3 May
2008
3 May
'08
7:29 p.m.
Thanks for your work on this Steve, I need to find the
one or two I
spotted again, but yours was cleaner than mine.
I tested the PCI NI-Analyser card again on the HP150 - which is the
working HP-IB machine I have. Since I didn't know better I captured
everything again - including timings. For someone writing an emulator,
this could be helpful, but I don't know if you want another huge capture
file of boring Amigo protocol stuff. I've attached what I did and
captures of the analyser interface - I'd welcome your advice on what I
should be capturing that would be useful for protocol understanding.
Hope all's well and thanks again,
Rodney Brown
Hey Rodney,
Hope you don't mind but, I'm gonna post this to the classiccmp list.
Someone there may want to follow the discussion.
What kind of system are you running the PCI Analyzer on? Windows? Linux?
I have FINALLY made some more progress with the ISA GPIB card in my
Linux box. Unfortunately, it uses the NEC7210 chip which has been a pain
in the rear to program. I found a few snippets of code online but really
didn't find anything that directly addressed my needs.
I have been able to read and write small data packets to the HP7980 but
that's about all. With the tape drive, you have to worry about block
sizes, file and record markers, rewind, readahead... Blah... Blah...
Blah... As of now, I have not yet figured out how to handle all that
stuff. So, I decided to take a break from the tape drive and concentrate
on the HPIB CS80 disks.
I suspect the HP150 is using the "Cold Load" command to read the tape on
bootup. I have not been able to read a tape using that method on my PC
(yet).
I was already familiar with the CS80 protocol so, I just had to learn to
program the 7210 chip. It took a while but, I can finally, with some
confidence, program the 7210 to do what I need. I was also able to write
some code to support a couple of HP instruments (multimeter / frequency
counter) over HPIB.
Programming the CS80 disk was fairly easy and, at this point, I can read
or write a raw binary file to any partition (including boot) of the HPIB
disk from my PC (linux). That disk is then bootable from the HP1000 with
a CS80 ROM.
I have created a small "library" of sorts with the common CS80 methods
and another with the HPIB methods in another. I'm not a "C" programmer
so, anyone looking at the code would probably think it's sloppy. But, it
works (sorta) :-)
The question is: What exactly are your goals at this point? If you're
still trying to boot the HP150 off the HP7980, I can probably provide
some additional insight into the tape protocol although, I cannot
provide any specific info for programming the HP150.
One of the problems with programming the tape drive is that you cannot
slow down the data interchange to see exactly what it is doing. With the
disk there is no "timeout" for any of the transactions. You can slow
down (or halt) the bus and watch the interaction between the disk and
the computer without causing any errors. You cannot do this with the
tape drive. It'll throw an error :-( Furthermore, when the computer
boots there is a large chunk of data that is passed across the bus this
bloats your analyzer's capture and makes it much harder to pick out the
important bits.
You might try to capture only the transactions where the ATN is
asserted. This will filter out the "data" and only leave the commands to
wade through.
If you have a HPIB card with the 7210 or 9914 I can perhaps provide some
insight into register level programming with those chips. I have not
given up on the AMIGO tape protocol. I just have been focused on other
stuff :-)
There is a steep learning curve when tackling all these protocols at the
same time. My suggestion would be to break up the project into smaller
chunks. The first thing is to figure out how to do the register level
programming on the HPIB card. You will need to know how to send and
receive "commands" and "data", how to conduct a "parallel
poll", and how
to change "states" on the card.
The tape and disk protocols are not simple, I would also suggest using a
simple device (HPIB meter etc...) to refine you programming. Use the
analyzer card to watch the bus to see exactly what happens when a
certain chunk of code is executed. Once you get predictable results,
it'll be easier to tackle a more complex protocol.
Talk to me about the HP150. What HPIB chip does it use? Does your
version of DOS/UNIX have the tools that you need to do register level
programming? I have HPUX 10.20 on a HP9000/800 that has HPIB support
built in. Does the 150 have this kinda support?
See ya,
SteveRob
steerex at ccvn.com
4 May
4 May
8:08 a.m.
