce much jitter, so the differences in the low-speed arena are marginal between the Premiums and nonPlextors. At higher burning speeds, the original Premium is fairly nice about keeping the jitter within a medium tolerance instead of letting it go crazy. While these burning behaviors should be true no matter what burning software you use, without the bells & whistles the base behaviors are hardly compelling enough reasons to choose the Premiums over cheaper or easier-to-find alternatives.
3.) Either of the Premiums will do perfect rips, as will nearly any real Plextor in the world. The main feature that sets Plextors apart is "Overread into Lead-In and Lead-Out", which is almost completely unheard of among nonPlextors. Pretend you're a police chief, there's a hostage situation, and you've got officers on the ground surrounding the perpetrator. You know the margin of 'kick' on your officers' firearms, but they aren't aware that there's a margin. You tell them to aim for the hostage, and shoot - knowing the bullets will actually hit the perpetrator. The officers refuse. This is how most nonPlextors behave. It's not that the Plextor "gun" doesn't have a margin - it's that the Plextors are more obedient. The Leads are like the hostages, the officers are the drives, and the shooting margins are the Read Offsets. There is something near the Leads that you're supposed to 'shoot', but you have to think you're barely aiming at the Leads in order get the very first or very last bit of data that other drives will refuse to read. Only in the ripping software where you can specifically say that your drive does full overreading (or you're ripping with PlexTools which already knows everything about every real Plextor), will this feature ever make a difference.
4.) The cacheless C2 ripping of the Premiums (when properly activated) is faster than any other secure ripping I have ever seen. For some reason the Premium is slightly faster on average about this than the Premium2, but this always depends on the album. Some albums the Premium2 will rip faster, and some albums the original Premium will rip faster. Despite this minor differences, either Premium will run circles around everything else, including the PX-760. When you're not doing cacheless C2, but when you're doing "cache-defeating" AccurateStream instead, the PX-760 might be faster than the Premiums, and this is a ripping process you may have to use on a very small handful of copy-protected CDs.
- I know enough not to trust the Premium-U's own external USB enclosure for Gap Detection methods B and C at least. Gap Detection isn't a heavily-explored feature when people are comparing external enclosures, so I believe everyone is on their own and in the dark for this. In any other manner of drive operations, most enclosures should be okay with most drive features, so while I can't recommend specific enclosures, I can at least say the odds are in your favor.
6.) I could imagine it being a problem with small-buffer drives burning at high speeds, but never with a Premium/Premium2 burning at 2x-20x. Shouldn't introduce any problems with reading, but who knows - maybe those adapters screw with -USEFUA, and maybe they mess with gap detection more than USB-to-IDE alone can already do.
7.) In EAC Options and Drive Options, leave everything unchecked that I don't specifically mention:
EAC Options > Extraction:
Fill Up missing offset samples with silence
Synchronize between tracks
Lock drive tray during extraction
Error recovery quality: High
EAC Options > General:
Use Alternate CD play routines
Disable 'CD' Autostart' for audio and data CDs while EAC running
On unknown CDs: open CD information dialog
Ask before overwriting files
Correct bug of wrong filename order in Windows multiple file dialog
Show status dialog after extraction
EAC Options > Tools:
Retrieve UPC / ISRC codes in CUE sheet generation
Use CD-Text information in CUE sheet generation
Automatically write status report after extraction
On extraction, start external compressors queued in the background: Your number of CPU cores (probably 2)
Do not open external compressor window
EAC Options > Filename:
I personally like %N. %T
and %N. %A - %T
But to each his own.
EAC Options > Write:
I uncheck EVERYTHING.
(except perhaps Disc-At-Once - which I haven't experimented enough with yet)
EAC Options > Write:
Use of SCSI interface:
"Native Win32" might be the only safe option on emulated EAC.
It's probably also the only selectable option.
Drive Options > Extraction Method:
Secure Mode with following drive features (recommended):
Drive caches audio data: UNCHECKED
Drive is capable of retrieving C2 error information: CHECKED
Drive Options > Offset / Speed:
Use red sample offset correction: +30
Overread into Lead-In and Lead-Out
Allow Speed reduction during extraction
CD-Text Read capable drive
Drive Options > Gap Detection:
Gap/Inex retrieval method: Detection method A
Detection accuracy: Secure
Drive Options > Writer:
Write samples offset: -30
Drive is able to write UPC/ISRC
Drive is able to write CD-Text (if you're a purist, you only want to enable this when you know your original CD actually had CD-Text, otherwise you are introducing CD-Text into a copy that was never in the original)
Go find a tutorial specific to which format you are interested in.
Also your commandline to EAC (in the form of a shortcut or whatever it is that you use to launch it) must include -USEFUA. If you cannot effectively implement this, you need to tell EAC that your drive does caching, and you should probably have it do AccurateStream instead of C2. The C2 behaves a little strange without -USEFUA, and AccurateStream behaves a little strange with functional -USEFUA.
You will also want to get into the habit of creating CUE sheets with each rip. There isn't a setting to automatically do this (like how the ripping logs can be auto-saved).
If you are going to do a lot of ripping but not a lot of burning, then don't waste your money on a Premium2. Most of their cost is for the burning features (slower burns, more adjustments with laser strength or pit size, etc..). The Premium2 is basically a Premium that is biased more toward burning (especially slow burning) - though either drive can "get the job done" with ripping or burning.