14 Mar
2005
14 Mar
'05
5:16 p.m.
From: "Jules Richardson"
<julesrichardsonuk at yahoo.co.uk>
On Mon, 2005-03-14 at 11:55 -0800, Dwight K. Elvey wrote:
What about cost? (irrespective of how the device physically connects to
the host machine)
I forsee four goals to make it useful:
o Cheap
Hi
When one considers how much hardware one saves over a PIC solution,
the DSP is cheap. They are in the $10 range.
From:
"woodelf" <bfranchuk at jetnet.ab.ca>
Look what you are doing is building a generic floppy disk controler.
The only high speed device what you use to sync the data/clock pulses
to the system cpu clock. The rest is software. I'd sooner use a CPLD
designed for generic bit sampling but a PIC would also work with
a digital data/clock seperator. Now would getting the people who do
cat-weasel create a USB version be a better goal?
Ben alias woodelf
PS. What about hard-sectored floppy disks, that too may need reading
too?
Hi
I think you are missing what I am saying. The SPI is just
a shift register that takes an external clock. It can be programmed
to automatically DMA transfer into memory. It is the perfect zero
additional logic circuit to use. You don't need to build a
data/clock separator or anything. Just sample the data.
One could even make the output SPI provide write data. These
chips are designed to load their programs from a single
flash or EPROM so the entire hardware requirement is almost
nothing.
I see others on comp.os.cpm talk about using a 50MHz variant
of a Z80. I think most miss the point. These DSP's are 30 MIPS+
not just 50 MHz clocks. They have enormous capabilities in
a relatively small package. It was like they were designed for
this project. You don't need to create a CPLD since the hardware
part is already done for you. o Simple to build by anyone with a few electronics
skills.
A little tough here. 50 mil spaces surface mount. Some soldering
skills required.
o Open 'source' (all schematics etc.
available)
No issue here.
o Easy / quick connectivity
Although, most computers come with a parallel port, using
the port on a newer XP for other than normal purposes may be a pain.
Catweasel seems to lose out on 1, 3, and 4 - and 2 isn't relevant in its
case. Can't comment on how nice its software API is as I haven't looked
at it yet, but doubtless a bunch of us on this list could come up with
something that'd cater for all tastes (plus the really low-level
software would all be open source anyway!)
Personally I'm not a fan of a USB version though; I'd rather have
parallel as pretty much any machine has a parallel port - USB limits me
to newer PCs and Macs (plus software interfacing *might* be harder).
Priorities seem to me to be (highest first):
o Reading disks
Should be easy.
o Writing back a disk image
I see only minor issues here ( pre-comp )
o Decoding disk data on host machine
Just requires some one to do it.
o Modifying disk data on host machine, re-encoding
back to floppy
If one can do the above, this one comes along, almost
for free.
Later
Dwight
Happily, that's probably order of complexity too, easiest first :) (I am
coming at this from a preservation point of view, rather than being able
to create disk images for use with emulators, say)
cheers
Jules
14 Mar
14 Mar
10:47 p.m.
Dwight K. Elvey wrote:
Catweasel: $90-ish. That is its disadvantage, but it offers a lot of
convenience that a $10 DSP (OK, plus $10 in random analog bits, plus $10
worth of sockets, and $xxx worth of time wirewrapping, etc).
Catweasel wins this one hands down. Just plug it in.
Except that it doesn't exist. A hypothetical thing has any advantage
that you wish to confer on it.
The catweasel design isn't open source, but I believe just about all the
drivers are.
There are a number of drivers for linux/amiga/others that exist, and the
driver for windows is coming along. Someone else has already suffered
the pain for us.
My own desire that motivated me to buy a catweasel was to collect and
archive information. The fact that the collection was done on a
"modern" machine (there are isa and pci versions of the card) doesn't
bother me at all.
These last three itmes were the main sticking point for a number of
people. The catweasel should be able to handle just about any format;
it is "just" a simple matter of software yet to be written. Yet it
hasn't been done for many formats.
The MK3 and earlier catweasels were deficient in their ability to write
hard sectored disks efficiently, but MK4 should fix that. At the HW
level read and writing are both simple; writing the flux pattern in
software is easier than decoding it.
