Date: Thursday, 31 January 1985
From: Robert Bloom
To: NorthStar users
Re: How to install the TEAC 55F's in a Horizon
SUBJECT: How to install the half-height 96tpi TEAC FD55f disk drives in a
NorthStar Horizon, with some side comments about installation of some BDOS and
BIOS patches and ZCPR2. Installation includes both the hardware and software
steps necessary to use two of the TEAC half-heights with one standard full
height drive (SHUGART [SSDD] or TANDON [DSDD - [quad']) at minimum cost.
Software patches include 'Archive' directory bit (t3) reset on file
modification, Plouffe's modifications to handle some non-standard but very
usefull disk formats (fully compatable with all standard formats), zpcr2
buffer area initiation on cold boots, and Rich Conn's zcpr2 system itself.
This report contains all the the actions that I found necessary to get the
drives on-line and should help any that are thinking about doing themselves.
Although I managed the actual swap by myself, contributions by Frank Wancho
and Al Plehn were invaluable.
This entire exercise was started by a message from Al Plehn on the virtues of
the TEAC FD55F 96-tpi drives. I had been thinking of replacing the SSDD with
the now NorthStar-standard QUAD (TANDON 100-2 DSDD) drives but had been
stopped by cost. Then I saw the ad in Byte in which California Digital had
the TEACs for only $139 (1), $135 (2-9), $129 (10+); this was even cheaper
than the TANDONs and had twice the capacity.
I ordered two on 14 Dec by their 800 number with a MasterCard number.
Received the acknowledgement copy via US Snail on 18 Dec and the actual drives
on 28 Dec. That's a two week delivery in the middle of the holiday season -
not bad at all.
The TEAC FD55F characteristics are, briefly: half-height, 96 tracks per inch,
80 tracks per side. Disks secured with a twist lever, not a swinging door.
Relative to TANDON (or SHUGART) drives, the TEAC drive controller edge
connector is slightly higher and upside down and the TEAC drives are slightly
deeper. They are hardware compatible with the standard NorthStar disk
controller, and software compatible with NorthStar's CP/M 2.2 Rev. 1.1.0
'CPMGEN'. CP/M formated capacity shows 784k free space with the standard
NorthStar FORMAT.COM formatter with a 4k block size. Plouffe's mod yeilds
786k free space with a 2k block size. The drives came with a technical manual
(~25 copied pages) which I had asked for (no cost) and two power connectors
with dettached plugs but no cables. Each drive had two 3mm and #6 mounting
holes top and bottom, the 3mm holes being in the preferred positions, the #6
holes are the same size as the old drives had.
Since I bought 2 of the TEACs, I decided to keep one of the old SHUGART SSDD
drives in the Horizon so that I still could write 48-tpi formats. Because of
the greater depth of the TEACs, they will only fit in the right side drive
position. (A power capacitor gets in the way on the left side drive - Al
Plehn suggests moving all the offending parts of the power supply back an inch
by drilling a few holes. The two new drives just barely fit in my standard
configuration.) As the disk controller cable has only two connectors, I
initially worked on getting just the two TEACs on-line - the old SSDD drive
was be just a dummy. Later in this message I tell how I got the SSDD drive
on-line so that all three can worked together.
Interesting note: the manual gives the mounting configuration as lever up if
mounted vertically, or the in-use light up if horizontally. (If mounted
horizontally, having the disk drive motor up or in-use light down is called a
no-no.) The vertical position as stated puts the R/O notch DOWN, opposite to
what I'm used to. This would put the edge connector right-side up, at least
compared to the TANDONs. As I have my Horizon vertically oriented on it's
right edge (drives on top), I really have the drives horizontal as suggested.
This means when I place the Horizon horizontal, the drives are now vertical
and upside down as regards the instruction and edge-card connector, but right-
side up as regards my 'instincts'.
Actually getting the drives to read and write disks was a bit more complicated
than I thought, but straight forward once you realize that the TEACs can't
write a 48-tpi disk or the 48-tpi drives a 96-tpi disk and CP/M has to be told
about the 96-tpi drives before using them. What follows is the steps I took
written as instructions. After switching to the two new drives one can still
read all the old disks, but can only reformat them to 96-tpi, not do any other
write operations on the 48-tpi disks. When one gets a longer controller cable
(or cross fingers and strech the old), you can put the old 48-tpi drive back
on-line and then write 48-tpi formats too.
