thinking of the "ultimate" retro x86 PCs - what bits to seek/keep ?

[Fred Cisin]

>>> And there was never a 720kB IBM standard, only on things like Apricots.
>> Could you explain?
>> 720K was just as much of an IBM "standard" as 360K.  Starting with PC-DOS
>> 3.20, ("/F:2").  Sure, other companies used the same media for different
>> formats, but nowhere near as much as 5.25".
>> "Convertible", PS/2, optional in and external to AT.

On Mon, 6 Jun 2016, Liam Proven wrote:
> I thought you might chime in on this.
> AIUI -- not sure -- the 720 kB *format* was mostly used on 5.25" DS/DD
> 80-track disks, no?

It was used on 5.25" DSDD 80 track, by MANY companies other than IBM at 
640K to 800K, and marketing people called it "quad density". 
Superbrain/Intertec called it "Super Density", and abbreviated it "SD"!! 
(therefore a Superbrain DSSD meant 800K)

IBM did not go there much.  They did use that on the PCJX, but not on 
anything in USA.

BUT, the 720K 3.5" was heavily used.
MS/PC-DOS 3.20 ushered in the official acceptance of 3.5" 720K.
(A few MS-DOS OEMs had had 3.5" already, such as Gavilan, DG-1, Toshiba)
That was the only drive choice for the "Convertible", IBM's not 
especially successful first attempt at a laptop, and the PS/2 model 25 
and 30.  (Didn't some of those use a cut-down model M keyboard w/o a 
numeric pad?)
IBM sold external 3.5" drives to add to PS/2, 5150, 5160, 5170, but they 
were not very popular - kinda expensive.   Some PS/2 owners added 
external 5.25", but most PS/2 owners didn't need an additional 3.5" drive.
IBM sold a 3.5" 720K drive with oversize faceplate and cradle to fit 5170.

With the PS/2-286, IBM added 1.4M with PC-DOS 3.30.
The 80286 and above PS/2s were available with either 720K or 1.4M drives, 
but with no significant price differential, nobody but the most ignorant 
college administrators (still in trauma from 360K/1.2M) would pick the 
720K.

Lack of a media detector in IBM's PS/2 1.4M drives meant that many 
users unintentionally, or without caring, formatted 720K disks to 1.4M.
The coervicity was close enough (600? V 720? Oersted) that it was believed 
by most that that was perfectly acceptable.

We all think that 2.8M support came sometime in DOS4.00 or 4.01, but 
Wikipedia level of non-authorities say 5.00.

> Apricot and other vendors shipped MS-DOS machines with only 3.5"
> DS/DD/80t drives. The same format was the standard diskette type of
> the Atari ST, too.
and early Amiga (#1000, etc.)

> But AFAIK IBM never shipped machines with DS/DD/80t track drives as
> standard, did it? *Possibly* as a 2nd drive, but not as the stock. I
> thought the IBM PC-compatibles came with:
> #1 DS/DD/40t, 5.25", 360 kB (PC, PC-XT etc., 1981)
(further subdivided with same FDC as SS, DS, 8spt, 9spt)
> #2 DS/HD/80t, 5.25", 1.2MB (PC-AT, 1984)
> #3 DS/HD/80t, 3.5", 1.4MB (PS/2, 1987)

That should be
#0 DS/DD/40t, 5.25", 360 kB (PC, PC-XT etc., 1981)
(further subdivided with same FDC as SS, DS, 8spt, 9spt)
#1 DS/HD/80t, 5.25", 1.2MB (PC-AT, 1984)
#2 DS/DD/80t, 3.5", 720K (Convertible,8086 PS/2, 1986)
#7 DS/HD/80t, 3.5", 1.4M (PS/2, 1987)
#9 DS/ED/80t, 3.5", 2.8M  (barely ever in PS/2)
(DRIVPARM/DRIVER.SYS numbering)

(#3 and #4 were 8" until about 6.00, but MICROS~1 doesn't like to talk 
about 8")
#5 was hard disk
#6 was tape
#8 was optical disk


>> If a good 150RPM ED drive were to have been readily available, then 2.8M
>> could have been retrofitted to all 1.4M systems, including Amiga, etc. But
>> would that, and the Barium-Ferrite disks have been worth it for just twice
>> the capacity?

150RPM drives (and 180RPM 1.2M (Weltec)) existed (Amiga), but were never 
any good.  They never got that kludge to work well enough.
"A kludge born of desperation"

> Wouldn't they have got much cheaper if every cloner had used them?

of course

> The cloners copied the 1.2MB and 1.4MB drives, but not the 2.8 ones.
> Was media expense the main problem?
yes   Barium-ferrite disks never got into sufficiently massive production.

> Or expensive FDC chips? Or both?
NOT expensive, nor chips.  FDC board/support circuitry.
The 765 or equivalent FDC CHIP had no problem with it.

A 5150/5160 FDC board runs at 250K bits per second data transfer rate
(THAT is Kilo, not Kibi)
A 5170 FDC board needs to handle
250K bits per second for 360K formats,
500K bits per second for 1.2M, and
300K bits per second for 360K disk in 1.2M drive.
Later, two speed drives came out, eliminating further need for that speed, 
but it was already in the designs.

A 2.8M FDC circuit needs to also support 1000K bits per second data 
transfer rate.


Therefore, adding a 2.8M drive would require replacing the FDC board, 
Existing inventories of HD/FDC cards would all need to be jettisoned.  And 
motherboards were just starting to come out that had FDC support on the 
board.

Yes, 2.8M was too little, too late.

--
Grumpy Ol' Fred     		cisin at xenosoft.com