--- On Sat, 5/3/08, Steve Robertson <steerex at ccvn.com> wrote:
I have created a small "library" of sorts
with
the common CS80 methods
and another with the HPIB methods in another. I'm not a
Steve, I'd love to see what you did with the CS80 methods under Linux if you're
willing to share. Some time back we talked about this (maybe 2 years ago) but as so often
happens, life got in the way and I haven't progressed my CS80 project very much. I
have a working build of the HPIB library under Linux and had started to poke around the
CS80 state model documentation. My goal is to provide an emulated CS80 disk that I can
boot my HP64000s and HP9000s from.
-Dave
3:17 p.m.
Steve Robertson wrote:
I have FINALLY made some more progress with the ISA
GPIB card in my
Linux box. Unfortunately, it uses the NEC7210 chip which has been a pain
in the rear to program. I found a few snippets of code online but really
didn't find anything that directly addressed my needs.
Is there some reason you want to access the 7210 directly rather than
using the Linux-GPIB library and driver?
http://linux-gpib.sourceforge.net/doc_html/index.html
I've successfully used that with several ISA, PCI, and PCMCIA cards,
including one that was 7210 based. My most recent use was to talk to
HP 9144 and 9145 cartridge tape drives.
To act as a device rather than the controller, poll ibsta() and use
the returned status to determine when the controller wants you to
talk or listen.
Eric
3:45 p.m.
I wrote:
To act as a device rather than the controller, poll
ibsta() and use
the returned status to determine when the controller wants you to
talk or listen.
Just for clarification, I didn't mean busy-wait polling. Obviously
one uses a loop with ibwait(), and checks ibsta() only when an
event occurs.
Eric
5:49 p.m.
Is there some reason you want to access the 7210
directly rather than
using the Linux-GPIB library and driver?
http://linux-gpib.sourceforge.net/doc_html/index.html
I've successfully used that with several ISA, PCI, and PCMCIA cards,
including one that was 7210 based. My most recent use was to talk to
HP 9144 and 9145 cartridge tape drives.
Yo,
For a couple of reasons.
1.) I took a look at the Linux-GPIB project but could not get a good
build on my linux box. I probably could have dinked around with it and
eventually got a clean build but, didn't want to mess with the kernel
too much.
2.) I looked at the sources and was able to grep some information but,
not being a C++ person, really couldn't follow all of it.
3.) I feel the need to know EXACTLY how these things work. I know it's
reinventing the wheel but, isn't that what we all do in this forum ;-)
Eric - Somewhere I had a link to your 9144 project but can't find it.
Please send the link again.
See ya,
SteveRob
steerex at ccvn.com
5 May
5 May
3:42 p.m.
Eric - Somewhere I had a link to your 9144 project but
can't find it.
Please send the link again.
http://svn.brouhaha.com/viewvc/read9144/trunk/
4 May
4 May
4:29 p.m.
Talk to me about the HP150. What HPIB chip does it
use? Does your
It's a TMS9914 (I've just checked the TechRef), directly on the 8088 bus.
In general HP's HPIB-capable computers of that vintage used the 9914.
Some peripherals used the Intel 8291, others bit-banged the protocol
using a microcontroller. Later computers used an HP custom chip called
'Medusa' IIRC. Earlier computers either used a microcontroller or
discrete TTL chips.
version of DOS/UNIX have the tools that you need to do
register level
THe HP150, being an 8088 machine, has no hardware protection at all, and
the TechRef gives the addreses for the 9914 registers (and tells you to
look at the data sheet for said chip :-)). There is an MS-DOS device
driver for it, which implelents a subset of the HPIB functions using
ioctl() calls (basically, according to the manual, it only supports
functions where the HP150 initiates the operation).
One disadvantage of the HP150 for this is that it only supports one HPIB
port. On the HP9000/200 machines you can have 2 or more HPIB interfacs,
it's best to have one for you system disks, etc and the other for talking
to the device you're hacking about with. That way, if you make a mistake
you'll not corrupt your system disk.
-tony
6:15 p.m.
On Sun, 2008-05-04 at 21:29 +0100, Tony Duell wrote:
Thanks Tony,
I'm sure Rodney Brown would like to have the register/address mappings
for the HP150.
I have a reference manual for a Microsys STD-BUS HPIB card that uses the
9914. It includes a reference with Z80 source code for all the common
functions. It's trivial to modify the source to "C" and compile for a PC
so, when I get a chance, I'll transpose the source and post it somewhere
for others to reference.
I have several later HP9000's and checked the chips in them. They have a
48 pin chip and are marked 1TL1-0008. A quick google search confirms
that's the "medusa" chip.