Reading through a lot of the postings on this thread makes me think of
somebody complaining that "linux is too big, bloated, and complicated,
and redhat charges for their version too. let's write our own version
that will be lean and mean and available to all for free." Thinking
about the design is fun and the one who actually does it will have some
rewards implementing it, but for most people, the already available
solution will be superior to the proposed design even if it is ever
done. Spending $60 more for a finished, professionally done board might
be a showstopper for some people, but futzing with the building your own
will preclude a great many more.
>From: "Jules Richardson"
<julesrichardsonuk at yahoo.co.uk>
...
What about
cost? (irrespective of how the device physically connects to
the host machine)
I forsee four goals to make it useful:
o Cheap
Hi
When one considers how much hardware one saves over a PIC solution,
the DSP is cheap. They are in the $10 range. o Simple to
build by anyone with a few electronics skills.
A little tough here. 50 mil spaces surface mount. Some soldering
skills required. o Open
'source' (all schematics etc. available)
No issue here. o Easy /
quick connectivity
Although, most computers come with a parallel port, using
the port on a newer XP for other than normal purposes may be a pain. >Catweasel seems to lose out on 1, 3, and 4 - and 2
isn't relevant in its
>case. Can't comment on how nice its software API is as I haven't looked
>at it yet, but doubtless a bunch of us on this list could come up with
>something that'd cater for all tastes (plus the really low-level
>software would all be open source anyway!)
#2 isn't relevant -- what do you mean? Sure, some people want to tinker
with HW for the sake of tinkering with HW. I have no fear of
electronics but with my limited time, I get more satisfaction in other
activities. I'd rather spend a few $$ and save hours of effort
collecting, building, troubleshooting so that I have time to do those
other things. I suspect that I'm in the majority otherwise this project
would have been done already.
Even the lowest level software of the catweasel is open source. So
there is no advantage in this respect for the putative design.
Personally
I'm not a fan of a USB version though; I'd rather have
parallel as pretty much any machine has a parallel port - USB limits me
to newer PCs and Macs (plus software interfacing *might* be harder).
Priorities seem to me to be (highest first):
o Reading disks
Should be easy.
o Writing back a disk image
I see only minor issues here ( pre-comp )
o Decoding disk data on host machine
Just requires some one to do it. o Modifying
disk data on host machine, re-encoding back to floppy
If one can do the above, this one comes along, almost
for free. 
11:34 p.m.
On Mon, 14 Mar 2005 21:47:28 -0600, Jim Battle <frustum at pacbell.net> wrote:
Reading through a lot of the postings on this thread
makes me think of
somebody complaining that "linux is too big, bloated, and complicated,
and redhat charges for their version too. let's write our own version
that will be lean and mean and available to all for free." Thinking
about the design is fun and the one who actually does it will have some
rewards implementing it, but for most people, the already available
solution will be superior to the proposed design even if it is ever
done. Spending $60 more for a finished, professionally done board might
be a showstopper for some people, but futzing with the building your own
will preclude a great many more.
I didn't know what a catweasel was. Yes, I agree that it is the right solution.
regards,
vax, 9000
15 Mar
15 Mar
7:31 a.m.
On Mon, 2005-03-14 at 21:47 -0600, Jim Battle wrote:
Except that it doesn't exist. A hypothetical thing has any advantage
that you wish to confer on it.
I think Dwight meant 'no issue' as in, no problem releasing any
schematics, source etc.
I don't forsee the hardware design being that complex. Parallel port
interfaces are easy (even I can do that bit ;) and driving the buffer
memory shouldn't be difficult.
Software-wise, remember that primary goal is archive of old media onto
modern systems (and restore back to floppy). It doesn't need
understanding of the data stream on the host machine to do that - it's
essentially just save and replay. Maybe an hour to write and test code
under each OS to do that.
Dwight K. Elvey wrote:
Catweasel wins this one hands down. Just plug it in.
Turn off your running machine, open the case, hope you have a free slot
(I certainly don't on this here desktop), start up, hope you don't have
IRQ troubles (PCs don't seem to handle loaded PCI buses well).
