10 Mar
2005
10 Mar
'05
7:49 p.m.
What we need so badly is a USB universal floppy disk controller. It has to
do 8" (single and/or double sides & density), 5.25" (both 360k and 1.2MB)
and 3.5", at least 2 drives of each type. Ideally totally configurable to
read/write any format, even "non-standard" formats (including 1k sectors on
8" drives, Microsoft's 3.5" "DMF" format, etc.) Can't anyone
design one?
It's way over my head, but I'd buy one in a flash.
I'm desperate enough that I've been taking apart 3.5" USB drives hoping to
find one that was a USB controller with a standard 3.5" drive. So far, they
have ALL been "integrated" devices.
10 Mar
10 Mar
8:25 p.m.
On Thu, 2005-03-10 at 19:49 -0500, Barry Watzman wrote:
What we need so badly is a USB universal floppy disk
controller.
Personally I'd rather have parallel or serial or SCSI interfaces, but
each to their own :-)
But yep, agreed. I certainly need *something* that can archive / restore
classic formats to/from modern(-ish!) media.
Can't anyone design one? It's way over my head
Over mine too really. I'd be able to build one given a schematic, but
probably couldn't design something from scratch.
Actually, I'd much prefer a schematic anyway for peace of mind as well
as curiousity value.
Given the know-how on the list I'm suprised there hasn't been a
collective effort in the past to build something.
Maybe the majority of us either:
1) only specialise in one machine and so do backups in whatever way's
easiest for our own hardware,
2) are lucky enough to be blessed with online archives of software so
backups aren't needed
3) aren't taking any backups at all! :-)
cheers
Jules
11 Mar
11 Mar
2:51 a.m.
In message <1110504341.4413.91.camel at weka.localdomain>
Jules Richardson <julesrichardsonuk at yahoo.co.uk> wrote:
But yep, agreed. I certainly need *something* that can
archive / restore
classic formats to/from modern(-ish!) media.
Shift register + counter + high speed RAM. Could even do it with an FPGA if
you wanted.
Buffering an entire track with 8x oversampling is going to be hellishly
memory intensive though.
3) aren't taking any backups at all! :-)
Aaghh! That reminds me - it's a good year since I backed up the HDD on my
machine... [fx: hard drive starts going kaclunk-kaclunk-kaboooom]
Later.
--
Phil. | Acorn Risc PC600 Mk3, SA202, 64MB, 6GB,
philpem at philpem.me.uk | ViewFinder, 10BaseT Ethernet, 2-slice,
http://www.philpem.me.uk/ | 48xCD, ARCINv6c IDE, SCSI
... --T-A+G-L-I+N-E--+M-E-A+S-U-R+I-N-G+--G-A+U-G-E--
12:16 p.m.
On Fri, 2005-03-11 at 07:51 +0000, Philip Pemberton wrote:
Oh boy. I rsync data from my desktop to an external hard disk
periodically (usually once or twice a week). Hadn't done it for a few
weeks, kicked off a run the other day, and the wholse SCSI subsystem
fell over.
Panic for a moment, thinking my desktop drive had died. Turns out I'd
set the SCSI data rate to the external disk higher than normal when I
was playing around with something else, and forgotten to set it back
again - so it was just a SCSI bus transfer problem, not a hardware
issue. Phew.
cheers
J.
In message <1110504341.4413.91.camel at
weka.localdomain>
Jules Richardson <julesrichardsonuk at yahoo.co.uk> wrote:
Well if worst-case is 1mbps data rate at 300rpm and 8x oversample, isn't
that (8 * 10240 * 1024) / 5 = 1677722 bits of memory maximum needed?
(div by 5 because 300rpm gives you a whole track in 1/5 of a second)
So 256KBytes of RAM should always be enough to buffer a whole track,
regardless of what physical drive or media you use.
Presumably there's no need to have any kind of microprocessor control;
it can all be done with a free-running clock to read from the drive or
write to it, and a few control lines to for read/write, start, stop etc.
Can static RAM can't cope with these kinds of speeds? Presumably cache
chips pulled from old PC motherboards would cope happily, even if old
80's parts wouldn't.