Before starting, inspect the drives for obvious problems. Program the drives
via the mini-jumps for: one drive as '1' (to be used as the A: drive), and
the other drive as '2' (to be used as the b: drive) and remove the terminating
resistor strip from drive 2 (not 1! - when you're finished drive 1 will be on
the end of the controller cable). I also removed both 'PM' jumpers and left
the others as them came. The manual does explain all of the options - my
problem was understanding the explanation!
1. With power off, remove the right-hand side old drive and temporarily
plug in a new drive (B) to the now-open power and controller cables. (The new
drive is 'upside down' with R/O notch at the bottom.) It need not be secured
in place as it will be moved later.
2. Power up and place a copy of the NorthStar CP/M distribution disk in
the A: (old) drive and a blank disk in the B: (new) drive. Boot your old
working version.
3. Run CPMGEN from the A: drive and create a new vanilla CP/M system
with drive one as a SSDD (D) or QUAD (Q) drive and drive two as a nine (N)
drive [fast stepping]. Place the new system on the A: drive disk and cold
boot.
4. Format the disk in drive two as a 'Nine' disk. Do this by running
FORMAT.COM and selecting either 'N' or '9' - it is not on the menu and you
might need a one byte patch. (See the NorthStar-users message archive if you
are not familiar with this patch - change '9'<39h> to 'N'<4Eh> at 0176h.) Or
alternately, and probably preferred, use NEWFMT.COM and format as 'N' - (*NOT*
'O' for octal with special 2k blocks!) NEWFMT.COM can be found in SIMTEL20 at
MICRO:<CPM.NSTAR>NEWFMT.14COM which you should get anyway as it is needed
later if you want Plouffe's modifications.
5. Run CPMGEN again and create a new system with both drives one and two
as N drives, and place it on the newly formatted disk in drive two (TEAC).
Mark this disk as a 'N' format boot disk and power down.
6. Disconnect the power and controller cables from both drives. Place
both both new drives in the right-hand slot in the correct orientation, i.e.
R/O notch and in-use lights up. I left the old (disconnected) left hand drive
in place. Before securing the drives to the chassis, attach power and
controller cables. I found it easist to run the controller cable to the
right-most drive first, and then to the middle drive. Note that the
connectors are now upside down relative to what it was. The controller cable
is pressed directly between the drives and a power connect strip with no
clearance. (I've only 3 cards in the buss and the supply runs cool - people
with hot supplies should take Plehn's advice above or risk a melted cable.)
Remember too that now drive #1 is on the end of the cable and should have the
terminating resistor, not drive #2!
7. Secure drives. The middle drive (which is the 1 or A drive) can be
secured directly from underneath from the same holes as the old drive came
from. I found the 3mm holes in perfect position - if you don't worry about
buggering up the threads, a #4 screw works (or *gasp* - a short self-tapping
screw), otherwise either use a correct 3mm screw or drill and tap the
baseplate for the #6 holes. There is no easy way to secure the right-most
drive - either drill the baseplate for another hole, fabricate a bracket to go
across the top of the drives (my favorite), or just let flop. The latter is
not as bad as it sounds, especially if the Horizon will be used vertically.
(One possible method of making a bracket: place all three drive in the
chassis where they belong. Place a piece of paper over the drives where you
want the bracket to go and punch holes in the paper where the screw hole are.
I used all six of the #6 holes on the top of the three drives. The bracket
will be nothing more than a flat piece of metal. Transfer punched holes in
the paper to a piece of aluminum and drill - and voila! One custom-made
NorthStar/TANDON/TEAC disk drive bracket!)
8. Power-up and boot with the 'N' drive boot disk. One can still read
all the old 48-tpi disks but a write request to any of these will cause a
"ILLEGAL ACCESS" and require a cold-boot to regain control.