Take case off, unplug board when you want to go to a different site. Go
through previous steps again.
To me that's a lot more hassle than simply plugging in a *simple* device
to the parallel port of a machine. Remember we're not talking about
complex electronics here - couple of connectors, some RAM and a bit of
glue logic and that's it.
(I just thought of another bonus. The parallel - or USB - cable can
probably be quite long, meaning the drives that you want to handle can
be up on the desk, say. With the catweasel, you're limited to floppy bus
length I assume which means hassle every time you want to use a
different drive with it)
o Simple
to build by anyone with a few electronics skills.
A little tough here. 50 mil spaces surface mount. Some soldering
skills required. o Open 'source' (all schematics etc.
available)
No issue here. The catweasel design isn't open source, but I
believe just about all the
drivers are.
Good. If at some stage someone writes a layer to take 'our' raw data
stream and do clock separation etc. then we can just use those open-
source drivers for manipulating specific format data.
My own desire that motivated me to buy a catweasel was
to collect and
archive information. The fact that the collection was done on a
"modern" machine (there are isa and pci versions of the card) doesn't
bother me at all.
I'm not knocking the catweasel although I know it sounds like I am. It's
great if you have one or two formats you want to read/write, you know
what drives you want to use, and you're just using a single machine.
But the financial outlay might be discouraging to some people
(particularly given the complexity of the board and lack of schematics).
My goal with a parallel/USB solution is something cheap and cheerful and
portable, primarily designed for archival purposes and not necessarily
complete decoding.
#2 isn't relevant -- what do you mean? Sure, some
people want to tinker
with HW for the sake of tinkering with HW. I have no fear of
electronics but with my limited time, I get more satisfaction in other
activities. I'd rather spend a few $$ and save hours of effort
collecting, building, troubleshooting so that I have time to do those
other things. I suspect that I'm in the majority otherwise this project
would have been done already.
I see it as akin to something like the SGI keyboard convertor that's
been mentioned on the list before (or several other "homebrew" type
projects that are floating around). A bit of simple hardware that anyone
with a few skills can quickly build - but there's potential for offering
a ready-built board too if there's a low-volume market for it.
But yes, it's more for the hardware people among us, not those who enjoy
playing with software or machines in as-is configuration.
As for not being done, I suspect it's because nobody's given it serious
thought and assumed it to be more complicated than it is. We're all
guilty of being sloppy about archival - I suspect a good 80% of
collectors aren't in the hobby for the preservation aspect, they just
want to play around with the machines. A project such as this is for the
other 20% who want to preserve the hardware and software and keep it
going for as long as possible.
[quick snip of things I agree with]
Reading through a lot of the postings on this thread
makes me think of
somebody complaining that "linux is too big, bloated, and complicated,
and redhat charges for their version too. let's write our own version
that will be lean and mean and available to all for free."
Yes. Except that it's a billion times more simple than taking on writing
an OS :-)
Thinking about the design is fun and the one who
actually does it will
have some rewards implementing it, but for most people, the already available
solution will be superior to the proposed design even if it is ever
done.
Agreed on the design front. It'd be nice to build something and see the
data being spat out at the other end; kind of like proof of concept code
at the start of a software project.
As for 'most people' I have no idea. I'm not sure how many catweasels
have been sold and to what sort of people. But I think the catweasel and
this cater for different markets anyway, so it's maybe unfair to compare
the two. The only reason I've brought the catweasel into this is because
I wanted something for the museum to archive all our media - and with a
bit of work a MK4 could probably do the job. But that doesn't mean to
say I'm not going to look into alternatives too.
Spending $60 more for a finished, professionally
done board might
be a showstopper for some people, but futzing with the building your own
will preclude a great many more.
More than likely :) That doesn't stop it from being a worthwhile project
with its own benefits, though.
cheers
Jules
7065
days inactive
7066
days old
3 comments
4 participants
participants (4)
-
dwight.elvey＠amd.com -
frustum＠pacbell.net -
julesrichardsonuk＠yahoo.co.uk -
vax9000＠gmail.com