HPUX 10.20 for the 9000's has HPIB support built in. Like the 150 you
use ioctl() calls to access the chip. I have done a little HPIB
programming on the 9000 but, was never really comfortable with the
ioctl() methods.
Thanks,
SteveRob
steerex [at] ccvn [dot] com
Talk to me
about the HP150. What HPIB chip does it use? Does your
It's a TMS9914 (I've just checked the TechRef), directly on the 8088 bus.
In general HP's HPIB-capable computers of that vintage used the 9914.
Some peripherals used the Intel 8291, others bit-banged the protocol
using a microcontroller. Later computers used an HP custom chip called
'Medusa' IIRC. Earlier computers either used a microcontroller or
discrete TTL chips.
version of DOS/UNIX have the tools that you need
to do register level
THe HP150, being an 8088 machine, has no hardware protection at all, and
the TechRef gives the addreses for the 9914 registers (and tells you to
look at the data sheet for said chip :-)). There is an MS-DOS device
driver for it, which implelents a subset of the HPIB functions using
ioctl() calls (basically, according to the manual, it only supports
functions where the HP150 initiates the operation).
5 May
5 May
2:43 p.m.
I'm sure Rodney Brown would like to have the
register/address mappings
for the HP150.
Sure.
The HP150 only decodes the bottom 8 address lines when doing an I/O
cycle, but the manual suggests setting the upper byte to 0 for
compatibility with future products. Expansion cards may, of course,
decode all 16 I/O address lines.
The I/O map is as follows :
0000-0003 : MPSC Datacomm controller (7201)
0004-000B : HPIB controller (9914)
000C-000F : Baud rate generator (8116T)
0010-0013 : Interrupt controller (8259)
0014-0015 : Datacomm 1 port control, clock select
0016-0017 : Datacomm 2 port control, manfuacturing test repeat
0018-001B : Keyoard/touchscreen controller (8041A)
001C-002F : Reserved, DO NOT USE
0030-003F : Internal printer interface
0040-005F : Real time clock (MM58167A)
0060-007F : I/O image DO NOT USE
0080-00FF : Resereved for accessory slots
Now, to expand each of thase :
MPSC (7201/8274)
0000 : Channel A data register (R/W)
0001 : Channel A control register (R/W)
0002 ; Channel B data register (R/W)
0003 : Channel B control register (R/W)
HPIB (9914)
0004 : Interrupt status 1/Interrupt Mask 1 (R/W)
0005 : Serial poll (R)
0006 : Bus Status/Auxilliary command (R/W)
0007 : Data In/Data Out (R/W)
0008 : Interrupt Status 0/Interrupt Mask 0 (R/W)
0009 : Address switch/address reg (R/W)
000A : Address status register (R)
000B : Command pass-through/parallel poll (R/W)
Baud rate generator (8116T)
000C : Baud rate select (W). High nybble -> port 2, low nybble -) port 1
Nybble : Baud
0 50
1 75
2 110
3 134.5
4 150
5 300
6 600
7 1200
8 1800
9 2000
A 2400
B 3600
C 4800
D 7200
E 9600
F 19200
Interrupt cnotroller (8259A)
0010, 0011 : Interrupt controller registers (R/W)
Datacom 1 control/clock select
0014 : Read
Bit
7 0
6 0
5 0
4 1
3 DM
2 0
1 OCR2
0 OCR1
0014 : Write
7,6 Clock select
1 OCD2
0 OCD1
Datacom 2 contorlmanufacturing test repeat
0016 : Read
4 POD (1 if datacom 1 daughterboard in place)
3 DM
2 MTST (1 if U62 pin 14 grounded -> repeat failing manufacturing test)
1 OCR2
0 OCR1
0016 : Write
1 OCD2
0 OCD1
Keyboard/touchscreen cotnroller (8041A)
0018 : Data register (R/W)
0019 : Status/cmmmand register (R/W)
Internal printer
0030 : write -> Character to print
0030 : read
2 Not PaperOut
1 Online
0 Ack
Reak tune clock (58167A)
0040 : Millisecond counter (R/W)
0041 : Tenth/hundreth second counter (R/W)
0042 : Seconds (R/W)
0043 : Minutes (R/W)
0044 : Hours (R/W)
0045 : Day (R/W)
0046 : Date (R/W)
0047 : Month (R/W)
0048 : Milisecond RAM (R/W)
0049 : Tenth/hundreth second RAM (R/W)
004A : Seconds RAM (R/W)
004B : Minutes RAM (R/W)
004C : Hour RAM (R/W)
004D : Day RAM (R/W)
004E : Date RAM (R/W)
004F : Month RAM (R/W)
0050 : Interrupt status (R)
0051 : Interrupt control (W)
0052 : Counter reset (W)
0053 : RAM reset (W)
0054 : Status bit (R)
0055 : Go command register (W)
0056-005F : Reserved.