That's it! What follows is the patches to the system for ZCPR2, and and
Plouffe's mods. Filename are from SIMTEL20 in <CPM.*> directories:
In MICRO:<CPM.NSTAR> you need:
FORMAT.DOC - if you want to know what it's all about
NEWCPY.16COM - replaces the NorthStar COPY.COM & handles the new formats
NEWFMT.14COM - replaces the NorthStar FORMAT.COM with new formats
NEWFRM.14ASM - the actual patches
NSGEN.COM - replaces the NorthStar SYSGEN.COM & handles the new formats
In MICRO:<CPM.ZCPR2> you need:
ZCPR2.ASM - the actual CCP replacement (need not change)
ZCPRHDR.LIB - all the the equates and addresses to change
(that's all you *need*, but there's lots of utilities <*.COM>, documentation
<*.WQ> and help <*.HLP> files in this directory too.)
In MICRO:<CPM.DSKUTL> you might also want:
ARCHIVE.COM - will backup all files with the archive bit (t3) unset and
reset it. A bdos patch will cause the archive bit to be unset whenever a file
is modified.
All of the following assumes a '64k' NorthStar CP/M system (actually 61k with
a split BIOS) and CPMGEN has been patched for the spurious $R/O BDOS error on
the A: drive. (The patch is for CPMGEN.COM and changes ANI 02 and JNZ 7992 at
2F31H to MVI A,99 and STA 7FAB.)
1. If ZCPR2 is wanted (recommended), edit ZCPRHDR.LIB for the following
addresses:
2. If you want the special disk formats (very recommended), edit
NEWFRM.ASM for the correct BIOS (F300) and CCP (D100) locations, set OCTAL
TRUE and QUAD FALSE. (I've not been able to get QUAD to work yet - I think
it's something to do with the split BIOS.)
; NEWFORM VERSION 1.4 *
; 12/02/82 *
; R. L. Plouffe *
3. Providing ZCPR2 is desired, one has to initialize certain buffers at
cold boot time. Create a new USER.ASM, call it ZUSR, by adding the following
code to the standard NorthStar-supplied user area code to initialize the ZCPR2
system on a cold boot: (This Z80 code is from Frank Wancho and requires
M80/L80 to assemble, convert to 8080 if necessary.)
aseg
z80
phase 0FA00H ;ORIGIN FOR 64KQD SYSTEM
mcmdbuf equ 0EC00h ; ZCPR2 Multi Command-line (MLC) Buffer
--- normal user area code (comment out serial printer code if parallel printer
--- and vis versa if you need more room.)
;WRAP IT UP BY SENDING A CR TO THE PRINTER & EXIT
;
CALL MULTI ; ZCPR2 Initialization
LD C,0DH ;A CARRIAGE RETURN
JP USERBAS-700H+15 ;GOTO PRINT JUMP IN BIOS VECTOR
;
;******* END OF STANDARD INITIALIZATION CODE *********
;
;**** ZCPR2 INITS ***********
ds userbas+130h-$ ; This fills to userbas+130h
;
; External PATH is here:
;
pinit: db '$','$' ; current disk/current user
; db '$',0 ; current disk/user 0
; db 1,'$' ; A:/current user
db 1,0 ; A0:
; db 1,10 ; A10: for finding NAMES.DIR
db 0 ; end of list
;
; Rest of space is available for PATH info up to userbas+200h
; So we use this space below for the one-time init of the MLC
; only done at cold boot
;
multi: ld de,mcmdbuf
ld hl,clinit
ld bc,5
ldir
ret
clinit: dw mcmdbuf+4
db 200
db 0
ds userbas+180h-$
end
4. If you would like the BDOS 'archive patch' to reset the archive bit
(t3) whenever a file is modified, you need to assemble the following code:
bdos$loc equ 0D900H ; base address of BDOS
wrsec equ bdos$loc+03b8h ; address of write sector routine
pntdir equ bdos$loc+055eh ; address of directory pointer routine
reset$archive equ bdos$loc+0df0h ; address of reset archive bit patch
;
ccp$base equ 1500h ; sysgen ccp base position
bdos$entry equ ccp$base+0806h ; sysgen bdos entry position
wrt$dir equ bdos$entry+05c8h; 'wrsec' call location, in DIR write routine
scratch equ bdos$entry+0deah; scratch RAM inside bdos for patch
org wrt$dir ; patch 'call wrsec'
;
call reset$archive ; call patch in scratch RAM
;
org scratch ; patch area for 'archive bit reset'
;
call pntdir ; point to directory entry in buffer
lxi d,11 ; make offset to 't3' in FCB
dad d
mov a,m ; get 't3' character from FCB
ani 07fh ; kill archive bit position
mov m,a ; return reset archive bit to FCB
call wrsec ; write directory sector
ret ; return to 'wrt$dir' routine
;
end
5. Now assemble the parts: (In the command lines below, 'x' should be
replaced with a drive letter indicating where the source is <first x> and
where the HEX file <second x> should be placed.)
