[Coco] Dead (??) 5 1/4" drives

Gene Heskett gheskett at shentel.net
Wed Mar 8 12:18:42 EST 2017


On Wednesday 08 March 2017 08:57:15 Mathieu Bouchard wrote:

> Le 2017-03-08 à 08:44:00, Mathieu Bouchard a écrit :
> > Le 2017-03-06 à 23:50:00, Barry Nelson a écrit :
> >>> I have 2,  5-1/4 drives, of which one can go. ( keeping at least
> >>> one for myself).
> >>> I do think they are both DS 80 track drive's, and they where given
> >>> to me from e working setup.
> >>> If you are interested in one let me know.
> >>> I think that $20 + Shipping would be a reasonable price.
> >>
> >> DS 80 track? That would make those 720k drives.
> >
> > Nope, because 3.5" DS 80-track includes three different densities :
> > 720k, 1440k, 2880k. They are distinguished by number of sectors : 9,
> > 18, 36 (with PC formatting @ 512-byte/sector).
>
> Woops, sorry for the dumb mistake about disk size. 5.25" DS 80-track
> include two different densities, QD and HD. Almost all those disks are
> HD, formatted at 1200k (15 PC sectors). QD disks are listed as 640k (8
> PC sectors) on Wikipedia, not 720k (9 PC sectors). I don't know why. I
> have never seen any. I can only say that 8-sector formatting was the
> norm in MSDOS 1.x, so it seems to me like a format that was dropped
> very early. Using the "physical composition" section of the article,
> it seems like it could be doing 720k if using a formatting analogous
> to the 360k (72% of unformatted size), or 800k if formatted like a
> 1200k (80% of unformatted size).
>
> https://en.wikipedia.org/wiki/List_of_floppy_disk_formats
>
The wiki would be wrong. In 5.25 disks, DD meant its was surface 
formulated for a 250 kilobyte data rate, which the coco's controllers 
do.

HD meant a 500 kilobaud data rate, and QD was an extremely fine grained 
coating formula, intended to be used on 1 megabaud data rate 
controllers. The 1200's were an early attempt at a 1440, on a 5.25" 
media at a 500 kilobaud data rate but with the disk speed turned up from 
300 RPM to 360 RPM because that increase in disk surface velocity got 
the playback signal up out of the noise floor far enough to be usable.

By the time head technology had progressed far enough to make the QD disk 
practical, giving 2880k, hard drives of decent sizes had become 
commodity items, and floppies slowly faded from the scene as CD R/W 
became competitive, replacing the mechanical precision track location 
needed for a floppy, with read it to see where you are and correct it 
electronic servo motors. Precision level in the angstrom range is much 
cheaper to do in smart sand. And several decimal places faster when the 
motor resembles a voice coil from a speaker. The modern blue-ray DVD r/w 
drive is the logical end point, increased capacity methods left to use 
are limited by the practical range of the laser diodes. Its possible to 
up-pump the blue ray into the UV, but while that would give a finer spot 
for more tracks, the up-pumping into UV range is presently limited in 
modulation rates by the fluorescent dyes available with limited rise and 
fall times so theres no huge net gain in capacity.  So a dual layer 
blue-ray may be the optical disks last hurrah. Already we can get more 
capacity in a USB key, and that price will continue to drop as time to 
develop layer stacking of the storage bit cell continues apace.

Technology marches on. If we can't keep up, well...  And we aren't, in 
some category's, even trying to keep up. Clock speed for instance. 
Technology today, could easily give us a 3GHz clocked 6309.  But the 
raspberry pi 3b could still outrun it with its 4 cores running at 1.2 
GHz. Context switching time is a pita, but I only know of 2 
architectures of cpu's that whupped that, the RCA 1802, and the TI-9900 
family, both of those could do the task switch by simply reloading the 
program counter in the 1802, or have it preloaded in another of its 16 
bit registers by simply switching the pc counter to another register.  
The 9900 had no concept of a program counter as that was all kept in 
memory, and the only internal register it had could be reloaded with the 
address in memory where another register image set lived. But both were 
scheduled, not interrupt driven AIUI. TI did have quick irq responses, 
so maybe I didn't get it all on the TI-9900's. The entry price to move 
beyond the 99/4A with the expansion box was at that time way too pricey 
for this old Iowa farm kid.  But I still have that 99/4A.

My favorite machine control software, LinuxCNC, is the exact same code 
that runs on an X86 boxen because its written in a scrambled eggs 
version of python and tcl/tk mixed.  Those languages have been fully 
translated, so I am running, on the arm based pi, the exact same code 
that runs here, or on 3 of the other 4 machines out there in the 
coyote.den's domain.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>


More information about the Coco mailing list