25 Jun
2014
25 Jun
'14
6:07 p.m.
Hello everybody,
I've always wanted to own a minicomputer, at least, ever since I learned what a
minicomputer was in my computer architecture class. Something about the user interface
was wonderfully refreshing, of course all we had was SIMH, which is awesome in its own
right, but wasn't the same. Anyway, last month I was finally able to pick up a
minicomputer (I could afford) off eBay, the HP 1000E/2113E.
It wasn't in the best shape, but I've been able to fix the fans, bring the power
supply back into spec, and design replacements for the irreplaceable front panel buttons
(http://www.thingiverse.com/thing:330821). As much fun as I'm having assembling and
toggling in trivial programs, I'm looking to get a complete system up and running,
something I can run a more-or-less useful operating system on. This brings me to my
current problem:
It looks like the original system peripherals are very few and far between, and when they
do show up for sale, they are astronomically priced. My plan is to put together a system
with whatever inexpensive contemporaneous parts I can find, regardless of how sacrilegious
it is.
I got a hold of two mostly working DEC RL02 disk drives, along with a SCSI 9-track tape
drive, and my (pipe) dream is to interface this all to the HP using an FPGA and HP True
In/Out card (i.e. HP Bizarro-Bus to TTL converter). As for software, I'm looking
towards writing a backend for GCC (understatement x1000) and running FreeRTOS (because
I'm familiar with it, I don't need runtime application loading, and I'm going
to need to write drivers anyway).
Believe it or not, the RL02 interface is probably the easiest part of this project.
I've already started working on a design in Verilog that should be able to provide the
necessary functionality over a wishbone interface, and I've gotten the drive to spin
up and do some simple tasks with just the FPGA controller connected. (Before anyone asks,
yes, I suppose one could make the RL02 a really stationary USB thumbdrive.)
At this point my question is, has anybody tried something like this before and succeeded
(or not)? Are there any helpful lessons to be imparted about the HP 1000 or RL02s or
working on this type of project in general? I think I have everything segmented off into
a monstrous series of small tasks, but I would love to hear from anyone who has already
learned from the mistakes I'm about to repeat.
Thanks,
Christopher Parish
26 Jun
26 Jun
10:54 a.m.
From: christopher.parish at parishcomputers.com
To: cctalk at classiccmp.org
Subject: HP-1000 Minicomputer System (w/DEC RL02) Guidance
Date: Thu, 26 Jun 2014 01:07:42 +0000
Hello everybody,
I've always wanted to own a minicomputer, at least, ever since I learned what a
minicomputer was in my computer architecture class. Something about the user interface
was wonderfully refreshing, of course all we had was SIMH, which is awesome in its own
right, but wasn't the same. Anyway, last month I was finally able to pick up a
minicomputer (I could afford) off eBay, the HP 1000E/2113E.
---snip---
Thanks,
Christopher Parish
As a project of love, I say go for it. Nothing impossible, just mostly time.
The only problem you may have is that you may need a buffer to keep
up with the drive.
I've done similar stuff in the past but not with fpgas. I've made a disk
setup, both floppy and hard drive for a NC4016 computer using
a XT floppy card and a XT hard drive card. I wrote my own drivers,
and did the 16 bit to 8 bit. I even added some hardware to swap
bytes of a 16 bit word to speed things up.
Dwight
11:06 a.m.
In article <639FFF5F324E6D409F42CF2B2F3BB56AB0FABEB1 at
MBX022-E1-NJ-4.exch022.domain.local>,
Christopher Parish <christopher.parish at parishcomputers.com> writes:
As for software, I'm looking towards writing a
backend for GCC
(understatement x1000) and running FreeRTOS (because I'm familiar with it,
I don't need runtime application loading, and I'm going to need to write
drivers anyway).
You may want to consider using clang instead of gcc for your compiler.
<http://clang.llvm.org>
--
"The Direct3D Graphics Pipeline" free book
<http://tinyurl.com/d3d-pipeline>
The Computer Graphics Museum <http://ComputerGraphicsMuseum.org>
The Terminals Wiki <http://terminals.classiccmp.org>
Legalize Adulthood! (my blog) <http://LegalizeAdulthood.wordpress.com>
11:10 a.m.
I second what Richard said, LLVM is a lot more friendly for compiler
writing.