Incidentally, the HP150-II (12" CRT model) is much the same except that
there's no internal printer interface. The keyboard/touchscreen
controller is replaced by an 8042 talking to the HP-HIL chip.
That should get you started :-)
[...]
I have several later HP9000's and checked the
chips in them. They have a
48 pin chip and are marked 1TL1-0008. A quick google search confirms
that's the "medusa" chip.
Yes. I think it also campe in a PLCC package, and was used in some later
disk units.
The HP9000/200 machines (the only ones I know much about) had a built-in
HPIB port using a 9914 chip. There was an expansion card (98624 IIRC)
that added another HPIB interface, also using a 9914, and a 'high speed
disk interface' using the 1TL1 Medusa. I certainly have the former card
in my collection (to go on the bench soon), I may well have the latter.
Anohter chip you might come acress in HP HPIB devices is the 1LH4-0001.
This is _NOT_ an HPIB controller, it's just the bus buffer. The pinout
(from the 82169 service manual) is :
------------
Gnd --| U |-- Vcc
D0 --| |-- Q0 ]
D1 --| |-- Q4 ]
D2 --| |-- Q1 ]
D3 --| |-- Q5 ]
D4 --| |-- Q2 ]
D5 --| |-- Q6 ]
D6 --| |-- Q3 ]
D7 --| |-- Q7 ] HPIB connector signals
CA/ --| |-- EOI ]
TA --| |-- REN ]
SRQ --| |-- DAV ]
ATN --| |-- NRFD ]
EOI --| |-- NDAC ]
DAV --| |-- IFC ]
NDAC --| |-- SRQ ]
NRFD --| |-- ATN ]
IFC --| |-- PPL/
REN --| |-- UPE/
Gnd --| |-- SC
------------
It appears that SC = System controller
CA/ = Contoller active
TA = Talker active
-tony
4 May
4 May
8:11 p.m.
On Sat, 2008-05-03 at 19:29 -0400, Steve Robertson wrote:
> Thanks for your work on this Steve, I need to
find the one or two I
> spotted again, but yours was cleaner than mine.
> I tested the PCI NI-Analyser card again on the
HP150 - which is the
> working HP-IB machine I have. Since I didn't know better I captured
> everything again - including timings. For someone writing an emulator,
> this could be helpful, but I don't know if you want another huge capture
> file of boring Amigo protocol stuff. I've attached what I did and
> captures of the analyser interface - I'd welcome your advice on what I
> should be capturing that would be useful for protocol understanding.
Hey Rodney,
Hope you don't mind but, I'm gonna post this to the classiccmp list.
Someone there may want to follow the discussion.
Not a problem, HP150 boots to Amigo 9133 NI Analyser captures available
on request (7-zipped 16Mb capture file, 168Mb text output)
What kind of system are you running the PCI Analyzer
on? Windows? Linux?
I have an ISA NI GPIB+ card in a Win ME box that will go to Jon Johnston
of hpmuseum.net, so his CE friends could capture traces from more
interesting hardware.
A PCI NI GPIB+ card that is in a W2k box, but may end up in my Linux
box, as just a GPIB card.
Unfortunately National Instruments don't release programming info for
their analyser cards - when I last went looking they don't release the
format of the capture files, though I have some of that decoded, though
not all the named state transitions.
I have FINALLY made some more progress with the ISA
GPIB card in my
Linux box. Unfortunately, it uses the NEC7210 chip which has been a pain
in the rear to program. I found a few snippets of code online but really
didn't find anything that directly addressed my needs.
I have been able to read and write small data packets
to the HP7980 but
that's about all. With the tape drive, you have to worry about block
sizes, file and record markers, rewind, readahead... Blah... Blah...
Blah... As of now, I have not yet figured out how to handle all that
stuff. So, I decided to take a break from the tape drive and concentrate
on the HPIB CS80 disks.
