PDA

View Full Version : FDC Pin 34 - READY Vs DISK CHANGE



IBM Portable PC
January 13th, 2014, 02:24 AM
The 8272 and 765 FDC's have a READY pin.

Other FDC chips like the Compaticard IV's 82077A have a DISK CHANGE pin.

This mean that unless your FDD has a DC/RDY or similar jumper, to switch pin 34 between these 2 incompatible signals (thanks IBM!), that you may have issues moving drives between controllers.

For example, I'm playing with a Gotek FDD emulator that works fine with an 8272, however it does not work with the 82077A. A JA jumper supposedly jumpers for READY on pin 34, however this does not appear to work, so I'll replace the CC IV with a 4 drive 765/8272 chip based card........

Chuck(G)
January 13th, 2014, 08:56 AM
The IBM standard design from the 5150 on have READY from the FDC (usually a D765 or 8272, though Zilog also sourced the chip) tied active. That is, a drive is never not "READY".

The whole rationale behind the READY signal in the 8272/765 is that it's an 8" controller, where the drive spindles are alway spinning. So READY really indicated that a disk was inserted in the drive. This was used by the "polling" mode of the FDC to periodically select each of 4 possible drives and generate an interrupt should the status change. To disable this, not only is READY tied active, but the 765 drive select outputs are no-connects, such that polling doesn't do anything.

So how does the PC determine that a drive isn't ready? Easy--by attaching a deadman timer to every operation. If an operation hasn't completed in about a second, the 765 can assumed to be hung, and a RESET signal is issued, and the FDC is respecified.

Note that the 5170 uses the 765/8272 but supports DISK CHANGED as well. This is done, like unit select and motor control, by external logic. Basically, a disk changed status is asserted by a drive when a disk is inserted or removed. The status is reset by moving the heads (a "step in" follwed by a "step out" is sufficient). The treatment of the status is up to the drive software--usually in the case of 360K and 720K drives, it's ignored--and for 1.2M and 1.44M, it's treated as an error and passed up to the calling program.

The 82077/8477 controller integrates the various functions of the 765 and supporting logic. In addition, it can support alternate PS/2 FDC modes. But in standard PC mode, the operation is the same as the discrete 765+glue version. If there's a problem with READY being asserted by a drive, it's the BIOS software, not the chip.

Dwight Elvey
January 13th, 2014, 10:47 AM
I have a computer that needs the DriveReady signal. Finding 3.5" drives that have
this signal is getting difficult.
I was wondering if anyone has made a circuit to recreate this signal. It seems
that most 3.5" drives will block the read data if the drive is not ready.
I'd think a simple oneshot chip could be used to work or even a diode and
filter capacitor might do the job. What to do with an unformatted disk becomes
a problem. These drives also block the index so maybe that might be a
better signal to monitor.
Dwight

Chuck(G)
January 13th, 2014, 11:02 AM
Yup, I've got one here that uses an 8-pin PIC (same as on the keyboard converter). It runs off the INDEX pulse of the drive. Source code attached.

However, for customers with CNC equipment that needs this signal, I simply supply jumpered Samsung 3.5, which are very configurable. In particular, some equipment requires READY on pin 34 and DISK CHANGED on pin 2, as well as IN USE on pin 4.

But, FWIW, here you go... The code commentary should be pretty self-explanatory.

You could also do the same with a couple of one-shots.

IBM Portable PC
January 13th, 2014, 01:03 PM
........... It seems
that most 3.5" drives will block the read data if the drive is not ready..........
Dwight

Just to clarify, DRIVE READY / DISK CHANGE are outputs from the FDD and of course conversely, inputs into the FDC chip.

So it's not the drive blocking a read, it's the controller, which has been told it can't read isn't it......

IBM Portable PC
January 13th, 2014, 01:09 PM
.....for customers with CNC equipment that needs this signal, I simply supply jumpered Samsung 3.5, which are very configurable. In particular, some equipment requires READY on pin 34 and DISK CHANGED on pin 2, as well as IN USE on 4...........

OMG!! As Steve Ciarcia says "my favorite language is solder". Perhaps just remove the FDC chip that wants a DISK CHANGE signal and build a little board to adapt an 8272/765 etc? Most FDC chips are very command compatible.......

Ref back to my original post, I'm pulling out the Compaticard IV and trying a couple of other 765/8272 cards............l

Chuck(G)
January 13th, 2014, 02:09 PM
Try the CC IV in a machine >286 with the BIOS disabled. I'll betcha it behaves much differently.

Dwight Elvey
January 13th, 2014, 04:35 PM
I'm sorry, I didn't mean to confuse things.
I was talking about the ready signal, pin 34, from the
disk drive.
That is an output of the drive.
Dwight

IBM Portable PC
January 13th, 2014, 11:54 PM
Here's a list of the few FDC chips and whether they expect/accept READY or DISK CHANGE:

READY on PIN 34 of the FDD interface:
(These were designed around the time the PC appeared i.e. they are in XT's and clones)
8272
NEC765
Z765A

DISK CHANGE on PIN 34 of the FDD interface:
(IBM wanted disk change detection and so most post XT chips don't expect READY on pin 34)
WD37C65C
I82077A
MCS3201
DP8473
UM8388

Chuck(G)
January 14th, 2014, 08:26 AM
I think you're missing the point. The second list of chips use a 765 "core", that is, they implement the 765 and add the extra logic present on the IBM 5170; some even acknowledge the NEC license on the chip labeling. You need only grab the 5170 floppy/hard disk controller schematics to see what's going on. As another example, consider the drive-select and motor-control logic as used in all PCs--it's not present in the first group. For all intents and purposes, the first group are the same chip--NEC and Intel had a very cozy relationship. The NEC floppy controller bearing the 8272 label was part of a cross-licensing agreement--I think Intel also licensed NEC's graphics chip at about the same time.

This is pretty much a case of integration, not implementation. The 765 dates from the 1970s; in particular, I seem to recall seeing the announcement around 1979. The second group dates from the 80s after the PC AT way of doing things had gained some ground. I particular, the first (that I'm aware of), the WD37C65 is in the WD 1984 storage products databook. It was an unusual product back then for WD--their own WD-developed line of floppy controllers (the WD17xx, 19xx and 27xx chips are based on a completely different design and date from the mid 1970s).