On 26 June 2014 12:06, Richard <legalize at xmission.com> wrote:
In article
<639FFF5F324E6D409F42CF2B2F3BB56AB0FABEB1 at MBX022-E1-NJ-4.exch022.domain.local
--
Thanks,
Kevin
,
Christopher Parish
<christopher.parish at parishcomputers.com> writes:
As for software, I'm looking towards writing
a backend for GCC
(understatement x1000) and running FreeRTOS (because I'm familiar with
it,
I don't need runtime application loading, and
I'm going to need to write
drivers anyway).
You may want to consider using clang instead of gcc for your compiler.
<http://clang.llvm.org>
--
"The Direct3D Graphics Pipeline" free book <
http://tinyurl.com/d3d-pipeline>
The Computer Graphics Museum <http://ComputerGraphicsMuseum.org>
The Terminals Wiki <http://terminals.classiccmp.org>
Legalize Adulthood! (my blog) <http://LegalizeAdulthood.wordpress.com>
1:14 p.m.
Hello everybody,
I've always wanted to own a minicomputer, at least, ever since I learned wh=
at a minicomputer was in my computer architecture class. Something about t=
he user interface was wonderfully refreshing, of course all we had was SIMH=
, which is awesome in its own right, but wasn't the same. Anyway, last mon=
th I was finally able to pick up a minicomputer (I could afford) off eBay, =
the HP 1000E/2113E.
Nice! The HP1000 is one familty of machines I've not found yet.
Annoyingly I'vealomst got one 3 times in my life, every time I lost out :-(
It looks like the original system peripherals are very few and far between,=
and when they do show up for sale, they are astronomically priced. My pla=
n is to put together a system with whatever inexpensive contemporaneous par=
ts I can find, regardless of how sacrilegious it is.
I got a hold of two mostly working DEC RL02 disk drives, along with a SCSI =
9-track tape drive, and my (pipe) dream is to interface this all to the HP =
using an FPGA and HP True In/Out card (i.e. HP Bizarro-Bus to TTL converter=
Personally I see nothing wrong with projects like this, provided you make
no permanent changes to the cvhassic machines. Something that plugs into
an I/O conencotr, or a new PCB that goes into an expansion slot seems ot
be a very reasoanble thing to do.
But personally I prefer to use the sort of technology that was around
when the machine was current. So I wouldn't use FPGAs (there are otehr
reasons why I detest these devices oto...) For me it would eb a rack of
PCBs stuffed with SSI/MSI chips :-)
). As for software, I'm looking towards writing a
backend for GCC (underst=
atement x1000) and running FreeRTOS (because I'm familiar with it, I don't =
need runtime application loading, and I'm going to need to write drivers an=
yway).
I've not looked, but is there anything useful on http://www.hpmuseum.net/
At this point my question is, has anybody tried something like this before =
and succeeded (or not)? Are there any helpful lessons to be imparted about=
the HP 1000 or RL02s or working on this type of project in general? I thi=
The RL is a bit strange.. A few thigns I rmeemebr from when I repaied a
feww :
The 'energiser clock must be present and the right frequency all the
time. Even the motor speed uses it as a reference.
To do a seek you send the _offest_ -- how many cylinders to move,not the
absolute cylidner number. This is different from just about every other
14" demountable drive I've seen
Be carefu lwith wrtie gate timing. I _think_ the drive hardware prevents
you from overwriting the servo bursts, but you get a fatal error if you
try. Best nto to risk it though.
-tony
1:36 p.m.
On Thu, Jun 26, 2014 at 2:14 PM, Tony Duell <ard at p850ug1.demon.co.uk> wrote:
Be carefu lwith wrtie gate timing. I _think_ the drive
hardware prevents
you from overwriting the servo bursts, but you get a fatal error if you
try. Best nto to risk it though.
I second that! You *really* don't want to risk overwriting a servo burst,
as there's no way for mere mortals to fix that. It's not even clear
whether *anyone* has the ability to fix that any more; even back in the day
the "solution" was generally to throw the pack away and buy a new one. So
even if the drive electronics has some protection against that, I'd
recommend ensuring that your design has protection against it as well.
3:44 p.m.
