Jules Richardson wrote:
OK, thinking about getting a catweasel board for the
museum.
...
2) Opinions of the board would be much appreciated,
particularly in the
context of what it *can't* do, how easy it is to code for, how easy it
is to get hold of others' code to handle a particular format (rather
than reinventing the wheel), how well the board copes with media errors
etc.
There are now four generations of catweasel cards -- the fourth is just
coming out now. It is an FPGA design where the FPGA gets set up via
software, so the developer can actually fix bugs and add features. The
mailing list was quite busy for a few weeks there but has been strangely
silent for a couple -- maybe I got unsub'd somehow.
The MK3 and earlier were OK for soft sectored/unsectored disks, but
reading hard sectored disks was significantly harder and writing them
much harder yet. The MK4 supposedly has/will have features to make
reading/writing hard sectored disks reasonable.
The "3rd party drivers" for TRS-80's that someone else mentioned in this
list was written by our own Tim Mann.
I think people have the wrong idea about this product. There isn't
corporation behind these cards -- it is a labor of love by one
individual supported by a number of hobbyists. It is a real company
making them, just a tiny one. Yes, you will have to write your own
software if you are expecting to read/write disk formats that aren't
already supported.
To give you some idea of the ethics of the company/individual (jens), I
bought a MK3 card a bit over a year ago. I wrote software to decode
some PTDOS disks that I have. I fed back some information to Jens about
how the MK3 fell short for reading/writing that particular format. He
had to spin the MK3 design to make the MK4 anyway (one of the key parts
was end-of-lifed), and he is incorporating some of that input into the
design. 9 months goes by. I get an email, apparently sent to a few
dozen people, saying: "What is your home address? I'm going to mail you
an MK4 card." A $100 card for free.
My only regret is at the moment I don't have the time to work on the MK4
card, but I will soon.
One pain about using the card was that I had to use it on an Win98
machine so I could do simple I/O to the thing. XP doesn't allow it
without drivers. Apparently under linux is isn't so hard. Anyway, this
time around they are working on a driver with a mostly common API
between linux/XP/otherthings so that you don't have to mess with the low
level IO and just program the thing without jumping through hoops.
Supposedly it will also be able to drive MK3 cards.
There are no hard real time constraints to programming it, making it
very simple. On MK3 boads you manually step it to the track you care
about then you tell the controller to read or write a track. All the
transition information gets captured in a RAM. When it is done you read
out the data or write the next track to the RAM. MK4 adds more logic to
tell apart the index hole from sector holes plus a state machine and
some other control bits to allow reading/writing individual sectors on
hard sectored devices.
If you expect it to be plug & play for some oddball format, you will be
disappointed. If you don't mind spending a week of evenings writing a
decoder/encoder in software, then it is a great card.