Having written a fair amount of software for the CW myself, I can comment
on a couple of technical issues.
Syncing to index is optional. You can either sync to the start of the
index pulse or simply initiate your transfer anywhere. Stopping can be
anywhere, up to 128K (the size of the RAM on the board) pulses later.
While LInux is nice, I prefer DOS and C under a DOS extender for a
platform. None of the timing uncertainty that a multitasking OS gives you.
2K/XP drivers are available for the later (MK 3) cards, but quite honestly,
I find the PnP nonsense to get in the way of doing productive work (the CW
software installs 8, count 'em, 8 drives for example). I run both of my
CW's on Win98-equipped machines booted into real (DOS) mode. If I need
networking or a GUI, I just type "win".
I've not found the signal drive characteristcs of the CW to be all that
robust, particularly for older drives on longer cables. Maybe Jens has
fixed this on the MK 4, but right now, I run the CW signals through a
separate buffer card with 7438 drivers and LS14's with 150 ohm pullups for
receivers (I know I could have simplified this, but I had a pile of the
chips and real estate wasn't important).
I imagine that a fair amount of Jen's cost is interfacing with PCI and
small production runs. AFAIK, he's still making them in Germany, which
can't be cheap, compared to outsourcing them to China.
I believe there's a fellow on the web who sells a 34-to-50 pin adapter with
a PIC installed to count seek pulses and handle RWC after track 40. Seems
like a worthwhile investment if you're going to handle a lot of 8 inchers.
Cheers,
Chuck