On Jun 26, 2014, at 13:36 , Eric Smith <spacewar at gmail.com> wrote:
I second that! You *really* don't want to risk
overwriting a servo burst,
as there's no way for mere mortals to fix that. It's not even clear
whether *anyone* has the ability to fix that any more; even back in the day
the "solution" was generally to throw the pack away and buy a new one. So
even if the drive electronics has some protection against that, I'd
recommend ensuring that your design has protection against it as well.
I've been curious about whether it would be possible to rewrite servo bursts on an
RL02 pack with a modified RL02 drive using ordinary RL02 heads. I think it would be
straightforward (though not easy) to modify the drive mechanics and replace the
electronics in order to position the head properly and write the servo bursts. The
outstanding question in my mind is whether special heads with a different (probably wider)
track width would be necessary, and that would be the part that I wouldn't know how to
fabricate.
I wonder if it would be possible to make the servo bursts and regular data tracks on an
old RL02 platter visible with magnetic developer fluid in order to measure the feature
widths under an ordinary microscope?
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
6:21 p.m.
On Thu, Jun 26, 2014 at 4:44 PM, Mark J. Blair <nf6x at nf6x.net> wrote:
I've been curious about whether it would be
possible to rewrite servo
bursts on an RL02 pack with a modified RL02 drive using ordinary RL02 heads.
I think so, but I don't know for certain. Even if the results didn't meet
the original manufacturing specifications (which we don't have), it might
be "good enough".
I think you'd need to do a good job of degaussing the entire surface before
writing the servo patterns. I don't think using any normal degaussing coil
will be sufficiently effective, but doing it with a normal head and
stepping in increments less than the track width might be viable.
6:49 p.m.
On Jun 26, 2014, at 18:21 , Eric Smith <spacewar at gmail.com> wrote:
I think you'd need to do a good job of degaussing
the entire surface before
writing the servo patterns. I don't think using any normal degaussing coil
will be sufficiently effective, but doing it with a normal head and
stepping in increments less than the track width might be viable.
As I understand it, the actuator in the RL02 is a DC motor with a band-and-track mechanism
much like floppy drives often have, and no positional feedback independent other than the
error signal derived from the servo bursts on the media. It seems to me that by adding a
positional transducer to the head and a suitable servo system, it should thus be possible
to position the head arbitrarily, with resolution limited by the transducer. I don't
know whether an off-the-shelf rotary optical encoder attached to the DC motor shaft would
be adequate, or if something like the diffraction plates used in drives like my Data
General 6045 (*) would be needed.
If the normal servo bursts are written with a head that has a wider track width than
normal, then maybe usable bursts could still be written with a narrower track width? And
if they're normally narrower, then it'd be a matter of being able to position the
heads enough off-track to write the bursts, but not so far as to trash adjacent tracks.
Without knowing the track and head geometries, I can only speculate, based on my minimal
knowledge of hard drive innards based on starting my career around 1992 at a company that
made chips for the hard drive industry.
(*) It uses a pack that appears to be mechanically the same as the RL02's pack, other
than a different sector pulse pattern on the hub. I believe that it positions the heads
based on feedback from a glass diffraction grating mounted to the positioner, rather than
on feedback read from the media. So in effect, it acts much like an open-loop stepper
motor drive despite using a linear voice coil motor. In contrast, the RL02 acts much like
a modernish drive with closed-loop servo derived from the media, even though its mechanism
looks a lot like a floppy-drive's positioner at first glance. I spend way too much
time geeking out over this stuff!
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
27 Jun
27 Jun
7:45 a.m.
On 6/26/14 6:49 PM, Mark J. Blair wrote:
It seems to me that by adding a positional transducer
to the head and a suitable servo system, it should thus be possible to position the head
arbitrarily, with resolution limited by the transducer.
If you're going to throw hardware at it, put a DSP in the read channel and positioner
and derive the track position on
the fly based on the recovered data with coarse positioning based on the gross amplitude
of the entire track.
I have actually thought about that for general 14" platter data recovery as opposed
to the open-loop method used to recover
the Cray OS diagnostic pack data.
The method should also work for Apple Widget disk data recovery, where most of the drive
mechanisms have failed
due to the glue failing on the glass graticule used for coarse servo positioning.
But to be honest, I don't see the logic in using an RL02 on an HP-1000 system in the
first place.
7:58 a.m.
On Jun 27, 2014, at 07:45 , Al Kossow <aek at bitsavers.org> wrote:
On 6/26/14 6:49 PM, Mark J. Blair wrote:
That might be a good approach for reading packs with damaged servo bursts, but the
application I had in mind was writing servo bursts to a degaussed platter.
