27 Nov
2019
27 Nov
'19
8:46 p.m.
Hi friends,
In the continuing saga of building a CP/M system with Pro-Log cards housed
in a Heathkit dual 8? floppy drive cabinet
I recently acquired a Pro-Log 7387 floppy disk controller card w/ manuals.
The 8272 chip on the card was fried (either before I got it, or by me doing
something stupid early on once I had acquired it), and so I replaced the
chip with another pulled from a 8-bit ISA floppy controller. Now I am
getting some more reasonable replies out of the card.
Now I'm stuck with "Missing Address Mark" errors, no matter if I'm using
a
real 8" floppy drive (8MHz FDC clock, appropriate settings in all the
software) or a GoTek floppy emulator (4MHz FDC clock, 720KB settings on the
GoTek & Floppy Disk Utility).
I can see the read data & read window lines working as I'd expect, so I'm
fairly confident that the read data separator and related logic is OK (I
even replaced the 74ls74 read data separator with no effect).
What am I missing that would cause a 8272 to always return Missing Address
Mark?
I'm using RomWBW, with the included FDU utility patched to talk to the FDC
at the correct addresses, and from the traces I've captured it appears that
side of things is working correct. I'm hoping someone out there has
experience troubleshooting floppy controllers?and can give me some pointers.
-David
------ Trace with GoTek running FlashFloppy and emulating a 720KB drive
follows -------
RetroBrew HBIOS v2.9.1-pre.5, 2019-11-23
PROLOG Z80 @ 3.686MHz
0 MEM W/S, 1 I/O W/S, INT MODE 1
512KB ROM, 2048KB RAM
SIO0: IO=0xF5 SIO MODE=9600,8,N,1
SIO1: IO=0xF7 SIO MODE=9600,8,N,1
MD: UNITS=2 ROMDISK=384KB RAMDISK=1920KB
FD: IO=0xC4 UNITS=2
Unit Device Type Capacity/Mode
---------- ---------- ---------------- --------------------
Disk 0 MD1: RAM Disk 1920KB,LBA
Disk 1 MD0: ROM Disk 384KB,LBA
Disk 2 FD0: Floppy Disk 3.5",DS/DD,CHS
Disk 3 FD1: Floppy Disk 3.5",DS/DD,CHS
Serial 0 SIO0: RS-232 9600,8,N,1
Serial 1 SIO1: RS-232 9600,8,N,1
PROLOG Z80 Boot Loader
Boot: (C)PM, (Z)System, (M)onitor,
(L)ist disks, or Disk Unit # ===> BOOT CPM FROM ROM
CBIOS v2.9.1-pre.5 [WBW]
Formatting RAMDISK...
Configuring Drives...
A:=MD1:0
B:=MD0:0
C:=FD0:0
D:=FD1:0
3623 Disk Buffer Bytes Free
CP/M-80 v2.2, 54.0K TPA
B> FDU
Floppy Disk Utility (FDU) v5.2, 08-Jan-2018 [HBIOS]
Copyright (C) 2017, Wayne Warthen, GNU GPL v3
SELECT FLOPPY DISK CONTROLLER:
(0) Exit
(1) Disk IO ECB Board
(2) Disk IO 3 ECB Board
(3) Zeta SBC Onboard FDC
(4) Zeta 2 SBC Onboard FDC
(5) Dual IDE ECB Board
(6) N8 Onboard FDC
(7) RC2014 SMC (SMB)
(8) ProLog 7387
=== OPTION ===> PL
===== PL ==============<< FDU MAIN MENU >>======================
(S)ETUP: UNIT=00 MEDIA=720KB MODE=POLL TRACE=00
----------------------------------------------------------------
(R)EAD (W)RITE (F)ORMAT (V)ERIFY
(I)NIT BUFFER (D)UMP BUFFER FDC (C)MDS E(X)IT
=== OPTION ===> FORMAT (T)RACK, (D)ISK ===> DISK
ENTER INTERLEAVE [01-09] (02):
RESET DRIVE...
PROGRESS: TRACK=4F HEAD=01 SECTOR=01
===== PL ==============<< FDU MAIN MENU >>======================
(S)ETUP: UNIT=00 MEDIA=720KB MODE=POLL TRACE=00
----------------------------------------------------------------
(R)EAD (W)RITE (F)ORMAT (V)ERIFY
(I)NIT BUFFER (D)UMP BUFFER FDC (C)MDS E(X)IT
=== OPTION ===> READ (S)ECTOR, (T)RACK, (D)ISK, (R)ANDOM ===> TRACK
ENTER TRACK [00-4F] (00):
ENTER HEAD [00-01] (00):
PROGRESS: TRACK=00 HEAD=00 SECTOR=01
READ: 46 00 00 00 01 02 09 2A FF --> 40 01 00 00 00 01 02 [MISSING ADDRESS
MARK]
CONTINUE? (A)BORT, (R)ETRY, (I)GNORE ===>
27 Nov
27 Nov
9:26 p.m.
On 11/27/19 8:46 PM, David Kuder via cctalk wrote:
Hi friends,
In the continuing saga of building a CP/M system with Pro-Log cards housed
in a Heathkit dual 8? floppy drive cabinet?
If you're talking about ST2, bit 0 being set as an error, this can occur
for several reasons.
A. You're trying to read an FM-encoded disk in MFM mode, or vice-versa.
B. The address headers on the track aren't what you expect (e.g. first
sector on the track is 63 instead of expected 1. Whatever formatted the
disk can put just about anything in the CHRN fields--and this is
sometimes used to separate different disk formats.