I suspect the HP150 is using the "Cold Load"
command to read the tape on
bootup. I have not been able to read a tape using that method on my PC
(yet).
I'm not surprised, from a quick read of the HP85 emulator, it looked
like that by default the Amigo disks work like a tape of fixed size
blocks.
I was already familiar with the CS80 protocol so, I
just had to learn to
program the 7210 chip. It took a while but, I can finally, with some
confidence, program the 7210 to do what I need. I was also able to write
some code to support a couple of HP instruments (multimeter / frequency
counter) over HPIB.
I don't have hardware or this kind of programming experience.
Programming the CS80 disk was fairly easy and, at this
point, I can read
or write a raw binary file to any partition (including boot) of the HPIB
disk from my PC (linux). That disk is then bootable from the HP1000 with
a CS80 ROM.
I have created a small "library" of sorts
with the common CS80 methods
and another with the HPIB methods in another. I'm not a "C" programmer
so, anyone looking at the code would probably think it's sloppy. But, it
works (sorta) :-)
I'd be interested to have a look. Ultimately I'm interested in
collecting enough information to be able to emulate an HP3000, but
emulating disk devices to keep machines running is important too.
Asking Bering or Reactive AEM to open source their code to emulate HP-IB
devices using SCSI disks could work.
When we start the MPE/iX machine in the office, I'll try decompiling the
HIO at .PUB.SYS and the SL.PUB.SYS from a Series 52 Sysdump I have.
They should give some hints.
The question is: What exactly are your goals at this
point? If you're
still trying to boot the HP150 off the HP7980, I can probably provide
some additional insight into the tape protocol although, I cannot
provide any specific info for programming the HP150.
I'd like to get the 7980 to someone who can use it (~35kg, Melbourne,
Au). I have a few tapes I'd like to image if the drive is behaving, if
possible reading to my/a Linux box. The HP150 is the working HP-IB
(MS-DOS) machine I have, supporting Amigo & SS-80 protocols I think. The
techref manual for it probably describes HP-IB programming, probably
using ioctl calls on an HPIB device analogous to the AGIOS calls for
accessing the inherent HP2623 terminal emulator. Tony Duell has the only
copy I know of.
One of the problems with programming the tape drive is
that you cannot
slow down the data interchange to see exactly what it is doing. With the
disk there is no "timeout" for any of the transactions. You can slow
down (or halt) the bus and watch the interaction between the disk and
the computer without causing any errors. You cannot do this with the
tape drive. It'll throw an error :-( Furthermore, when the computer
boots there is a large chunk of data that is passed across the bus this
bloats your analyzer's capture and makes it much harder to pick out the
important bits.
The 7970s probably wouldn't have cared, but a 7980 is a different beast.
You might try to capture only the transactions where
the ATN is
asserted. This will filter out the "data" and only leave the commands to
wade through.
If you have a HPIB card with the 7210 or 9914 I can
perhaps provide some
insight into register level programming with those chips. I have not
given up on the AMIGO tape protocol. I just have been focused on other
stuff :-)
The 7980 is CS-80 as you know
There is a steep learning curve when tackling all
these protocols at the
same time. My suggestion would be to break up the project into smaller
chunks. The first thing is to figure out how to do the register level
programming on the HPIB card. You will need to know how to send and
receive "commands" and "data", how to conduct a "parallel
poll", and how
to change "states" on the card.
Understood
The tape and disk protocols are not simple, I would
also suggest using a
simple device (HPIB meter etc...) to refine you programming. Use the
analyzer card to watch the bus to see exactly what happens when a
certain chunk of code is executed. Once you get predictable results,
it'll be easier to tackle a more complex protocol.
Talk to me about the HP150. What HPIB chip does it
use? Does your
version of DOS/UNIX have the tools that you need to do register level
programming? I have HPUX 10.20 on a HP9000/800 that has HPIB support
built in. Does the 150 have this kinda support?
I don't know what HP-IB chip the HP150 uses - but it looks like Tony has
answered.
My HP-UX boxes are 700 series without HP-IB - I should look out for such
an 800 series - though I remember those as being large.
See ya,
SteveRob
Thanks again Steve
5917
days inactive
5919
days old
9 comments
5 participants
participants (5)
-
ard＠p850ug1.demon.co.uk -
david_comley＠yahoo.com -
eric＠brouhaha.com -
rdbrown＠pacific.net.au -
steerex＠ccvn.com