It seems to me that by adding a positional
transducer to the head and a suitable servo system, it should thus be possible to position
the head arbitrarily, with resolution limited by the transducer.
If you're going to throw hardware at it, put a DSP in the read channel and positioner
and derive the track position on
the fly based on the recovered data with coarse positioning based on the gross amplitude
of the entire track. The method should also work for Apple Widget disk data
recovery, where most of the drive mechanisms have failed
due to the glue failing on the glass graticule used for coarse servo positioning.
I didn't know that the Widget drives used closed loop servo. Interesting!
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
2:40 p.m.
On Jun 26, 2014, at 13:36 , Eric Smith <spacewar at gmail.com> wrote:
It possibly is, but the modifications to the drive would be quite major.
The main problem is makign that bsolute opostion sensor that is accrate
enough and stae enough to be sueable. I belive the origianl factory
formatting set-uy used some kid of interferometer.
I am not susre you need a wider head. The servo bursts are offset from
the data tracks, but may well eb the same width (if anything, they'd be
narrower, I think).
As an asaide, I did like a methrd used for one of the HP drives which had
fixed and remvoeable platters. It was possible to rewrite the servo
infroamtion o nthe fixed platter in the field. What you did was to use
soem special electroncis (and head?) to take the positioning infroamtion
form the servo surface of the removable pack, and then wrtie the servo
pattern o nthe approrpate sruface of the fixed disk pack. I have never
seen the setup to do this, though.
FWIW, given a servo-tracked drive it is normally possible to fairly
accrately offset the haeds from the position where the servo thinks they
should be. The RK07 made ths a user feature (one of the cotnroller
registers can set an offstr) to recover mis-recorded disk packs IIRC
I have this vision of a modified RL02 drive with the platter from a
dismantled pack on the same spidnle and the normal pack and 4 heads o
nthe positioner. The 2 on the now-fixed paltter are used for positioning,
you can then use the heads on the removeable pack can write servo
signals, etc.
-tony
I second that! You *really* don't want to
risk overwriting a servo burst,
as there's no way for mere mortals to fix that. It's not even clear
whether *anyone* has the ability to fix that any more; even back in the day
the "solution" was generally to throw the pack away and buy a new one. So
even if the drive electronics has some protection against that, I'd
recommend ensuring that your design has protection against it as well.
I've been curious about whether it would be possible to rewrite servo
bursts on an RL02 pack with a modified RL02 drive using ordinary RL02
heads. I think it would be straightforward (though not easy) to modify
26 Jun
26 Jun
1:24 p.m.
On Jun 25, 2014, at 18:07 , Christopher Parish <christopher.parish at
parishcomputers.com> wrote:
Before anyone asks, yes, I suppose one could make the
RL02 a really stationary USB thumbdrive.
I think I would enjoy giving one of my RL02 drives a USB interface.
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
27 Jun
27 Jun
2:07 a.m.
New subject: AW: HP-1000 Minicomputer System (w/DEC RL02) Guidance
Hello Christopher,
to create replacement buttons via thingiverse is a very nice job and I'll try this for
some of my machines.
Nevertheless I do have some complete front panel pcbs including buttons for the E/F/MX
series.
Andreas
-----Original-Nachricht-----
Betreff: HP-1000 Minicomputer System (w/DEC RL02) Guidance
Datum: Thu, 26 Jun 2014 08:07:54 +0200
Von: Christopher Parish <christopher.parish at parishcomputers.com>
An: "cctalk at classiccmp.org" <cctalk at classiccmp.org>
Hello everybody,
I've always wanted to own a minicomputer, at least, ever since I learned what a
minicomputer was in my computer architecture class. Something about the user interface
was wonderfully refreshing, of course all we had was SIMH, which is awesome in its own
right, but wasn't the same. Anyway, last month I was finally able to pick up a
minicomputer (I could afford) off eBay, the HP 1000E/2113E.
It wasn't in the best shape, but I've been able to fix the fans, bring the power
supply back into spec, and design replacements for the irreplaceable front panel buttons
(http://www.thingiverse.com/thing:330821). As much fun as I'm having assembling and
toggling in trivial programs, I'm looking to get a complete system up and running,
something I can run a more-or-less useful operating system on. This brings me to my
current problem:
It looks like the original system peripherals are very few and far between, and when they
do show up for sale, they are astronomically priced. My plan is to put together a system
with whatever inexpensive contemporaneous parts I can find, regardless of how sacrilegious
it is.