It's probably still beyond my abilities to design something that'd
work :-) But on the surface it doesn't sound *that* complicated for
someone who's used to designing digital circuits.
But yep, agreed. I certainly need *something*
that can archive / restore
classic formats to/from modern(-ish!) media.
Shift register + counter + high speed RAM. Could even do it with an FPGA if
you wanted.
Buffering an entire track with 8x oversampling is going to be hellishly
memory intensive though. 3)
aren't taking any backups at all! :-)
Aaghh! That reminds me - it's a good year since I backed up the HDD on my
machine... [fx: hard drive starts going kaclunk-kaclunk-kaboooom] 
1:17 p.m.
>>>> "Jules" == Jules Richardson
<julesrichardsonuk at yahoo.co.uk> writes:
Jules> On Fri, 2005-03-11 at 07:51 +0000, Philip Pemberton wrote:
> ...
> Shift register + counter + high speed RAM. Could even do it with
> an FPGA if you wanted. Buffering an entire track with 8x
> oversampling is going to be hellishly memory intensive though.
Jules> Well if worst-case is 1mbps data rate at 300rpm and 8x
Jules> oversample, isn't that (8 * 10240 * 1024) / 5 = 1677722 bits
Jules> of memory maximum needed?
Jules> (div by 5 because 300rpm gives you a whole track in 1/5 of a
Jules> second)
Jules> So 256KBytes of RAM should always be enough to buffer a whole
Jules> track, regardless of what physical drive or media you use.
Even if you want 8 bits per sample, that's still only 1.5 MB, and the
data rate is 7.5 MB/s (since a track is 200 ms by your calculation).
That's a trivial bandwidth requirement; anything better than antique
DRAM will do the job easily, and the space requirement is tiny, too.
I'm wondering if you can do this job with a microcontroller similar to
a PIC, running the sampling loop in software. If not quite, it should
be pretty close.
paul
5:10 p.m.
In message <1110561380.6178.63.camel at weka.localdomain>
Jules Richardson <julesrichardsonuk at yahoo.co.uk> wrote:
I really should look into buying an external USB hard drive casing and a
server-rated IDE/ATA hard drive (probably a Maxtor MaXLine II or a Seagate
Barracuda)
Well if worst-case is 1mbps data rate at 300rpm and 8x
oversample, isn't
that (8 * 10240 * 1024) / 5 = 1677722 bits of memory maximum needed?
(div by 5 because 300rpm gives you a whole track in 1/5 of a second)
And therein lies the error in my calculations. I came up with 1MByte per
track...
Presumably there's no need to have any kind of
microprocessor control;
it can all be done with a free-running clock to read from the drive or
write to it, and a few control lines to for read/write, start, stop etc.
Why not stuff it all in a CPLD? Maybe create a logic-only version too...
Can static RAM can't cope with these kinds of
speeds? Presumably cache
chips pulled from old PC motherboards would cope happily, even if old
80's parts wouldn't.
70nS SRAM should work fine, and most SRAM is pretty cheap too. I've got a
512kbit here that cost about ?4 and is rated to 70nS IIRC.
Aaghh! That
reminds me - it's a good year since I backed up the HDD on my
machine... [fx: hard drive starts going kaclunk-kaclunk-kaboooom]
Oh boy. I rsync data from my desktop to an external hard disk
periodically (usually once or twice a week). Hadn't done it for a few
weeks, kicked off a run the other day, and the wholse SCSI subsystem
fell over. Panic for a moment, thinking my desktop drive had
died. Turns out I'd
set the SCSI data rate to the external disk higher than normal when I
was playing around with something else, and forgotten to set it back
again - so it was just a SCSI bus transfer problem, not a hardware
issue. Phew.
You're using SCSI in a desktop system? Heh. I should probably see about
trying that sometime.
Later.
--
Phil. | Acorn Risc PC600 Mk3, SA202, 64MB, 6GB,
philpem at philpem.me.uk | ViewFinder, 10BaseT Ethernet, 2-slice,
http://www.philpem.me.uk/ | 48xCD, ARCINv6c IDE, SCSI
... DOS means never having to live hand-to-mouse
7:22 p.m.