assemble the archive overlay to BDOS:
ASM ARCPATCH.xxZ
assemble the user area overlay with ZCPR2 inits
M80 =ZUSR/M
L80 ZUSR,ZUSR.OVR/N/E
Plouffe's mod's for 2k block size on 96-tpi drives
ASM NEWFRM.xxZ
Finally assemble ZCPR2 proper
MAC ZCPR $Ax Hx PZ SZ
6. Actual assembly and patch: The order of overlaying the various
items is important and went as follows to create NEWCPM.COM (a SYSGEN-like
program containing the modified SYSGEN image):
NCPMGEN <-- REMEMBER TO USE PATCHED VERSION!
ETC ... CREATE A VANILLA SYSTEM OF SIZE REQUIRED
SAVE VANILLA SYSTEM ON DRIVE 1
NSGEN <-- TO READ SYSTEM INTO MEMORY (USE THE NEW
SYSGEN-LIKE PROGRAM!
1 <-- FROM DRIVE 1
^C <-- EXIT WITHOUT WRITTING TO ANOTHER DISK
SAVE 51 SAVECPM.COM <-- SAVE THE SYSTEM INTO A NORMAL FILE
DDT SAVECPM.COM <-- PUT IT INTO DDT
IARCPATCH.HEX <-- ADD THE ARCHIVE BDOS PATCH
R <-- NO OFFSET NEEDED, SYSGEN IMAGE ADDRESS
HARD CODED IN
IZUSR.OVR <-- AND THE USER AREA
R3100 <-- OFFSET NEEDED BECAUSE ZUSR.OVR IS NOT A
HEX FILE, SYSGEN IMAGE ADDRESS IS 3200H,
LOAD ADDRESS 0100H SO USE 3200-0100 = 3100
INEWFRM.HEX <-- SPECIAL MODS FOR QUAD/NINE DRIVES
R3800 <-- SYSGEN IMAGE @3200 - SYSTEM LOCATION @FA00
GIVES OFFSET OF 3800
F1500,1CFF,00 <-- CLEAR THE OLD CCP OUT
IZCPR.HEX <-- ADD IN THE ZCPR2
R4400 <-- SYSGEN IMAGE @1500 - SYSTEM LOCATION @D100
GIVES OFFSET OF 4400
^C <-- EXIT DDT
SAVE 51 NEWCPM.COM <-- AND SAVE THE PATCHED SYSGEN IMAGE
NEWCPM <-- RUN IT
<CR> <-- SYSTEM IN MEMORY SO NO DISK READ
1 <-- WRITE IT BACK OUT TO DRIVE 1
<CR> <-- COLD BOOT TO BRING IN NEW
7. Now with the new system in place, one can format disks with the
'O'ctal format of NEWFMT which gives a 2k block size. HOWEVER: beware using
the octal format with any CP/M without the NEWFRM patch - sometimes
it'll work
sometimes it won't, and you might not know the difference intil it's too late.
Also, never try to copy 'O' format disks with the old NorthStar COPY.COM or
use the NorthStar SYSGEN.COM - use NEWCPY.COM and NSGEN.COM instead.
Try it, you'll like it!
That was the easy part - I did have some difficulty with the following. I
don't recommend the following procedure (it would better to buy or make a new
cable) but if you have confidence in yourself and a restricted pocketbook you
can try the following. (The controller cable just barely fits in this
procedure - I would recommend buying 3 drive connectors, one 34-pin connector
and at least 24 inch of 34-wire ribbon cable and make your own. In general
you can follow the procedure below but fit the cable as you go - it's lots
easier to wind it 'over around and through' the power supply if you have
plenty to spare.)
1. Buy one or two 34-pin edge card connectors from somebody. (I got
mine from Radio Shack.) Make sure that they are 'through' connectors, not the
cable termination type.
2. Take everything apart again, i.e. remove all three disk drives and
take the disk controller cable off of the controller card. My cable was 17.5
inches from controller connector to the end drive connector.
3. *VERY CAREFULLY* remove the middle drive connector from the cable.