I got a hold of two mostly working DEC RL02 disk drives, along with a SCSI 9-track tape
drive, and my (pipe) dream is to interface this all to the HP using an FPGA and HP True
In/Out card (i.e. HP Bizarro-Bus to TTL converter). As for software, I'm looking
towards writing a backend for GCC (understatement x1000) and running FreeRTOS (because
I'm familiar with it, I don't need runtime application loading, and I'm going
to need to write drivers anyway).
Believe it or not, the RL02 interface is probably the easiest part of this project.
I've already started working on a design in Verilog that should be able to provide the
necessary functionality over a wishbone interface, and I've gotten the drive to spin
up and do some simple tasks with just the FPGA controller connected. (Before anyone asks,
yes, I suppose one could make the RL02 a really stationary USB thumbdrive.)
At this point my question is, has anybody tried something like this before and succeeded
(or not)? Are there any helpful lessons to be imparted about the HP 1000 or RL02s or
working on this type of project in general? I think I have everything segmented off into
a monstrous series of small tasks, but I would love to hear from anyone who has already
learned from the mistakes I'm about to repeat.
Thanks,
Christopher Parish
2 p.m.
Sounds like a rather nontrivial project, cool stuff.
Have you seen Reinhards work? Not quite what you are doing but
maybe it can be of help.
http://www.pdp11gy.com/indexE.html#file:///E:/homepage/indexE.html
/P
On Thu, Jun 26, 2014 at 01:07:42AM +0000, Christopher Parish wrote:
Hello everybody,
I've always wanted to own a minicomputer, at least, ever since I learned what a
minicomputer was in my computer architecture class. Something about the user interface
was wonderfully refreshing, of course all we had was SIMH, which is awesome in its own
right, but wasn't the same. Anyway, last month I was finally able to pick up a
minicomputer (I could afford) off eBay, the HP 1000E/2113E.
It wasn't in the best shape, but I've been able to fix the fans, bring the power
supply back into spec, and design replacements for the irreplaceable front panel buttons
(http://www.thingiverse.com/thing:330821). As much fun as I'm having assembling and
toggling in trivial programs, I'm looking to get a complete system up and running,
something I can run a more-or-less useful operating system on. This brings me to my
current problem:
It looks like the original system peripherals are very few and far between, and when they
do show up for sale, they are astronomically priced. My plan is to put together a system
with whatever inexpensive contemporaneous parts I can find, regardless of how sacrilegious
it is.
I got a hold of two mostly working DEC RL02 disk drives, along with a SCSI 9-track tape
drive, and my (pipe) dream is to interface this all to the HP using an FPGA and HP True
In/Out card (i.e. HP Bizarro-Bus to TTL converter). As for software, I'm looking
towards writing a backend for GCC (understatement x1000) and running FreeRTOS (because
I'm familiar with it, I don't need runtime application loading, and I'm going
to need to write drivers anyway).
Believe it or not, the RL02 interface is probably the easiest part of this project.
I've already started working on a design in Verilog that should be able to provide the
necessary functionality over a wishbone interface, and I've gotten the drive to spin
up and do some simple tasks with just the FPGA controller connected. (Before anyone asks,
yes, I suppose one could make the RL02 a really stationary USB thumbdrive.)
At this point my question is, has anybody tried something like this before and succeeded
(or not)? Are there any helpful lessons to be imparted about the HP 1000 or RL02s or
working on this type of project in general? I think I have everything segmented off into
a monstrous series of small tasks, but I would love to hear from anyone who has already
learned from the mistakes I'm about to repeat.
Thanks,
Christopher Parish
3831
days inactive
3832
days old
15 comments
11 participants
participants (11)
-
aek＠bitsavers.org -
ard＠p850ug1.demon.co.uk -
aswood＠t-online.de -
christopher.parish＠parishcomputers.com -
dkelvey＠hotmail.com -
ethan.dicks＠gmail.com -
krfkeith＠gmail.com -
legalize＠xmission.com -
nf6x＠nf6x.net -
pontus＠Update.UU.SE -
spacewar＠gmail.com