On Fri, 2005-03-11 at 22:10 +0000, Philip Pemberton wrote:
Well personally I prefer logic-only because I know I'll have the parts
here, plus I don't have any sort of programmer for making more exotic
things (ok, so I do have an ancient PAL/EPROM programmer, but without
manuals I've only ever figured out the EPROM side of it :)
But also I think that keeping it simple (and understandable) should be a
goal to enable it to buildable by anyone with a few electronics skills.
If it means it encourages more people to build it and back up classic
media, it's got to be a good thing :-)
OK so four of those would be enough. That's not *too* costly I suppose,
although it probably makes it enough for people to think twice about
it.
I still like the idea of SRAM from old PC motherboards; didn't they
normally have 64 or even 128KBytes? Free then; with other parts costs
it'd probably make the whole thing do-able for a fiver or so (more with
a few frills like a case, PSU etc.)
Presumably
there's no need to have any kind of microprocessor control;
it can all be done with a free-running clock to read from the drive or
write to it, and a few control lines to for read/write, start, stop etc.
Why not stuff it all in a CPLD? Maybe create a logic-only version too... Can static RAM
can't cope with these kinds of speeds? Presumably cache
chips pulled from old PC motherboards would cope happily, even if old
80's parts wouldn't.
70nS SRAM should work fine, and most SRAM is pretty cheap too. I've got a
512kbit here that cost about ?4 and is rated to 70nS IIRC. You're using SCSI in a desktop system? Heh. I
should probably see about
trying that sometime.
Yeah, I bailed on IDE for PC systems about 8 or so years ago and have
only ever used SCSI since. It's nice having the DVD drive, CDROM, DAT
etc. up on the desk and the rest of the machine tucked away underneath.
Handy for imaging old Unix workstation drives too, reading old QIC24
tapes (as per last week when I slogged through heaps of them) etc.
cheers
Jules
8:20 p.m.
In message <1110586962.6194.110.camel at weka.localdomain>
Jules Richardson <julesrichardsonuk at yahoo.co.uk> wrote:
The Altera ByteBlaster and Xilinx Parallel Download Cables are literally
three-component devices - a DB25 connector, a 74LS244 and an IDC header. Plug
the cable into the target board, apply power, run the downloader, hit
"Program", then wait a few seconds for the fuse data to download.
Not to mention the fact that the output from a floppy drive is 1-bit binary.
Adding a 74LS164 shift register would mean you could use a slower RAM at the
expense of slightly more complicated logic and data decompiling.
Writing to disc would also be a very useful feature, lower priority than
reading of course. USB Option Board anyone?
Why not stuff
it all in a CPLD? Maybe create a logic-only version too...
Well personally I prefer logic-only because I know I'll have the parts
here, plus I don't have any sort of programmer for making more exotic
things (ok, so I do have an ancient PAL/EPROM programmer, but without
manuals I've only ever figured out the EPROM side of it :) But also I think that keeping it simple (and
understandable) should be a
goal to enable it to buildable by anyone with a few electronics skills.
If it means it encourages more people to build it and back up classic
media, it's got to be a good thing :-)
Only problem is, you might not be able to get TTL logic chips up to the
necessary speeds.
70nS SRAM
should work fine, and most SRAM is pretty cheap too. I've got a
512kbit here that cost about ?4 and is rated to 70nS IIRC.
OK so four of those would be enough. That's not *too* costly I suppose,
although it probably makes it enough for people to think twice about
it. I still like the idea of SRAM from old PC
motherboards; didn't they
normally have 64 or even 128KBytes? Free then; with other parts costs
it'd probably make the whole thing do-able for a fiver or so (more with
a few frills like a case, PSU etc.)
I've got a TAG RAM here from a Pentium board and it's only 16k. I'm still
trying to find a few more of them...
Yeah, I bailed on IDE for PC systems about 8 or so
years ago and have
only ever used SCSI since. It's nice having the DVD drive, CDROM, DAT
etc. up on the desk and the rest of the machine tucked away underneath.
I haven't seen any SCSI DVD writers (or readers for that matter) for ages. I
think a few of the first writers were SCSI, but now they all seem to be
IDE/ATA.