C. You're not on the right track.
D. If you're using an 8272 and *not* the 8272A, you could be running
into the extra-long "blind spot after index" of the older chip, meaning
that the first sector on the track may not be seen. Even on the
8272A/NEC 765A, there are some formats written without an IDAM by other
controllers, which causes the 8272A to miss the first sector.
That's right off the top of my head--to see exactly what you're looking
at, try doing a READ ID and see what you get back, if anything.
--Chuck
10:08 p.m.
I'll agree with Chuck, that FM/MFM seems a likely possibility.
index blindness can prevent seeing the first sector, but it should see
other sectors.
What is on the disk that you are trying to read?
You are using an 8" drive, but you mention settings for 720K.
720K (MFM) is normally 250K bts per second (300 K at 360 RPM in a 1.2M)
8" (MFM) is normally 500K bits per second.
8" FM is 250K bits per second.
Is the data transfer rate consistent with what you are trying to read?
Do you have a scratch disk that you can spare?
Set up the code to FORMAT a track (eventually appropriate loops to format
the whole disk).
After formatting a track, and without doing any stepping to another track,
try to read it. FM/MFM and data transfer rates and even sector
header (CHRN) should therefore be compatible.
If you have another machine, try reading THAT track on the other machine
and check what it sees.
28 Nov
28 Nov
7:06 a.m.
I've tried using the 8" drives as 8" drives with the clock input to the
8272 set correctly, and likewise with the 720KB 3.5" emulator (and even
360KB 5.25" emulation). On physical and virtual disks I'm getting the same
error trying to read or write data. Formatting steps through all the tracks
without error, and I can see a write status indicator on the GoTek
display...
Across all of the formats I've tried, they use MFM and not FM, and I can
see the MFM bit set in the command byte. I can try building FDU and the
BIOS to support FM encoded disks, and see if that works. I'd love to see
some kind of progress, even if it is single density.
Thanks for the help,
-David
On Thu, Nov 28, 2019, 1:08 AM Fred Cisin via cctalk <cctalk at classiccmp.org>
wrote:
I'll agree with Chuck, that FM/MFM seems a likely
possibility.
index blindness can prevent seeing the first sector, but it should see
other sectors.
What is on the disk that you are trying to read?
You are using an 8" drive, but you mention settings for 720K.
720K (MFM) is normally 250K bts per second (300 K at 360 RPM in a 1.2M)
8" (MFM) is normally 500K bits per second.
8" FM is 250K bits per second.
Is the data transfer rate consistent with what you are trying to read?
Do you have a scratch disk that you can spare?
Set up the code to FORMAT a track (eventually appropriate loops to format
the whole disk).
After formatting a track, and without doing any stepping to another track,
try to read it. FM/MFM and data transfer rates and even sector
header (CHRN) should therefore be compatible.
If you have another machine, try reading THAT track on the other machine
and check what it sees.
8:24 a.m.
On 11/28/19 7:06 AM, David Kuder via cctalk wrote:
I've tried using the 8" drives as 8"
drives with the clock input to the
8272 set correctly, and likewise with the 720KB 3.5" emulator (and even
360KB 5.25" emulation). On physical and virtual disks I'm getting the same
error trying to read or write data. Formatting steps through all the tracks
without error, and I can see a write status indicator on the GoTek
display...
You'd likely be better off with the 1.44MB or 1.2MB emulator, as 8"
disks used the 500KHz clock rate, not the 250KHz of the 360/730K.
FWIW,
CHuck
9:34 p.m.
After digging in the manual for the controller card, it appears they have set it up to
only support single density FM formatting. Even still, formatting the disk 568KB FM with
either clock speed 8MHz or 4MHz (and 500KHz / 250KHz WCLK respectively), same outcome with
missing address mark. I think I?m going to scrap the floppy card and design a fresh one
using the FDC37C78. Less things to go wrong.
-David
On Thu, Nov 28, 2019, 1:08 AM Fred Cisin via cctalk <cctalk at classiccmp.org
<mailto:cctalk at classiccmp.org> > wrote:
I'll agree with Chuck, that FM/MFM seems a likely possibility.
index blindness can prevent seeing the first sector, but it should see
other sectors.
What is on the disk that you are trying to read?
You are using an 8" drive, but you mention settings for 720K.
720K (MFM) is normally 250K bts per second (300 K at 360 RPM in a 1.2M)
8" (MFM) is normally 500K bits per second.
8" FM is 250K bits per second.
Is the data transfer rate consistent with what you are trying to read?
Do you have a scratch disk that you can spare?
Set up the code to FORMAT a track (eventually appropriate loops to format
the whole disk).
After formatting a track, and without doing any stepping to another track,
try to read it. FM/MFM and data transfer rates and even sector
header (CHRN) should therefore be compatible.
If you have another machine, try reading THAT track on the other machine
and check what it sees.
10:07 p.m.
On 11/28/19 9:34 PM, David Kuder via cctalk wrote:
After digging in the manual for the controller card,
it appears they have set it up to only support single density FM formatting. Even still,
formatting the disk 568KB FM with either clock speed 8MHz or 4MHz (and 500KHz / 250KHz
WCLK respectively), same outcome with missing address mark. I think I?m going to scrap
the floppy card and design a fresh one using the FDC37C78. Less things to go wrong.
Is there a manual for the card online?
If you're doing your own, I'd recommend a NSC DP8473--it has the benefit
of full FM and MFM support, including reading and writing 128 byte MFM
data. Not even the NSCPC8477 can do that--and you can forget the Intel
chips like the 82077/78 for FM.
--Chuck
1855
days inactive
1856
days old
6 comments
4 participants
participants (4)
-
cclist＠sydex.com -
cisin＠xenosoft.com -
dk＠thewaffleiron.net -
dkgrizzly＠thewaffleiron.net