(This might not be possible without damaging the connector.) One will want to
reuse the plug so don't break any pins. (I found myself needing to solder two
of the little buggers back together after they broke. Sort of like trying to
solder two wires together end-to-end WITHOUT overlapping them. Not the job
one could do without *a lot* of patience, or substantial soldering skill. So
either don't break it or buy the extra edge connector. My repaired connector
failed a couple of days later, and had to be replaced anyway.) In any case,
clearances are so tight the middle connector has to be removed one way or the
other.
4. Crimp on the removed (if salvaged) or a new connector 1-3/4 inches
from the end drive connector. This moved connector and the end connector will
connect to the two TEAC drives. I found it easist to place the connectors in
a vise to crimp them on.
5. Look at steps 7 and 8 below and make a test fit of the cable on the
computer. After you are familiar with how the cable has to twist and turn,
crimp the new connector approximately 8 inches from the CONTROLLER end of the
cable and facing the same direction as the other drive connectors. You should
mark the exact location during the test fit.
6. This will leave the cable with the controller connector on one end, a
drive connector on the other, and two drive connectors in between, all facing
the same way:
NOTE: IF YOU ARE MAKING UP A BRAND NEW CABLE DO NOT USE THESE DIMENSIONS
- THIS IS A MINIMUN LENGTH CABLE. With the luxury of cutting and fitting to
length, one would probably want to add several inches before and after the
connector to the old drive.
7. Place disk controller in the 4th or 5th buss slot from the rear or
whatever is closest to the back of the disk drives. Attach cable to the
controller and run the cable directly down to the SSDD drive, you will have to
take out the old twists in order to get it into position. It is easier to do
if the drive is not attached.
8. Make two 45 degree turns with the free end of the cable (between the
first and second drive connectors) to flip it upside down, and attach the
second connector to the third drive (TEAC). Finally, connect the end
connector onto the middle drive (TEAC). This is practically impossible to do
with the drives connected to the chassis so leave them all loose.
9. Solder or crimp one end of 4 short (4 to 6 inch) wires from your
spare parts box to 4 supplied power connector pins. It will be easier if you
use one black, one white, one green and one red wire to match the current
colors in the plugs. Using one of the drive power supply cables as a color
guide, push the proper color-coded pins into the supplied power connector
housing. Carefully remove the four pins from one of the drive power supply
cables and solder on the free ends of the new connector cable, matching colors
appropriately. Push the pins back into the housing, using the other cable as
a color guide. NOTE: the connectors are keyed to their jacks to prevent
misattachment. Therefore it is important that you install the pins in the
housing correctly.
10. Attach the old, unaltered power cable to the old drive and the new,
split power cable to the two new drives.
11. Double check that the power connections are right, and the
controller cable is in the correct orientation. Ignore the markings on the
cable connectors themselves (the connectors can and are often attached upside
down), refer back to the controller to find which edge of the cable have the
low numbers (pins 2 and 34 are marked on my controller card). Follow that
edge to the three drives and make sure that the pin markings on the drive
(ciruit card) match up.
12. Providing everything is correct, cross fingers and power up.
13. Check out the three drives to make sure that all will read and
write. Note that drive C: (the old one) will only read and write 48-tpi
formats; drives A: and B: will read 48-tpi and read and write 96-tpi. If need
be, generate a new system with the three drives identified appropriately.
14. Finally, attach drives to chassis with the screws underneath or your
homemade brackets. Make sure that the cable connectors stay in place when the
drvies are pushed in, the old drive connector is especially difficult if it
does not fit tightly.
bob bloom
p.s. After five days of use, one of the new TEAC drives died. The motor no
longer started either with in parallel with the others or upon a drive select
signal although the in-use indicator did light. California Digital was very
cordial, gave me a return authorization number and said to send it back. I
sent it back on Jan 9, and I'll report the results when something more
happens. For the time being, I'm running one old SSDD and one of the TEACs.
Coming attractions:
I'm currently working on getting ZCPR3 working on the new drives. All of
the ZCPR3 buffers fit very nicely above the disk controller address space,
leaving continuous memory for a 58k CP/M.
I also am trying to do the same modifications to a 18Mbyte hard disk
system, which is complicated because of the larger BIOS and multi-user
operating system (TSS/C).