I did have a SCSI card in my RiscPC, but I replaced it with a 100BaseT LAN
card, seeing as I wasn't using the SCSI (no SCSI devices to use with the
card).
Later.
--
Phil. | Acorn Risc PC600 Mk3, SA202, 64MB, 6GB,
philpem at philpem.me.uk | ViewFinder, 10BaseT Ethernet, 2-slice,
http://www.philpem.me.uk/ | 48xCD, ARCINv6c IDE, SCSI
... Hi! I'm a tagline virus! Steal me & join in the fun!
12 Mar
12 Mar
3:02 p.m.
On Sat, 2005-03-12 at 01:20 +0000, Philip Pemberton wrote:
Hmm, just grabbing a random 486 board from the pile here; it's got 8
61256 chips on it which is presumably 256KB in total. I suppose by the
Pentium era the cache was all in the CPU or something though...
In message <1110586962.6194.110.camel at
weka.localdomain>
Jules Richardson <julesrichardsonuk at yahoo.co.uk> wrote:
The Altera ByteBlaster and Xilinx Parallel Download Cables are literally
three-component devices - a DB25 connector, a 74LS244 and an IDC header. Plug
the cable into the target board, apply power, run the downloader, hit
"Program", then wait a few seconds for the fuse data to download.
Forget all I said then - that's nice and easy ;-)
Why not
stuff it all in a CPLD? Maybe create a logic-only version too...
Well personally I prefer logic-only because I know I'll have the parts
here, plus I don't have any sort of programmer for making more exotic
things (ok, so I do have an ancient PAL/EPROM programmer, but without
manuals I've only ever figured out the EPROM side of it :) I still like
the idea of SRAM from old PC motherboards; didn't they
normally have 64 or even 128KBytes? Free then; with other parts costs
it'd probably make the whole thing do-able for a fiver or so (more with
a few frills like a case, PSU etc.)
I've got a TAG RAM here from a Pentium board and it's only 16k. I'm still
trying to find a few more of them... I did have a SCSI card in my RiscPC, but I replaced it
with a 100BaseT LAN
card, seeing as I wasn't using the SCSI (no SCSI devices to use with the
card).
I unearthed about ten Acorn SCSI cards the other day :-)
seeya
J.
4:13 p.m.
On Fri, Mar 11, 2005 at 05:16:20PM +0000, Jules Richardson wrote:
Well if worst-case is 1mbps data rate at 300rpm and 8x
oversample, isn't
that (8 * 10240 * 1024) / 5 = 1677722 bits of memory maximum needed?
<snip>
Can static RAM can't cope with these kinds of
speeds? Presumably cache
chips pulled from old PC motherboards would cope happily, even if old
80's parts wouldn't.
Easily and cheaply gotten SRAM should easily be able to cope with speeds
like 1 megabit a second or even 1 megabyte a second or even 10 megabytes
a second, or even more.
It's probably still beyond my abilities to design
something that'd
work :-) But on the surface it doesn't sound *that* complicated for
someone who's used to designing digital circuits.
But don't look at me. I just look at designs, I don't really make them.
--
Joshua D. Boyd
jdboyd at jdboyd.net
http://www.jdboyd.net/
http://www.joshuaboyd.org/
11 Mar
11 Mar
12:35 a.m.
Barry Watzman wrote:
What we need so badly is a USB universal floppy disk
controller.
I was just thinking about that the other day. I found this chip which
looks pretty nice:
http://www.smsc.com/main/catalog/usb97cfdc2_01.html
I'm desperate enough that I've been taking
apart 3.5" USB drives hoping to
find one that was a USB controller with a standard 3.5" drive. So far, they
have ALL been "integrated" devices.
You might have better luck with ones that have a "765B Floppy Disk
Controller", such as this:
http://www.cmicomputer.com/cgi-bin/quikstore.cgi?product=fd1.44u&detail…
-James Nugen
7068
days inactive
7069
days old
10 comments
6 participants
participants (6)
-
jdboyd＠jdboyd.net -
jnugen＠one.woovis.com -
julesrichardsonuk＠yahoo.co.uk -
philpem＠dsl.pipex.com -
pkoning＠equallogic.com -
Watzman＠neo.rr.com