Early 3.5" Floppy Drives

[Chuck Guzis]

On 12/15/2005 at 1:18 PM Allison wrote:

>;) your assumption is double density.  8" SSSD is not that fast.
>I never said formats were the same or even dive interface only that
>the data rates fly.

Nope.  I'm just going by the 765 data sheet:

"Pin 19 - CLK - Single-phase 8 MHz (or 4 MHz for mini-floppies) squarewave
clock"

IOW, if you supported 3.5  DD (or SD) floppies, you weren't going to be
able to do an A1 8" floppy without changing the clock.

Same thing obtains for the WD 179x - "Pin 24 - CLOCK - This input requires
a free-running 50% duty cycle square wave clock for internal timing
reference.  2 MHz +/- 1% for 8" drives, 1 MHz +/- 1% for mini-floppies."

The 279x has a clock divider that's programmed by pin 17 (5/8).

Look at the grandaddy of the 8" XT drive controllers, the CompatiCard I.
It uses port 7F2H to change the 765 clock from 4 to 8 MHz for 8" support.
MFM/FM doesn't enter into the equation--that's programmed into the read and
write commands and the data separator.

But an old XT-era 720K 3.5"/5.25" controller couldn't do this
clock-switching trick unless it could also support 1.2/1.44 media.

Cheers,
Chuck
 















>
>765A write clock rate by drive and density, bit rate is clock/2.
>
>Size   density  format   writeclock
>-----------------------------------
>8"       DD      MFM     1000khz
>8"       SD       FM      500khz  (8"SSSD 241k CP/M standard)
>
>5.25     DD      MFM     500khz (40track is 360k, 80track 720k)
>5.25     SD       FM     250khz
>
>3.5"     HD      MFM    1000khz (1.44mb) (looks like 8" different CHS)
>3.5"     DD      MFM     500khz (720k)   (same rate as 5.25 DD and 8" SD)
>3.5"     ??       FM     250khz (not used obsolete)  
>
>None of this has anything to do with rotation rate of the media.
>Actual data storage capability is format dependent.
>
>One example that was known the to CP/M world was 5.25" 80track (FD55F)
>two sided at either 720k or ~780k  I was sometimes called QD as it
>was really the same as the 360k but twice the tracks (48 tpi vs 96).
>So happens that the 3.5" drive can be plugged in and used exactly
>as if it were a FD55F for the same 720k as I do it all the time
>from a CP/M system to DOS and the CPM80 side has a utility that 
>read/writes DOS FAT files. I'd have used 1.44 but the WD1770 literally
>cannot run at the required rate (not rated to either!).
>
>I'll let you all in on a dirty trick.  The 765(A) outputs a signal on 
>pin26 called FM, that is used to select data rates /2 ALWAYS.  If you lift
>the pin the data rates for FM mode are now twice as fast and suitable
>for many other uses like 8" media.  For the integrated flavors of 765
>the same effect can be had by twiddling bits in the drive control
register.
>If all else fails, you can double the the 8 or 16 mhz clock source
>used to 16/32 as needed.  I have taked the 9.6mhz out and used higher
>on one board 16mhz so that switching to AT 5.25HD got me 8"DD instead
>without futzing with drivespeed (rotation rate) that means nothing to 
>most 3.5, 5.25 (including FD55GFR with the jumper pulled) and 8" drive.
>
>>That was the beauty of the NEC APC line--from 8" right down to 3.5", the
>>data format didn't vary one iota.  The NEC 9801 floppies still record
>1.3MB
>>on a 3.5" drive spinning at 360 RPM.
>>
>>But the  PC-XT 8" drive controllers were a special beast, honest.
>
>Not really.  I can take the stock IBM XT long board and with one change
>make it do DSDD 8" (other than cable adaptor).  Common parts cost 'bout 
>$1, acutally cheaper now than 20 years ago then it would have been 1.89.
>Replace the 8mhz clock source with 16mhz.
>
>Thats how it's done.
>
>Allison